Introduction

At Intelligence at the Frontier (March 14-15, 2026), six floors of a San Francisco technology co-working hub (each floor a sector-focused community) received seed funding and chose their own governance mechanism for allocating it. The experiment was designed to produce a comparative dataset: same building, same weekend, same attendee pool, different governance approaches. This report documents what happened, what we learned, and what the data means for future governance experiments.

What follows: Section 2 places the experiment in the broader governance literature. Section 3 documents methodology. Section 4 documents the experimental design and where it deviated from plan. Section 5 carries floor-by-floor narratives. Section 6 compares mechanisms, spending, and cross-floor scoring. Section 7 covers what on-chain tools were deployed and why governance stayed off-chain anyway. Section 8 documents lessons and failure modes. Section 9 describes the permanent treasury this experiment activated. Appendices preserve full floor narratives (A), treasury architecture detail (B), tooling deployment detail (C), and replicable templates (D). The headline finding sits at the top of Section 6: every floor chose some variant of concentrated authority, and the floor with the deepest pre-existing governance practice produced the most output.

Key Recommendations

For communities running future comparative governance experiments:

1. Secure an anchor technology partner before scoping the experiment. A governance tooling partner brings a platform, implementation support, and institutional motivation. Generic funding does not replace this.
2. Budget at least $3,000–$5,000 per participating community. Below this threshold, the rational response is to minimize governance overhead. On-chain adoption has an even higher threshold.
3. Provide a dedicated governance facilitator. On-chain governance tools do not adopt themselves. Non-crypto-native communities need hands-on onboarding and a product owner for the governance infrastructure.
4. Allow 8–12 weeks for governance design before the event window. The governance patterns that played out during the event were established weeks earlier.
5. Track self-funding explicitly. It muddies mechanism comparison and should be controlled for or at minimum documented.
6. Distinguish between governance effectiveness and governance distribution. A motivated oligarchy outperformed a transparent absence. Whether that is a success depends on what the experiment is trying to measure.

Background & Related Work

The Public Goods Governance Problem

Governance of shared resources is among the oldest problems in political economy. Ostrom's Governing the Commons (1990) challenged the inevitability of Hardin's tragedy of the commons by documenting communities that successfully self-governed shared resources through eight design principles including collective-choice arrangements, monitoring, and graduated sanctions. Her work, validated across 91 empirical studies (Cox, Arnold & Villamayor-Tomás, 2010), established that self-governance is possible but demanding: successful commons regimes evolved over decades or centuries of repeated interaction, stable membership, and gradually developed institutional trust [1][2].

The web3 ecosystem has produced a new generation of governance mechanisms designed to solve public goods provision at digital scale. Buterin, Hitzig, and Weyl's quadratic funding (2019) proposed a theoretically optimal mechanism for decentralized public goods provision [3]. Conviction voting (Commons Stack, 2020) introduced time-weighted preference signals. Optimism's Retroactive Public Goods Funding distributed over 65M OP tokens across six rounds (2021–2025). Octant, developed by the Golem Foundation and co-funder of this experiment, has distributed over 2,340 ETH through its participatory allocation mechanism, which uses quadratic funding [4]. These mechanisms represent theoretical advances, yet each faces persistent empirical challenges in deployment.

The Gap Between Design and Practice

The empirical record reveals a consistent pattern: governance systems designed for participation converge toward concentrated authority regardless of formal structure.

In the DAO ecosystem, Fritsch, Müller, and Wattenhofer (2022) found that just 8 delegates could control Compound and 11 could control Uniswap, with Gini coefficients exceeding 0.99 for token-weighted governance [5]. Individual voting participation rates fall below 2% (Liu, 2023), while a top decile of voters controls 76.2% of voting power in typical proposals (Cong et al., 2025) [6]. Fan, Ai, and Liu (2024) studied 87 DAOs over six years and found that oligarchy tendency increases significantly within the first three weeks, even in systems whose participants explicitly value decentralization [7].

Quadratic funding, the mechanism most discussed in the governance design literature, has faced persistent sybil attacks (fake-identity votes used to game funding) in Gitcoin's $38M+ deployment, with over 50% of votes potentially fraudulent in unsecured rounds [8]. Conviction voting lacks published participation or outcome data from any deployment. Optimism's RetroPGF participants described the process as requiring a median 16 hours of work, far exceeding what small communities would tolerate [9].

These challenges are not limited to blockchain contexts. Freeman's "Tyranny of Structurelessness" (1972) documented that attempts at flat governance in social movements produced informal elites rather than genuine equality [10]. Schneider (2024) extended this analysis to digital spaces, coining "implicit feudalism" to describe the persistent bias, cultural and technical, toward building communities as fiefdoms, noting that even platform designers who value democracy default to hierarchical governance in practice [11]. The open-source software community's BDFL (Benevolent Dictator for Life) model has proven durable, with O'Mahony and Ferraro (2007) documenting that even Debian, after 13 years of deliberate democratic design, required blending bureaucratic and autocratic mechanisms [12].

What This Experiment Adds

Most governance research examines a single mechanism in a single community. Comparative governance data, where multiple communities operating under similar conditions choose and implement different mechanisms simultaneously, is rare. This report documents six communities in one building, over one weekend, with the same attendee pool, each freely choosing their governance approach. The experimental design draws on Ostrom's comparative case study methodology but applies it to a controlled physical environment rather than geographically dispersed commons.

Three features distinguish this experiment from prior work. First, the communities were non-crypto-native (with one exception), providing data on governance adoption outside the web3 echo chamber. Second, the experiment was embedded in a real world event with real budgets and real operational consequences. Third, the AI-assisted governance component (the S-Process deployed through Simocracy) connects to an established methodology: Andrew Critch developed the S-Process at the Survival and Flourishing Fund, which has distributed $34.92M using computational preference aggregation where recommenders express utility functions over applicants and an optimization algorithm aggregates them, with funders retaining final veto [13]. Critch was formerly chair of Funding the Commons, making this deployment a direct institutional lineage from his governance research to practitioner implementation.

The behavioral economics literature predicts what we observed. Samuelson and Zeckhauser (1988) established that individuals disproportionately choose defaults [14]. Iyengar and Lepper (2000) demonstrated that excessive choice triggers decision paralysis [15]. Dunbar's social brain hypothesis (1993) predicts that groups of 5-20 fall within the trust layers where informal governance is cognitively natural and formal participatory mechanisms add overhead without benefit [16]. At these scales, as Hackman (2002) found, team effectiveness depends more on context, structure, and support systems than on individual capabilities or formal governance design, a finding this report's data strongly supports [17].

The question this report investigates: at the scale of 5–20 people with short timelines and small budgets, is concentrated authority the functionally rational choice, and what conditions would need to change for participatory mechanisms to become viable alternatives.

Methodology

Research Design

This study employed a comparative case study design, observing six communities within a single physical environment (Frontier Tower, San Francisco) as they received seed funding and chose governance mechanisms for allocating it during a two-day event (March 14-15, 2026). The design was quasi-experimental: communities freely chose their own approaches, with the constraint that they document their governance process.

Data Sources

Five categories of data inform this analysis:

Structured debriefs. A standardized 24-question survey was administered to all six floor leads within two weeks of the event. The survey covered governance mechanism choice, budget allocation, active governance participants, transparency self-assessment (1–5 scale), cross-floor contribution, tools used, and forward-looking governance interest. Five of six floor leads completed the survey (Floors 6, 8, 11, 12, 14). Floor 9's data was compiled from transcript evidence and direct communication. A $200 honorarium was provided per respondent.

Event platform data. Session counts, scheduling data, and attendance records were exported from the Funding the Commons platform (platform.fundingthecommons.io). Ticket attribution data (registration codes, check-in rates) was compiled from the FtC platform and Luma.

Meeting and call transcripts. Pre-event floor lead coordination meetings (January-March 2026), budget meetings, and post-event retrospectives were transcribed through Fireflies.ai. The March 16 Simocracy allocation session was transcribed separately. These transcripts provided governance context, decision rationale, and direct quotes not captured in the structured debriefs.

Pre-event workspace documentation. Each floor's Notion workspace was compiled prior to the event, documenting planned governance approaches, budget proposals, programming commitments, and design aspirations. The gap between workspace documentation and actual governance outcomes became a data source in its own right.

Post-event financial records. Budget allocations, disbursement records, and self-reported spending breakdowns were compiled from FtC financial records and floor lead debrief responses.

Analytical Framework

The analysis employed three frameworks:

Mechanism comparison. Each community's governance approach was categorized by type (council, benevolent dictator, delegation, default to entity), participation depth (number of active governance participants), and transparency (self-reported 1-5 scale). These were compared across a common set of output metrics: session count, cross-floor contribution score (0-18 scale across six dimensions), and spending patterns.

Cross-floor contribution scoring. A rubric was developed to measure value created beyond each community's own space, scored 0-3 across six dimensions: cross-floor operational support, volunteer mobilization, sponsorship/revenue generation, programming beyond own space, coordination meeting attendance, and resource sharing. Scores combined self-reported data from debriefs, platform data, and organizer observation.

Correlation testing. Five relationships specified in the research plan were tested: governance participation vs. programming output, budget size vs. governance complexity, transparency vs. governance effectiveness, cross-floor contribution vs. governance engagement, and governance mechanism type vs. spending priorities. Given N=6, results are reported as observed patterns rather than statistically significant findings.

Limitations

This study has several limitations that should be weighed when interpreting the findings.

Small sample size. Six participating communities over two days cannot produce statistically significant results. Patterns are reported as suggestive observations warranting investigation at larger scale.

Budget shortfall. The experiment was designed around $3,000-$4,000 per community. A denied Ethereum Foundation grant reduced most allocations to $1,400. Floor leads described feeling demoralized, likely suppressing governance engagement. The findings reflect governance behavior under budget disappointment, not under originally designed conditions.

Self-funding confound. At least three communities had participants who spent personal funds beyond the seed allocation, up to 4x in one case. Mechanism comparison is unreliable when the budget constraint being compared varies by this much due to individual investment.

Self-reported data. Transparency scores, governance participation counts, and budget breakdowns rely on floor lead self-reporting, which may reflect aspiration or impression management rather than observed behavior.

No control for community maturity. Communities entered the experiment with vastly different levels of governance experience, organizational infrastructure, and leadership stability. These pre-existing differences likely explain more variance in outcomes than mechanism choice.

Researcher positionality. The author served as both event producer and governance experiment designer, creating potential conflicts between event success and research objectivity. Floor lead relationships with the author may have influenced debrief responses.

Building and event maturity. Frontier Tower was less than one year old at the time of the experiment, and the governance experiment ran inside the production of a 1,000-person two-day festival. The community-level trust and operational norms that comparative case studies typically rely on (Ostrom, 1990) had not yet had time to develop, and floor leads were operating under high time pressure with competing demands for attention. The findings reflect governance behavior under conditions a stand-alone experiment would not produce.

Experiment Design

Floor leads and builders at a coordination table during the weekend. The experiment ran across the six floors documented in this section.

The Thesis

Most governance experiments run one mechanism at a time, making comparison hard. Intelligence at the Frontier attempted something different: six floors, one building, one weekend, same attendee pool, different governance approaches. The goal was a comparative dataset, not proof that one mechanism outperforms another.

The Setting

Frontier Tower is a 16-floor technology hub in central San Francisco. Nearly every floor houses a distinct community: AI researchers, biotech startups, a robotics workshop, an Ethereum community hub (the first permanent Ethereum community space), a human flourishing foundation, an arts and music collective, and more.

The Ethereum-hub positioning was a central motivation for this experiment. The hypothesis was that proximity to a crypto-native community would produce at least baseline governance fluency across floors, and that Ethereum Foundation institutional interest would underwrite on-chain governance infrastructure. Both hypotheses were tested by what happened, with results documented in Sections 7 and 8.

Before this experiment, the building had no shared financial infrastructure. No community treasury, no mechanism for collective decision-making across floors, and no financial incentive for floors to collaborate. Each floor competed independently for the same pool of potential members. The structure rewarded isolation over coordination.

The experiment was designed to test a building-level hypothesis alongside the floor-level governance comparison. The contractual agreement between Funding the Commons and Frontier Tower introduced revenue sharing across the building's authority layers: 2% of event revenue and 20% of event profit would flow into a tower-wide community treasury co-signed by floor leads, building management, and Funding the Commons as co-steward. The premise: aligning financial incentives across floor leads through a shared treasury would convert previously zero-sum floor-to-floor dynamics into positive-sum coordination. The floor-level governance experiment is what this report's main analysis documents in Sections 5 through 8. The building-level evidence is suggestive rather than dispositive, given two days of operation and the absence of a controlled comparison; Section 9 presents what the data shows and its limitations.

The Rules

Each participating floor received a seed budget. The base allocation was $1,400 per floor, with variations: floors with minimal session commitments received half ($700), and floors taking on additional production responsibilities received more (Floor 14 received an extra $1,000 for producing Earth Commons programming on a second floor; Floor 6 received an additional $1,600 for booking and running nighttime music events building-wide).

The governance requirements were intentionally minimal: choose a mechanism, document the process. Any decision-making approach was acceptable: consensus, delegation, voting, council, benevolent dictator, or something invented for the occasion. No prescribed tooling, no mandatory on-chain transactions, no reporting templates beyond basic documentation.

What Actually Happened vs. What Was Planned

The experiment deviated from the original design in several ways. This section documents those deviations honestly, because the gap between design and reality is itself useful data.

Participation: 6 floors instead of 8+. Eight floors were originally expected to participate. Two dropped out. Floor 7 (Makerspace) disengaged entirely after the floor lead stopped attending coordination meetings. Floor 4 (Robotics/Physical AI) focused exclusively on the hackathon and did not participate in the conference programming or governance experiment, though they contributed physical space and equipment.

Budget shortfall. A denied Ethereum Foundation grant reduced per-floor allocations from a planned $3,000-$4,000 to $1,400. The effect on participation, and the full story of the grant denial, is documented in Section 8 Lesson 4.

On-chain infrastructure: mostly off-chain. The original design specified on-chain wallets per floor, on-chain transaction records, and governance documentation published to IPFS/GitHub. In practice, governance stayed mostly off-chain. Floor-level decisions were primarily made through informal coordination: WhatsApp groups, Telegram, in-person meetings, and floor lead discretion. Low penetration of Ethereum and crypto knowledge outside Floor 12 was a factor that surprised the organizers, given that the building is marketed as the first Ethereum community hub in the United States. On-chain tools added complexity without sufficient incentive for non-crypto-native floors to learn them, especially with small budgets at stake.

Simocracy, Hypercerts, and the AI agent were deployed but not central to floor governance. Simocracy and Hypercerts were integrated, but neither was the primary mechanism any floor used for allocation decisions. The Frontier Tower Agent was built during the hackathon rather than being operational beforehand. All three became central post-event, in the community treasury infrastructure that follows this experiment. The full René backstory and the agent's deployment are documented in Appendix C and Section 8 Lesson 8.

These deviations are not failures of the experiment. They are the experiment. That governance defaulted to off-chain processes despite on-chain tooling being available is itself a finding about adoption friction, not a flaw in the design. How communities govern when given resources and freedom, including whether they adopt the tools provided or default to what they already know, is precisely what this report documents.

What Was Funded and By Whom

The experiment was funded by two sponsors:

Octant (Golem Foundation): $5,000 directed to floor allocations, with a commitment to publish a comparative governance report within 30 days of the event.

Hypercerts Foundation: $5,000 directed to the tower-wide community treasury, to be governed collectively by floor leads through Simocracy and the Frontier Tower agent.

Total external funding: $10,000. FtC contributed an additional $5,300 from event revenue and operational budget. Of the combined total, $10,300 was distributed to participating floors. Separately, $8,000 was committed to a tower-wide community treasury, sourced from $3,000 in event profits (2% of approximately $150,000 in event revenue) and the $5,000 Hypercerts Foundation allocation. As of this writing, the $8,000 is committed but not yet deposited; the treasury infrastructure is still being integrated with the Simocracy system. An additional $1,200 is being distributed to floor leads for survey completion incentives ($200 per floor lead).

Per-Floor Allocations

What This Report Documents

The experiment tested floor-level governance: how six communities allocated seed funding over a two-day event. That is the comparative dataset this report analyzes (Sections 5 through 8). Two additional infrastructure layers were designed around the experiment but were not the controlled experiment itself: a tower-wide community treasury, and an AI agent deploying as a co-signer alongside human-governed floors. Both are follow-on infrastructure. Section 9 elaborates, and presents observational evidence relevant to the building-level hypothesis described in The Setting, framed as suggestive data for future research rather than as proven findings, given that the experiment was not designed as a controlled comparison of building business models. Appendix B documents the full treasury architecture as originally designed.

Research Questions

With 6 participating floors and the data collected through floor lead debriefs, event records, platform data, and post-event infrastructure, the report investigates:

Mechanism diversity. What governance mechanisms did different floor communities choose, and why? Did any floors design novel approaches, or did they default to existing community patterns?

Participation depth. How many people in each floor community participated in allocation decisions? Was governance concentrated in a single leader or distributed across the community?

Spending patterns. What did different communities prioritize with their resources? Do technical floors spend differently than creative or wellness communities? What share went to permanent infrastructure versus one-time event costs?

Cross-floor contribution. Did floors that participated in the governance experiment also contribute value beyond their own space? Is there a relationship between governance engagement and building-wide collaboration?

The on-chain gap. Why did governance stay mostly off-chain despite the availability of on-chain tools? What does this tell us about minimum viable governance tooling for non-crypto-native communities, and what level of budget makes the friction of learning new tools worthwhile?

Three additional dimensions emerged as findings during analysis rather than as pre-specified research questions: transparency as a measure of process formality (Section 6), non-participation and disengagement (Sections 5 and 8), and self-funding as a confounding variable (Sections 6 and 8).

Floor-by-Floor Analysis

Six floors participated. The table below is the core of this section; full narratives for each floor are in Appendix A. Three floors (14, 8, 9) have complete debrief and transcript data; three (6, 12, 11) combine debrief data with platform records and pre-event workspace documentation.

Mingzhu He (Floor 14, Human Flourishing) in conversation with David Dao (Protocol Labs Research, builder of Simocracy) during the conference weekend.

Floor 14: Human Flourishing (Mingzhu He & Judy Zhou). Most-programmed floor by a wide margin: 47 sessions on Floor 14 plus 31 on Floor 16 (Earth Commons), roughly 34% of all sessions at the event. The community entered with pre-existing governance: three standing circles (People, Programs, Place), a documented backlog of community priorities, and months of prior coordination practice. The council was real (12–14 active members), but purchasing decisions collapsed to 3–5 people (full preference-aggregation story in Appendix A). Spending skewed heavily to permanent materials (~72%: sound panels, recording devices, a laptop), with members self-funding beyond the allocation as a community pattern one member flagged at the March 24 retro. Floor 14 was the only floor to build new governance tooling during the experiment (a participatory budget tool integrated with Hypercerts). Cross-floor contribution score: 14–15/18, the highest. The floor externally described as having the most advanced governance also produced the most candid documentation of where participation broke down within its own community. Effective oligarchy and participatory aspiration coexisted in the same community.

Floor 9: AI & Autonomous Systems (Devinder Sodhi). Benevolent dictator, most direct description in the dataset: "I knew I needed to do more work on the Annex and the 10 Floor Lab anyway." One person identified needs, made decisions, executed. Devinder spent roughly $6,000 against a $1,400 allocation, self-funding ~$4,600 to compensate contributors and buy equipment for Floors 9 and 10 (full breakdown in Appendix A). Output was 22 platform sessions plus ~25 volunteers deployed building-wide; Devinder was a top sponsorship lead generator alongside Dhruv from Floor 4. Cross-floor contribution: 13–14/18. What looks like "benevolent dictator governance of $1,400" reads better as "one motivated person investing $6,000", the self-funding confound is analyzed in Section 8 Lesson 6.

Floor 6: Arts & Music (Gage Olesen). Benevolent dictator, event-coordinator logic: "We booked an event, so we used an event coordinator to coordinate it." Three allocation participants, transparency 2/5. Floor 6 received the largest seed ($3,000) and spent 100% on artist compensation, the only floor where the entire allocation went to a single category, and the only floor where that category was people rather than things. The floor's AV equipment was deployed to every stage in the building, making Floor 6's internal decisions a form of infrastructure provision for the whole event. Whether that deployment was a floor decision or a unilateral call by the lead is the kind of question future experiments should track: when one floor's spending creates shared infrastructure, the governance has consequences well beyond the floor. Output: 25 platform sessions; cross-floor contribution 11–12/18. In his debrief, Gage flagged his own limitation: given more time, he would have opened the floor to programming proposals from other members.

Floor 12: Ethereum & Decentralized Tech (Gökhan Er, Mia Deng, Yoofi Annan). Delegation to intern. The debrief documents leadership transition as the operative governance fact: Mia Deng was out, Gökhan Er was running ETH SF the same week, and Yoofi was onboarded late without having attended early coordination meetings. Gökhan delegated budget authority; Yoofi allocated "meritocratically" based on what he assessed the floor needed. One person decided, transparency 2/5. The delegation held in routine matters but reverted in at least one higher-stakes instance, when a decision diverged from the original lead's preferences. Section 9 names this pattern conditional delegation; Lesson 10 develops the design implication. The pre-event workspace documented ambitious plans (a $13,000 quadratic funding match pool, full multisig treasury, Agentic Governance Brief) that did not materialize. When the people who designed a governance model are not the people executing it, the design often does not survive the handoff. Output: 25 platform sessions; cross-floor contribution 7-8/18.

Floor 11: Health & Longevity (Laurence Ion). Benevolent dictator with informal consultation, chosen because "pre-existing and we didn't have time to set up anything formal." Four participants, transparency 2/5. Laurence offered payment terms to top contributors and they agreed, a pattern that produced the second-highest session count (28) but the lowest cross-floor contribution score (3–4/18). Floor 11 spent ~$1,900 against a $1,400 allocation (self-funded ~$500), with ~79% going to compensating people. Laurence was the only lead to articulate a forward governance pathway rather than either defending or replacing the current mechanism: "Progressive decentralization! Expand from 1 to 3 people and keep expanding from there." The contrast between Floor 11's near-blank workspace documentation and its strong programming output is a useful caution against using written documentation as a proxy for governance quality.

Floor 8: Neuro & Biotech (Elliot Roth). Default to nonprofit. Zero governance participants, transparency 5/5, because there was nothing to observe. $700 allocation (half, minimal session commitment), $150 spent on workshop supplies, remainder returned to the floor's nonprofit. One session, one facilitator, no cross-floor operational support, no sponsorship leads. Floor 8 is a control, not a failure case. The comparison with Floor 14 is the starkest in the dataset (12–14 governance participants vs. 0; 78 sessions vs. 1; $2,400 vs. $700), and the question it surfaces is what conditions explain the 78x difference in session output: pre-existing infrastructure, leadership energy, community size, domain culture, or budget alone.

Floors that did not participate. Floor 7's lead (Tony) stopped attending coordination meetings before the event; that coordination gap cost René $2,500 when he couldn't verify the building's exclusive alcohol policy, and René built the Frontier Tower Agent in direct response. Floor 4 (Robotics/Physical AI, Dhruv Diddi) focused on the hackathon, providing space, robots, and mentorship, and generating sponsorship leads comparable to Floor 9. Governance is one channel through which floors contribute to a building-wide commons; production is another.

Comparative Analysis

Six floors. Six governance mechanisms. Same building, same weekend, same attendee pool. This section compares what happened across floors and tests what, if anything, the data can tell us about the relationship between governance design and outcomes.

A necessary caveat before the analysis: N=6, over two days, with budgets smaller than planned. The patterns described here are observations, not statistically significant findings. With six data points, any correlation could be coincidence. We report them because they are suggestive enough to warrant investigation in future experiments with larger samples, and because even pattern-level findings from a comparative governance dataset this unusual are worth documenting.

The Mechanism Comparison

Three of six floors chose benevolent dictator governance, one used a council, one delegated to an individual, and one defaulted to an existing organizational structure. No floor chose quadratic funding, conviction voting, participatory budgeting, or any mechanism from the governance design literature, despite the building housing an Ethereum community hub with deep fluency in those concepts.

The difference across floors lay in how many people influenced the person making decisions, Floor 14's council of 12-14 versus Floor 12's single delegate versus Floor 8's zero participants.

Spending Patterns

How floors spent their allocations reveals a rough taxonomy of governance priorities.

Two patterns emerge. First, the people-vs-infrastructure split: Floor 14 invested primarily in permanent materials (sound panels, recording devices, a laptop), while Floors 6 and 11 invested primarily in compensating people. Floor 9 split between the two. This maps loosely onto floor culture: the community with pre-existing governance infrastructure (Floor 14) spent on physical infrastructure; the communities with centralized leadership spent on the people doing the work.

Second, self-funding distorts the comparison. Three of six floors had leads or members who spent personal money beyond the allocation. Floor 9's actual operating budget was roughly 4x its seed allocation. Floor 11 exceeded its allocation by ~$500. Floor 14 members self-funded as a community pattern. When the analysis compares "outcomes per dollar of seed funding," it must account for the fact that some floors were operating with substantially more than the seed.

Cross-Floor Contribution Scoring

The cross-floor contribution score measures value created beyond a floor's own space. Scored 0-3 across six dimensions (max 18), using self-reported data from debrief forms, platform data, and organizer observation.

Scores are estimates with partial data; ranges reflect uncertainty in some dimensions.

The spread is wide: Floor 14 scored roughly 7-8x Floor 8's contribution. The top three contributors (F14, F9, F6) account for the vast majority of cross-floor value, and each contributed in different ways: Floor 14 through programming and mentorship, Floor 9 through volunteer deployment and sponsorship, Floor 6 through AV infrastructure.

Five Correlations Tested

The research plan specified five relationships to investigate. Here is what the data shows, with appropriate caveats about small sample size.

1. Governance participation vs. programming output

Finding: Weak positive relationship, heavily confounded by pre-existing infrastructure.

Floor 14 (12-14 participants, 78 sessions) and Floor 8 (0 participants, 1 session) sit at opposite extremes, which suggests a correlation. The middle floors do not follow the pattern: Floor 11 (4 participants, 28 sessions), Floor 6 (3, 25), Floor 12 (1, 25), Floor 9 (1, 22). Programming output appears driven by community size, leadership energy, and pre-existing organizational capacity rather than by budget-decision participation. Floor 14 had all three. Floor 8 had none. The governance mechanism was downstream of those factors.

2. Budget size vs. governance complexity

Finding: Suggestive but not clean. Budget size correlated with governance engagement more than governance complexity.

No floor at any budget level chose a complex governance mechanism. Floor 6 ($3,000) used benevolent dictator governance, the same as Floors 11 and 9 ($1,400 each). Floor 14 ($2,400) was the only floor with multi-person governance, but that structure pre-existed the experiment and would have operated the same way at any budget level.

What budget size affected was engagement. Floor 8 ($700) had zero governance participants; floor leads who expected $3,000-$4,000 and received $1,400 described feeling demoralized. On-chain tooling adoption was also budget-related: no floor considered the tools worth the setup cost at $1,400 (see Section 7).

3. Transparency vs. governance effectiveness

Finding: Inverse relationship in this dataset. The most transparent floor produced the least output.

Floor 8 scored 5/5 on transparency and produced 1 session with 0 governance participants. Floor 14 scored 3/5 and produced the most programming of any floor. Floors 6, 11, and 12 all scored 2/5 and produced between 25-28 sessions each.

This is not evidence that opacity produces better governance. Transparency scores measure process formality, not governance quality. When there is nothing to observe (Floor 8), everything is technically visible. When governance involves active deliberation with disagreements (Floor 14), the process is inherently messier to score as fully transparent. Floors with formal processes are more aware of where their transparency falls short; floors with no process score themselves highly because there is no gap between what happened and what was visible.

4. Cross-floor contribution vs. governance engagement

Finding: Strong positive relationship, but causation is unclear.

The three floors with the highest cross-floor scores (F14: 14-15, F9: 13-14, F6: 11-12) were the three floors whose leads attended 75%+ of coordination meetings. The three lowest (F12: 7-8, F11: 3-4, F8: 1-2) had lower or later attendance. Meeting attendance may be a proxy for a deeper trait (leads invested enough to show up are invested enough to contribute beyond their own space), or the meetings themselves may be causal, since they are where cross-floor relationships form and needs get communicated. Both explanations are consistent with the data.

5. Governance mechanism type vs. what the money was spent on

Finding: The mechanism did not determine spending priorities. Floor culture did.

All three benevolent dictators spent their money on completely different things: Floor 6 on artists, Floor 9 on labor and equipment, Floor 11 on compensating contributors. The council governance at Floor 14 produced the most infrastructure-heavy spending. Floor identity predicted spending patterns more reliably than governance structure. The arts floor paid artists; the AI floor bought equipment; the wellness floor compensated practitioners; the flourishing floor invested in infrastructure. Governance mechanisms shaped how decisions were made, not what decisions were made.

The most participatory governance moment in the experiment happened after the event ended. Section 7 documents it.

The Governance Spectrum

What the experiment observed was a spectrum ranging from "one person decides everything" to "one person decides after consulting several others." Under the conditions that held, small budgets, short timelines, high operational pressure, no dedicated governance facilitator, concentrated authority was the rational default. Whether different conditions (larger budgets, longer design phases, on-site facilitation, on-chain tooling) would produce genuinely different mechanisms is the question for future experiments.

On-Chain Infrastructure & Tooling

The experiment was designed to run on Ethereum. It mostly ran on WhatsApp.

That framing is not a complaint. The tooling existed and was available; floors were free to use it. The gap between the on-chain design and the off-chain reality is itself the finding this section documents. What was deployed, what was not, and why governance stayed off-chain despite on-chain tools being available is a story about adoption conditions, not about the tools themselves. Full deployment details are in Appendix C; this section summarizes what ran and carries the argument about why.

What ran, what didn't

Simocracy (simocracy.org/ftc-sf) was deployed with digital twins for all eight floor communities, built through 30-minute interviews with floor leads conducted by David Dao and Sejal Rekhan. Engagement during the event was ample, with 18 proposal submissions submitted by floors, and a number of sims created by participants of the festival. The system's climactic moment came on March 16, the day after the event ended, when all floor leads used it to collectively allocate the $5,000 Hypercerts Foundation contribution.

Hypercerts are a contribution-attestation protocol from the Hypercerts Foundation. Unlike NFTs, which track ownership of a unique digital asset, Hypercerts encode structured records of who contributed what to a piece of work, with attribution split among multiple parties and verifiable on-chain. In this experiment, James Farrell integrated Hypercerts into the FtC event platform; all 176 speakers received Hypercerts documenting their participation, published March 17 via the AT Protocol. Floor 14's participatory budget tool minted proposals as Hypercerts, the only floor to use the protocol for internal allocation decisions.

The FtC event platform (platform.fundingthecommons.io) was the only digital tool with universal adoption across all six participating floors. It served an immediate practical need (scheduling 228 sessions) rather than requiring floors to learn a new governance paradigm. This pattern, where practical coordination tools achieve adoption and governance-specific tools do not, is one of the experiment's clearest findings about tooling.

The Frontier Tower Agent (Frontier Road, frontierroad.replit.app) was not operational during the event. It was built during the hackathon and is deploying post-event alongside the community treasury infrastructure.

On-chain floor wallets and per-floor transaction records never materialized. Floor-level governance happened through WhatsApp, Telegram, and in-person meetings. The tower-wide Gnosis Safe treasury is pending deposit and integration with Simocracy; status is in Section 9.

The March 16 Simocracy session

The most participatory governance moment in the experiment occurred the day after the event ended. Floor leads met with David Dao on Floor 14 and used Simocracy to collectively allocate the $5,000 Hypercerts Foundation contribution through the S-Process (mechanics in Appendix C).

Figure 1. Simocracy in session, March 16, 2026. Sims walk a digital twin of Frontier Tower while the S-Process allocates $4,270 of $5,000 across eight floor-lead proposals (Gifting Games, Regen Room, Interactive AI Mural, Motion Capture, Ethereum Node, Flourishing Endowment, Vertical Garden). $730 held back for future rounds.

Figure 2. David Dao demonstrating Simocracy to floor leads, walking through the digital twin tower and the S-Process allocation. The interview process that produced each floor lead's sim ran for thirty minutes per participant.

The AI distributed approximately $4,695 across the proposals and held back $305 for future rounds, judging the current proposal set insufficient to warrant spending the full amount. The largest allocations (Gifting Games, motion capture, the Regen Room) roughly matched the group's top human picks. But the AI deprioritized infrastructure and endowment proposals because it judged that small amounts would not meaningfully advance them, reproducing the same logic floor leads had used when governing $1,400 allocations. Small budgets produce cautious governance regardless of whether the governor is human or algorithmic.

The group voted. Multiple participants preferred the human distribution, one of whom would have received most funding for her favorite proposal if human distribution would have been chosen. Zero voted for the AI-only allocation. Several preferred holding all $5,000 for a future round on the argument that the proposal quality did not warrant funding at any split. That argument surfaced the session's central insight: the binding constraint was the quality of the proposal set, not the evaluation mechanism.

The direction several participants preferred for ongoing treasury governance was an AI-advised, human-run model: AI reads all proposals carefully, explains its reasoning, and surfaces tradeoffs humans cannot sustain at scale; a human holds final allocation authority. The alternative (six hours of weekly meetings to read through proposals) is not realistically sustainable. The AI makes a governance process possible that otherwise would not exist.

Rather than distributing the $5,000 to individual floors in this round, the group pooled it into the tower-wide community treasury and committed to govern it through Simocracy going forward. Appendix C documents the full Phase 1 / Phase 2 session structure and per-proposal allocations.

Why governance stayed off-chain

Five factors contributed to off-chain governance being the universal default:

Budget size relative to tooling friction. At $1,400 per floor, the time cost of setting up on-chain wallets, managing keys, and executing transactions on Ethereum exceeds the perceived benefit. Multiple floor leads described the budget as too small to justify learning new tools. The on-chain governance thesis likely requires a minimum budget threshold, above which governance tools start earning their complexity.

Crypto literacy concentrated on Floor 12. Despite the building being marketed as the first Ethereum community hub in the United States, crypto fluency barely extended beyond the Ethereum floor. Leads from wellness, biotech, arts, and longevity communities had limited experience with wallets or on-chain governance tools. Several had barely used Coinbase.

No dedicated governance facilitator. On-chain tooling does not adopt itself. FtC attempted to confirm leading governance experts to serve as an on-site facilitator but could not secure their participation due to time constraints. Without a dedicated product owner walking non-technical leads through wallet setup and transaction workflows, adoption was left to individual initiative. No floor took that initiative independently.

Informal coordination is fast; on-chain coordination is slow. Floor leads needed to make spending decisions quickly during a two-day event. A WhatsApp poll or hallway conversation takes minutes. Setting up a governance proposal on-chain, waiting for signers, and executing a transaction takes longer and requires technical overhead that does not serve urgent operational needs.

No forcing function. The experiment required transparency and documentation, but not on-chain execution. Given a choice between familiar tools and unfamiliar ones, every floor chose familiar. The only on-chain adoption that occurred was either externally deployed (Simocracy by David Dao, Hypercerts by James Farrell) or built by a floor that already had governance infrastructure and technical sophistication (Floor 14's custom Hypercerts tool).

For future experiments, these five levers need to move together. Raising the budget alone will not overcome the crypto-literacy gap. Providing a facilitator without raising the budget produces better-informed refusal. Making on-chain execution mandatory without reducing tooling friction produces either non-compliance or bad UX. The combination of small budgets, optional tools, no facilitator, and non-crypto-native participants produced predictable results: governance happened where the people already were.

Lessons & Failure Modes

Hackathon builders at the event. Lesson 8 documents how the Frontier Tower Agent (René) was built during this hackathon in response to a coordination failure that occurred earlier in the week.

This section documents what we learned, what we would do differently, and where governance broke down or defaulted to convenience. The lessons are grounded in specific data from the experiment, not abstract principles. Where the findings have implications for future governance experiments beyond Frontier Tower, we say so.

Lesson 1: Governance experiments need an anchor technology partner, not just funding

The change we would prioritize is securing an anchor governance technology partner before the experiment begins. The original design assumed the Ethereum Foundation would serve this role: funding, tooling access, institutional credibility. When the EF grant was denied, no single partner replaced that function. Octant and Hypercerts Foundation provided funding. David Dao deployed Simocracy. James Farrell integrated Hypercerts into the FtC platform. But these contributions arrived late and were layered onto an experiment already designed around a different anchor. What generic funding does not replace is a partner that brings a platform floor leads would be onboarded onto from day one, a product team providing implementation support, and institutional motivation to make the experiment succeed because their technology is being tested.

For future experiments: Confirm the anchor governance technology partner before scoping the experiment. Let that partner's tooling shape the experiment design, not the other way around.

Lesson 2: The six weeks before the event mattered more than the two days during it

The event gave plenty of data, but the governance experiment was not a two-day exercise. It ran for approximately six weeks, from the first coordination meeting in late January through the event on March 14–15. The patterns that played out on the event weekend were established weeks earlier: Floor 14's council governance was activated by early February, Devinder's benevolent dictator pattern on Floor 9 by mid-February, Floor 7's disengagement by early March. Floors that governed well during the event were the ones that had been governing (or at least meeting and planning) for weeks. Floors that did not engage during the lead-up did not suddenly activate during the event.

The event is the showcase. The experiment is the lead-up. Our six weeks were compressed by late-arriving tooling partners and the budget uncertainty created by the EF grant denial; with a confirmed anchor partner and secured budget from the start, the design phase could have been more deliberate.

For future experiments: Budget at least 8–12 weeks of active governance design before the event window.

Lesson 3: Budget size determines governance ambition

Floor leads who expected $3,000–$4,000 received $1,400. Several described feeling demoralized. This was not a minor detail; it shaped how much energy floor leads invested in governance design. When the budget is small, the rational response is to minimize governance overhead: one person decides, money gets spent on obvious needs, the community moves on. Floors 8 and 9 are the clearest cases. Floor 14 is the partial exception: their engagement was high despite a comparable budget, but Floor 14 entered with pre-existing governance infrastructure and leadership intrinsically motivated by the experiment regardless of budget size. Most communities do not start from that position. The same budget logic shapes on-chain tooling adoption (see Section 7).

For future experiments: The minimum budget to produce sustained governance engagement is likely $3,000–$5,000 per participating community. Below that threshold, expect rational defaults to minimal governance.

Lesson 4: Ethereum's reach problem is real

This finding is politically sensitive, and we are reporting it because it is relevant.

The building is marketed as the first Ethereum community hub in the United States. Floor 12 houses the Ethereum community, received a separate Ethereum Foundation grant, and has floor leads fluent in governance design concepts. Despite this, Ethereum and crypto knowledge was concentrated almost entirely on Floor 12. Floor leads from wellness, biotech, arts, and longevity communities had limited experience with wallets, transactions, or on-chain governance tools. Several had barely used Coinbase. The Ethereum floor's presence had not translated into broader Ethereum adoption throughout the building.

The gap is not abstract. During pre-event planning, we identified a team in the building that had deployed an autonomous vending machine system but could not access bank accounts for their AI agents. One floor away, the Ethereum community had exactly the tools to solve that problem: programmable wallets, smart contract-based treasury management, cryptographically bounded agent authority. The connection was never made. Meanwhile, during the weeks surrounding Intelligence at the Frontier, AI developer communities in San Francisco were actively integrating Solana for agent-to-agent payments, and Google Pay was becoming a default for agentic commerce. These choices were being made inside the building that houses Ethereum's own community hub.

The governance experiment was designed, in part, to address this: by putting real Ethereum infrastructure in the hands of non-crypto-native AI and robotics builders, the experiment would have demonstrated Ethereum's relevance to problems they already face. The Ethereum Foundation declined to fund the experiment. The author believes this was a missed opportunity, on the assumption that funding at the planned level would have addressed three constraints documented elsewhere in this report: budget-driven demoralization (Lesson 3), the absence of institutional signal that might have motivated floor leads to take on-chain tooling more seriously, and the absence of a dedicated governance facilitator (Lesson 1). Whether the grant would have produced those outcomes is unknowable. The governance infrastructure was designed from the start to be replicable: Frontier Tower's management company is exploring expansion to other cities, and the treasury model, incentive alignment structure, AI agent layer, and governance playbooks were designed as a deployable template for Ethereum community hubs globally.

The Foundation linked FtC's application with Floor 12's existing grant and assessed that it had already invested sufficiently in the building. Floor 12 produced the conditional delegation pattern documented in Section 9: authority delegated under low stakes, reclaimed when decisions diverged from leadership preferences. The experiment designed to extend Ethereum's tools beyond Floor 12 to the AI builders and other communities in the building was not funded.

This is not primarily a criticism of Floor 12, whose pre-event workspace documented governance designs (Section 5, Appendix A) and whose floor hosted some of the most well-attended sessions. Nor is it primarily a criticism of the Ethereum Foundation, which makes hundreds of funding decisions with imperfect information. It is an observation about a systemic gap: between Ethereum's physical presence in a building and actual adoption by builders in that building, between governance theory and governance practice, and between Ethereum Foundation’s stated goal of reaching AI developers and the institutional decisions that would make that goal achievable. Industry observers have argued that the next 12 to 18 months are decisive for which chains become default infrastructure for AI agent coordination, with Solana, Google Pay, and other platforms actively competing for that role. The builders making those choices are, in some cases, in this building. The experiment did not produce conditions under which they engaged with Ethereum tooling.

For future experiments: Do not assume that proximity to a crypto-native community translates to crypto literacy. On-chain governance experiments with non-crypto-native participants require dedicated onboarding, managed wallets, and a governance facilitator who can bridge the gap. And the institutions that fund governance infrastructure should understand that the highest-value experiments are often the ones running where their technology has not yet penetrated, not the ones preaching to communities already converted.

Lesson 5: Effective governance and distributed governance are not the same thing

Floor 14 led on programming output (78 sessions), cross-floor contribution score, and was the only floor to build new governance tooling. Floor 9 led on volunteer mobilization and sponsorship revenue, including the $3,000 commission earned from a single introduction. But their governance was not distributed in the way the experiment intended to test. WhatsApp polls reached 5 voters out of 60+ community members; purchasing decisions were made by 3–5 people. The floor described its own governance as "value-based rather than process-based." This is a motivated oligarchy that produced excellent results.

Meanwhile, Floor 8 scored 5/5 on transparency because there was nothing to observe. Floor 9's benevolent dictator delivered the broadest cross-floor operational contribution (see Section 5).

The experiment did not demonstrate that participatory governance produces better outcomes than concentrated authority. With N=6, it cannot. What it demonstrated is that the relationship between governance structure and output is mediated by pre-existing infrastructure, leadership energy, and domain culture in ways comparable to or larger than mechanism choice.

For future experiments: Distinguish governance effectiveness (did good things get funded?) from governance distribution (did many people participate?). They are different metrics and may trade off against each other.

Lesson 6: Self-funding breaks the experiment's constraints

Several floor leads spent personal funds beyond their allocations because they were also producing the festival itself. The governance experiment ran inside a $150,000 production, and operational pressure surfaced unbudgeted costs (equipment, AV, last-minute programming gaps). Floor and festival budgets blurred at the operational level.

At least two floors had participants who spent significantly beyond the seed allocation out of pocket. Devinder (Floor 9) invested approximately $4,600 beyond his $1,400 allocation. Floor 14 members self-funded expenses not approved through the governance process, a pattern one member described as a community tendency.

When a floor lead self-funds the majority of operations, the governance question ("how do we decide?") becomes moot. Devinder did not need to convene a process because he was spending his own money. This is the clearest path to benevolent dictatorship: personal financial investment creates both the authority and the justification for unilateral decision-making. Floor 9's output cannot be attributed to a governance mechanism governing $1,400; it is better attributed to a person investing $6,000.

For future experiments: Track self-funding explicitly. It is a confounding variable that makes mechanism comparison unreliable unless controlled for.

Lesson 7: Coordination meeting attendance predicts everything

The simplest predictor of a floor's event contribution was whether its lead attended the weekly coordination meetings. Floors that attended most or all meetings (14, 9) produced the highest session counts, the most cross-floor contribution, and the strongest governance engagement. Floors that attended "some" (8, 12) produced either minimal governance or governance that operated as conditional delegation under stress. Floor 7, whose lead stopped attending entirely, dropped out.

Meeting attendance is a proxy for commitment, but it may also be causal: the meetings were where cross-floor relationships formed, where governance design was discussed, and where FtC communicated tooling options and experiment expectations. Floor leads who skipped meetings missed the context that would have grounded their governance participation.

For future experiments: Make coordination meeting attendance a condition of receiving the full seed allocation.

Lesson 8: Consequential infrastructure came from a documented failure

René spent $2,500 on alcohol for an investor event, then learned the building's exclusive vendor contract prohibited outside alcohol service that date. His floor lead (Tony, Floor 7) had been skipping the meetings where the policy was communicated. René had no way to independently verify the information. The next day, René built the Frontier Tower Agent, the hackathon-winning coordination tool designed to close exactly this kind of information gap. Jakob (Frontier Tower co-founder) watched the pitch and announced FrontierOS. René received a $500 prize and an $1,800 annual membership.

The tool that would have prevented the coordination failures documented throughout this report was built by someone who suffered from one of them. This is the literal sequence of events. Floor 14's governance pre-existed the experiment because the community had already experienced coordination problems and built solutions. The Frontier Tower Agent was built because René lost money. Communities that had not experienced painful coordination failures did not build governance tools during the experiment.

For future experiments: Document failure cases. They are the strongest generators of governance innovation.

Lesson 9: Distributed micro-budgets to community leads outperformed contracted staffing

This finding sits outside the experiment's stated design but is consistent enough with the data to warrant documentation.

The $10,300 distributed across six community leads produced more programming hours, more volunteer mobilization, and more sponsorship lead generation per dollar than any other line item in the event's $150,000 budget. The same dollars spent on contracted staffing produced significantly less. Media contractor, deco and graphic design, speaker curation contractor, marketing consultant, and remote operations support together cost approximately $50,000 and produced significantly narrower outputs measured by the same axes. Floor 9's lead alone generated a $3,000 sponsorship commission for the Bittensor introduction, a single deal that exceeded twice his floor's seed allocation and covered the entire cost of the allocations used to run this experiment. Floor 4's lead generated $2,250 across two introductions. Floor leads collectively produced over 150 sessions, deployed roughly 25 volunteers building-wide, and ran AV infrastructure on every stage except Floor 2 (Funding the Commons stage).

The pattern is conditional. Three factors were present in this experiment that may not hold in others: communities with permanent physical space and member-based business models (which gave floor leads a reason to invest beyond the seed allocation), mission alignment that positioned Funding the Commons as a credible non-corporate convener (which made participation feel reciprocal rather than transactional), and brand exposure floor leads could convert into community membership growth (which created a return on labor invested). Where these conditions hold, the decentralized production model is replicable. Where they do not, the comparison breaks down.

In the decade that the author of this report spent producing music festivals at scale, this pattern was not a design choice; it was the only way to run a festival without contractor budgets. Many festival organizers know this intuitively. The experiment at Frontier Tower is, in this regard, a documentation exercise for a pattern many event producers already operate by but few have recorded.

For future experiments: Distributing micro-budgets to embedded community leads is a more efficient production model for community-embedded events than traditional contracting, conditional on the three factors named above. Future research should test the model in adjacent contexts: residency programs, university campus events, civic convenings, and multi-stakeholder convenings where participating communities already have skin in the game.

Lesson 10: Delegation cannot be evaluated under low-stakes conditions alone

Delegation governance held while routine. In higher-stakes instances at both the floor and building levels, decisions diverging from the original lead's preferences were reversed or vetoed. Section 9 names this pattern conditional delegation and develops it in full.

The design implication, developed in Section 9, is that without pre-agreed rules for when override is permitted, every delegation is conditional, and the test comes only when stakes rise.

For future experiments: Design delegation mechanisms with explicit, pre-agreed rules for when override is permitted. Single-event experiments cannot reliably generate the high-stakes test; longer-running experiments with deliberately introduced high-stakes decisions are needed to evaluate whether a delegation mechanism is genuine.

Forward: The Permanent Treasury

A floor session during the conference weekend. The communities documented in this report continue to operate the floors after the experiment ended; the permanent governance layer is being built around them.

The governance experiment at Intelligence at the Frontier was designed to activate infrastructure, not produce a one-time report. Two weeks after the event ended, the infrastructure is taking shape.

What Exists Now

The $8,000 community treasury is committed: $3,000 from event profits (2% of approximately $150,000 in revenue) and $5,000 from the Hypercerts Foundation. The funds are not yet deposited because the governance layer that will manage them is still being built. That governance layer is the point.

The Simocracy system is live at simocracy.org/ftc-sf, with sims created for each of the eight floor communities. Frontier Road (the AI agent built by René during the hackathon) is live at frontierroad.replit.app. The two systems are being integrated so that the community treasury is governed by a structure where each floor lead's sim holds one vote, the Frontier Tower Agent holds one vote, and Funding the Commons (which will create its own sim) holds one vote. The technical implementation, whether through a Gnosis Safe multisig, a Simocracy Senate mechanism, or another multi-signature structure, is still being finalized. The governance architecture is clear; the specific technology is a detail being resolved.

The contractual foundation is in place. The FtC x Frontier Tower Event Partnership and Revenue Sharing Agreement establishes 2% of event revenue and 20% of event profit flowing to the community treasury as a standing commitment for tower-wide events. This is permanent funding, not grant funding that expires. It is a business model that generates its own sustainability as long as the building hosts collaborative events.

What Happens Next

Three things need to happen for the experiment to become permanent infrastructure:

First, the treasury needs to be deposited and governed. The $8,000 is committed. The Simocracy system and Frontier Road agent are live. Integrating these into a functional governance structure where floor leads, the AI agent, and FtC collectively manage real funds is the immediate next step. David Casey, based at Frontier Tower, is co-stewarding this process alongside David Dao and the floor leads.

Second, the governance model needs a second event to test repeatability. The 2%/20% revenue model generates treasury capital when the building hosts collaborative events. If the next tower-wide event feeds the same treasury through the same contractual mechanism, the model demonstrates sustainability beyond a single activation. FtC is evaluating a recurring programming series (Frontier Frequencies) and future conference programming at the tower that would feed the treasury on an ongoing basis.

Third, the model needs to be documented for replication. Frontier Tower's management company is exploring expansion to other cities. The governance infrastructure being piloted here, the treasury structure, the incentive alignment model, the AI agent layer, and the Simocracy governance process, is designed from the start to be deployable at other sites. Appendix D contains the practical materials. But the real test of replicability comes when a second community hub adopts and adapts the model.

The Larger Question

The experiment began with a practical problem: a building full of distinct communities with no shared financial infrastructure and no incentive to collaborate. That problem has been partially solved. A treasury exists. A governance mechanism exists. An AI agent exists. Floor leads have experience governing real funds and documented what worked and what did not.

The larger question is whether this model, shared treasury funded by collaborative activity, governed by a mix of human representatives and an AI agent, with transparent on-chain infrastructure, can work as a general pattern for multi-community coordination. Frontier Tower is one building in San Francisco. The design challenges that floor leads and building members collaboratively documented (fragmented communities, misaligned incentives, conditional delegation, underused collective capacity, see Appendix C) are not unique. They describe university campuses, innovation districts, co-working networks, and urban neighborhoods.

The conditional delegation pattern observed at Floor 12 was not unique to that floor. A version of the same pattern was observable at the level of the building itself. The building's governance is nominally distributed. Floors operate as autonomous communities with their own leadership, programming, and budgets. In practice, ultimate authority over building-level decisions, including staffing, platform access, vendor selection, and event policies, sits with management. When floor and event-partner decisions align with management preferences, autonomy holds. When they diverge, management retains the option to override. The override is structural, and it operated multiple times during the experiment's planning and execution. Whether each instance produced better or worse operational outcomes than the original decisions would have is contested between event partners and management. The structure at Floor 12, delegated authority that reverts when stakes rise, is the structure at the building scale, with the threshold for reclamation set by the principal rather than the participants.

We name this pattern conditional delegation: governance that appears participatory in routine matters and reverts to concentrated authority in consequential ones. The defining feature is that the threshold for reversion is unilateral.

For governance researchers, the design implication is that delegation mechanisms cannot be evaluated under low-stakes conditions alone. The diagnostic test is what happens when the delegate makes a decision the delegator disagrees with. The participatory language is real. The underlying authority is not relocated.

Observing this pattern at two scales within the same building suggests it is not a one-off. The conditions that produce it, distributed authority layered over principals who retain override power without bounded rules for when override is permitted, are common across many institutional contexts: building owners and tenant communities, university administrations and student-led affinity groups, parent corporations and regional offices, foundation funders and grantee programs.

The experiment generated data relevant to a building-level hypothesis that the floor-level governance comparison was not designed to test directly.

Building management collected rent from floor communities and a percentage of event revenue, the standard co-working model. Management benefited when any floor ran a successful event. The structural gap relevant here was at the floor layer: floors had no financial stake in each other's success, no incentive to collaborate on programming or cross-promote, and no mechanism by which value created jointly across the building accrued to a resource floors governed together. The contractual agreement under which the experiment ran introduced a shared treasury: 2% of event revenue and 20% of event profit flowed into a tower-wide community treasury co-signed by floor leads, building management, and Funding the Commons as co-steward. The hypothesis was that aligning financial incentives this way would convert the previously zero-sum floor-to-floor dynamic into positive-sum coordination.

The event ran under that structure for two days. Approximately 25% of event revenue (around $37,000) flowed back into the building through floor budgets, the community treasury, member contractor fees, referral fees, and management fees. Two floor leads earned $5,250 in sponsorship commissions across three introductions. Six first-time sponsor relationships were brought into the building. 25+ member conversions are projected over the 90 day window post-event, representing $45,000 or more in annualized membership revenue. One new sponsor is in active discussion about longer-term tenancy.

These outputs are suggestive, not dispositive. The experiment was not designed as a controlled comparison of building business models. Two days of operation under a new structure cannot establish whether the model produces better outcomes than the prior one at scale, given that the conditional delegation pattern described above remained available to building management throughout: the revenue-sharing structure was contractual but the override authority above it was not constrained by the contract. A test of the revenue-sharing model at scale would also be a test of whether the override pattern can be bounded by rules set in advance.

We treat this as future research. There is comparatively little controlled data on revenue-sharing community treasuries as alternatives to landlord-tenant or management-fee models for community hubs. We would welcome co-publication with a research partner whose primary focus is comparative business model evaluation in community-embedded organizations.

A second pattern, less central to the experiment's design but more immediately replicable, deserves separate attention. The micro-budget staffing & production model documented in Section 8 Lesson 9 is portable in ways the governance findings are not. Where a convening has community partners with permanent space, mission alignment with the host, and the ability to convert event exposure into community growth, distributing small budgets to those partners is more efficient than equivalent contracted staffing. This describes the operating conditions of United Nations country offices working with civil society organizations, university programs working with student-led affinity groups, and city governments working with neighborhood associations. The experiment did not test this hypothesis directly, but the data is consistent enough with a decade of festival production practice that institutional event producers and program designers should treat the pattern as worth piloting in their own contexts.

The experiment produced a dataset, not a proof (see Section 6 for the sample-size caveat). But it also produced infrastructure that continues operating and a community of floor leads who now have firsthand experience governing shared resources. The next iteration will run with better data, more time, and the advantage of knowing what happened when six floors tried this for the first time.

Author Contributions

This report follows the CRediT taxonomy (https://credit.niso.org/) for contributor attribution.

David Casey (Funding the Commons): Conceptualization, Methodology, Investigation, Formal analysis, Writing, Original Draft, Visualization, Supervision, Project administration, Funding acquisition.

James Farrell (Funding the Commons): Software, Data curation. Built the Funding the Commons event platform used to schedule, document, and analyze the 228 sessions and 176 speaker contributions referenced throughout the report; integrated Hypercerts to produce the speaker activity records.

David Dao (Protocol Labs / GainForest / AI4PG): Initial thought partner, Software, Investigation, Methodology. Built and operated Simocracy; designed and ran the S-Process allocation session on March 16, 2026 documented in Section 7 and Appendix C; co-developed the AI-assisted governance methodology adapted from the Survival and Flourishing Fund.

Sejal Rekhan: Investigation. Conducted floor lead interviews used to construct Simocracy sims.

Tereza Bízková (Funding the Commons): Writing, Review & Editing.

Kim Buisson (Funding the Commons): Project administration, Resources, Writing, Review & Editing.

Ryan Rising (Funding the Commons): Project administration, Resources.

Eugene Leventhal (Octant / Golem Foundation): Writing – Review & Editing, Funding acquisition. Octant co-funded the experiment and commissioned this comparative governance report.

Floor leads who completed debrief surveys (Investigation): Judy Zhou (Floor 14, Human Flourishing), Elliot Roth (Floor 8, Neuro & Biotech), Gage Olesen (Floor 6, Arts & Music), Yoofi Annan (Floor 12, Ethereum & Decentralized Tech), Laurence Ion (Floor 11, Health & Longevity), and Devinder Sodhi (Floor 9, AI & Autonomous Systems, whose data was compiled from transcripts and direct communication).

Use of AI Tools

Claude (Anthropic) was used by the author as a thinking partner, drafting assistant, and structural editor throughout the writing process. Claude contributed to section structure, comparative analysis frameworks, language refinement, and consistency checks across the document. The author retained full editorial control over all claims, data, and conclusions. All factual claims, dollar figures, governance descriptions, and analytical conclusions were authored or verified by the author. Floor lead debrief responses, transcripts, and platform data were authored entirely by the cited human contributors.

Acknowledgements & how to cite

Funders

Octant (Golem Foundation) directed $5,000 to the floor allocations that made the comparative governance experiment possible. The contribution carried a commitment to publish a comparative governance report within 30 days of the event; this paper is that report.

Hypercerts Foundation directed $5,000 to a tower-wide community treasury, governed collectively by floor leads through Simocracy and the Frontier Tower agent. The Hypercerts protocol also served as the contribution-attestation layer across the event.

Funding the Commons contributed $5,300 from event revenue and operational budget to the floor allocations, plus $3,000 (2% of event revenue) to the community treasury.

Methods, lineage, and tooling

The S-Process methodology used in the March 16 allocation session was developed by Andrew Critch at the Survival and Flourishing Fund, which has distributed $34.92M using AI-aggregated preference functions paired with human veto authority. Simocracy was built by David Dao as part of the AI4PG (AI for Public Goods) research initiative, a coalition including GainForest, Hypercerts Foundation, Octant, Seer, Funding the Commons, and Protocol Labs. Sejal Rekhan co-conducted the floor-lead interviews that produced the digital twins.

The Frontier Tower Agent (René, Frontier Road) was built during the conference hackathon and is deploying as the AI co-signer on the community treasury alongside human floor leads and Funding the Commons.

Author, contributors, and disclosures

David Casey is CEO of Funding the Commons and the author of this report. He served as both the producer of Intelligence at the Frontier and the designer of the comparative governance experiment, a dual role disclosed in Section III (Limitations). Floor lead relationships with the author may have influenced debrief responses; the report works to surface this rather than obscure it.

Contributors: Kim Buisson (COO, Funding the Commons) led the research framing, structural editing, and production of this report. Tereza Bízková (Head of Communications, Funding the Commons) led copy editing and voice review. David Dao (Protocol Labs Research, GainForest) co-led the methodology, built and operated Simocracy, and ran the March 16 S-Process allocation session documented in Section VII and Appendix C. Voice and structural editing was assisted by AI tools and reviewed by humans.

Reproducibility and reuse

Appendix D contains five replicable templates for future comparative governance experiments: governance mechanism menu, floor treasury setup checklist, lead debrief questionnaire, cross-community value contribution scoring rubric, and experiment design checklist. The structured floor debriefs and full event transcripts are available on request to accredited researchers and funders.

How to cite

Casey, David (2026). Comparative Floor Governance at Funding the Commons SF. Funding the Commons Research, 2026.01. April 2026.

Comments, corrections, and follow-up inquiries: contact@fundingthecommons.io.

References & primary sources

[1] Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

[2] Cox, M., Arnold, G. & Villamayor-Tomás, S. (2010). "A Review of Design Principles for Community-Based Natural Resource Management." Ecology and Society, 15(4), 38.

[3] Buterin, V., Hitzig, Z. & Weyl, E.G. (2019). "A Flexible Design for Funding Public Goods." Management Science, 65(11), 5171–5187.

[4] Golem Foundation. (2023). "Announcing Octant." Retrieved from https://golem.foundation/2023/08/08/announcing-octant.html

[5] Fritsch, R., Müller, M. & Wattenhofer, R. (2022). "Analyzing Voting Power in Decentralized Governance." Blockchain: Research and Applications, 5(3).

[6] Cong, L.W., Rabetti, D., Wang, C.C.Y. & Yan, Y. (2025). "Centralized Governance in Decentralized Organizations." Harvard Business School Working Paper.

[7] Fan, J., Ai, Y. & Liu, X. (2024). "DAO's Governance Paradox." 74th Annual ICA Conference.

[8] ChainScore Labs. Gitcoin Grants sybil analysis reports, Rounds 8–23.

[9] Optimism Foundation. (2024). "RetroPGF Round 3 Results." Retrieved from https://community.optimism.io

[10] Freeman, J. (1972). "The Tyranny of Structurelessness." Berkeley Journal of Sociology, 17, 151–165.

[11] Schneider, N. (2024). Governable Spaces: Democratic Design for Online Life. University of California Press.

[12] O'Mahony, S. & Ferraro, F. (2007). "The Emergence of Governance in an Open Source Community." Academy of Management Journal, 50(5), 1079–1106.

[13] Survival and Flourishing Fund. S-Process methodology documentation. Retrieved from https://survivalandflourishing.fund

[14] Samuelson, W. & Zeckhauser, R. (1988). "Status Quo Bias in Decision Making." Journal of Risk and Uncertainty, 1(1), 7–59.

[15] Iyengar, S.S. & Lepper, M.R. (2000). "When Choice Is Demotivating." Journal of Personality and Social Psychology, 79(6), 995–1006.

[16] Dunbar, R.I.M. (1993). "Coevolution of neocortical size, group size and language in humans." Behavioral and Brain Sciences, 16(4), 681–735.

[17] Hackman, J.R. (2002). Leading Teams: Setting the Stage for Great Performances. Harvard Business School Press.

[18] Putnam, R.D., Leonardi, R. & Nanetti, R.Y. (1993). Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press.

[19] Olson, M. (1965). The Logic of Collective Action: Public Goods and the Theory of Groups. Harvard University Press.

[20] Rozas, D., Tenorio-Fornés, A., Díaz-Molina, S. & Hassan, S. (2021). "When Ostrom Meets Blockchain." SAGE Open, 11(1).

[21] Barbereau, T. et al. (2022). "Decentralised Finance's Timocratic Governance." Technology in Society, 73.

[22] Hsiao, Y.-T. et al. (2018). "vTaiwan: An Empirical Study of Open Consultation Process in Taiwan." SocArXiv.

[23] Gudiño-Rosero, J.F., Grandi, U. & Hidalgo, C.A. (2024). "LLMs as agents for augmented democracy." Philosophical Transactions of the Royal Society A, 382(2285).

[24] Messias, J. et al. (2023). "Understanding Blockchain Governance." arXiv:2305.17655.

[25] Halfaker, A., Geiger, R.S., Morgan, J.T. & Riedl, J. (2013). "The Rise and Decline of an Open Collaboration System." American Behavioral Scientist, 57(5), 664–688.

Primary Data Sources

[P1] Floor lead debrief survey responses (5 of 6 respondents), March–April 2026. Unpublished primary data, Funding the Commons.

[P2] Pre-event floor lead coordination meeting transcripts, January–March 2026. Transcribed via Fireflies.ai.

[P3] Simocracy allocation session transcript, March 16, 2026. Conducted by David Dao with all floor leads.

[P4] FtC event platform session data, platform.fundingthecommons.io. Exported March 2026.

[P5] Luma ticket attribution and registration data. Exported March 2026.

[P6] Pre-event floor workspace documentation (Notion). Compiled March 2026.

[P7] FtC x Frontier Tower Event Partnership and Revenue Sharing Agreement. February 2026.

Full Floor Narratives

This appendix preserves the full narrative detail for each of the six participating floors. Section 5 carries the condensed summary and comparison table; the narratives below are the source material those summaries were drawn from, reproduced uncut for readers who want the full picture of how governance played out on each floor.

Floor 14: Human Flourishing (Mingzhu He & Judy Zhou)

Seed funding: $2,400 Sessions: ~78 across two floors (47 on Floor 14 + 31 on Floor 16 Earth Commons) Active governance participants: 12-14 (core group), but ~3-5 made purchasing decisions Transparency score: 3/5 (self-reported) Ticket attribution: 70 code-attributed registrations (45 checked in). Judy self-reported 95 comp tickets.

The Community

Floor 14 houses the Human Flourishing Foundation, a community organized around wellbeing, somatic practice, and regenerative systems. The floor operates with pre-existing governance infrastructure: three circles (People, Programs, Place) that function as standing committees, coordinated through a core group of 12-14 active members.

What distinguishes Floor 14 from every other floor in this experiment is that the community had already diagnosed the building's governance gaps before the experiment began. Their Notion workspace documents eight design challenges for "Frontier Tower as a Commons," including: no real-time decision-making over shared resources, no shared coordination mechanism, no transparency on finances, and power/responsibility/incentive mismatches across floors. This was not a community encountering governance for the first time. They had a framework, a language, and a backlog of ideas about how the building should work.

Governance Mechanism: Core Council with a Preference Aggregation Problem

Floor 14 used a council-based model built on their pre-existing committee structure. Three circles (People, Programs, Place) mapped directly onto budget categories. Decisions were coordinated through a core group of 12-14 members via WhatsApp and Zoom, drawing from a backlog of wishlist items maintained before the experiment.

The story underneath is more interesting than "pre-existing governance applied successfully."

The floor tried participatory decision-making and it failed. Mingzhu described the problem in a March 5 floor leads meeting: WhatsApp polls got only 5 voters out of a 60+ person floor community. "Because using WhatsApp polls don't work. Like five people vote on it and we just end up buying things anyway." The result was a de facto oligarchy of three or four people making all purchasing decisions.

David Casey suggested a quorum rule during the same meeting: set the threshold at 20%, and whoever shows up decides. "Law of two feet." Floor 14 adopted this approach.

Judy Zhou's debrief described the governance as "value-based rather than process-based," with a transparency score of 3/5. Some decisions were made last minute. Given more time, she said the floor would want broader participation.

The floor described externally as having the "most advanced and documented floor treasury governance system" by an external observer also produced the experiment's most explicit account of participation breakdowns in its own decision-making. The governance was effective (78 sessions, two floors programmed), but it was not distributed the way the experiment aimed to test. Whether effective oligarchy is a governance success depends entirely on what you're measuring.

Budget Allocation

Floor 14 received $2,400 in seed funding ($1,400 base plus $1,000 for producing Earth Commons programming on Floor 16). Spending breakdown from the debrief:

Sound panels: $600

Recording devices: $400

New laptop for AV: $200

Office supplies: $100

Earth Commons opening ceremony supplies: $150

Speaker stipends: ~$800 (estimated; may include overage covered out of pocket)

Roughly 72% of the reported allocation went to permanent materials and equipment rather than one-time event costs. This is the highest infrastructure-to-event ratio of any floor in the experiment. The floor leads covered overages out of pocket.

Floor 14 is also the only floor where self-funding beyond the allocation was described as a community pattern, not just a leadership decision. At the March 24 retro, a floor member noted: "I think our floor has the tendency to self fund things that aren't approved by the group." This suggests the governance mechanism (council approval) and the actual spending behavior (individuals buying what they thought was needed) diverged.

Programming Output

Floor 14 produced the most programming of any floor by a wide margin. 47 sessions ran on Floor 14 across Saturday and Sunday. On Sunday, the floor also took over programming for Floor 16 (the d/acc Lounge), running it as "Earth Commons" with an additional 31 sessions. Combined output: approximately 78 sessions across two floors, representing roughly 34% of all sessions at the entire event.

Judy reported 30+ hours of programming and 8-10 floor members who spoke or facilitated. Session formats ranged across panels, workshops, circles, healing sessions, embodiment practices, and the Earth Commons opening ceremony. The format diversity is itself a governance output: the circle structure allowed different committees to program in parallel without bottlenecking through a single curator.

Cross-Floor Contribution

Floor 14's contribution extended well beyond their own space. They programmed an entire additional floor (Floor 16 Earth Commons on Sunday), mentored the AI and Ethereum floors on programming and production, hosted a pre-FtC speaker happy hour and post-FtC afterparty, sourced vendors for the broader event, and generated sponsorship leads through speaker networks. They attended 75%+ of floor lead coordination meetings.

Preliminary cross-floor contribution score: 14-15 out of 18, the highest of any floor.

Tools and Infrastructure

Floor 14 is the only floor that actively built new governance tooling during the experiment. They used the FtC platform (described as "a little bit clunky"), engaged with Simocracy ("high engagement and enjoyed, looking forward to continued collaboration"), and built their own participatory budget tool integrated with Hypercerts. This organic governance innovation, building tooling rather than just using or ignoring what was provided, is unique in the dataset.

What Persists

Floor 14 is the strongest case for the experiment generating lasting infrastructure. The governance mechanisms were pre-existing and will continue. The physical equipment purchased with the seed funding remains on the floor. The Earth Commons programming created a template for cross-floor collaboration. Judy expressed clear interest in ongoing governance participation and in working with sponsors on future governance design experiments.

Floor 8: Neuro & Biotech (Elliot Roth)

Seed funding: $700 (half allocation; minimal session commitment) Sessions: 1 (~3 hours) Active governance participants: 0 Transparency score: 5/5 (self-reported) Ticket attribution: 2 code-attributed registrations (2 checked in)

The Community

Floor 8 houses a neurotech and biotech community organized as a nonprofit. Their workspace documented a vision around "enabling access to tools to heal people and the planet" with a four-segment framework (people + planet × discovery + play), but the operational planning for the experiment was minimal.

Governance Mechanism: Default to Nonprofit

Elliot Roth's debrief is the most concise in the dataset: "Default to all returning to the nonprofit." Zero people actively participated in allocation decisions. The mechanism was pre-existing and required no deliberation. Budget transparency scored 5/5, which in this context means the process was fully visible because there was effectively no process to observe.

Floor 8 received the lowest allocation ($700, half the standard) because they committed to only one workshop. The budget meeting transcript from March 17 shows Ryan Rising noting this was debatable, since Elliot "did participate in meetings leading up to some extent or another. Not as much as some floor leads."

Budget Allocation

$700 received. $150 spent on workshop supplies. The remainder returns to the floor's nonprofit. No governance decisions were required beyond this default.

Programming Output

One workshop, which lasted three hours. Elliot was the sole speaker and facilitator. No floor members volunteered for event shifts. No cross-floor operational support. No sponsorship leads. No programming on other floors. Attended "some" (25-75%) of the floor lead coordination meetings.

What This Tells Us

Floor 8 is, in our framing, a control case. It is a control. It shows what minimal participation looks like in the experiment's framework: a floor that received funding, chose the path of least governance resistance, produced a single session, and defaulted remaining funds to its existing organizational structure.

The comparison with Floor 14 is the starkest contrast in the dataset. Similar experiment conditions. $2,400 vs. $700 in funding. 12-14 governance participants vs. zero. 78 sessions vs. 1. New governance tooling built vs. none engaged with. The question the comparative analysis should investigate: what conditions produced this 78x difference in session output? Pre-existing governance infrastructure, community size, leadership energy, domain culture, or budget alone?

Floor 9: AI & Autonomous Systems (Devinder Sodhi)

Seed funding: $1,400 base allocation Actual spending: ~$6,000 (self-funded ~$4,600 beyond allocation) Sessions: 22 on platform Governance mechanism: Benevolent dictator Ticket attribution: 42 code-attributed registrations (32 checked in), plus ~11 invite-attributed

The Community

Floor 9 is the AI and autonomous systems hub, led by Devinder Sodhi with Clovis V and Lexi as co-leads. The floor runs an active compute lab (Floor 9 proper plus an annex on Floor 10) and serves as a working space for AI researchers, builders, and startup founders. This is not a community that formed for the experiment. In the prior year, Floor 9 ran 62 events, 48 of which were cross-floor. Devinder described the floor's role as "backbone" of the tower.

Two co-leads brought divergent visions: Clovis focused on post-labor economics and AI fund management experiments; Devinder on tech anxiety, education, and future of work. In practice, Devinder ran operations.

Governance Mechanism: Benevolent Dictator

Devinder's description of his governance process is the most direct in the dataset: "It was relatively easy. I knew that I needed to do more work on the Annex and the 10 Floor Lab anyway and I knew that I needed a lot of help with setup. So just hiring people and building out."

One person identified operational needs, made spending decisions, and executed. No committee, no vote, no formal process. The mechanism was pre-existing in the sense that Devinder already functioned as the floor's operational leader.

What Devinder spent on is where this gets interesting. He invested approximately $6,000, roughly 4x the seed allocation, covering the difference out of pocket. The spending went to paying people (including Lexi for approximately 70 hours at ~$30/hour) and buying equipment for Floors 9 and 10. His stated goal: "I just wanted it to be a space people could utilize long term. And I wanted to make sure that the people putting in the work were in some way shape or form compensated."

This is the only floor where the governance question ("how do we decide?") was inseparable from the operational question ("what needs to get done?"). Devinder did not govern a budget. He ran a production operation and funded it. The self-funding complicates the experiment's "identical constraints" framing: Floor 9's output reflects a $6K budget, not a $1,400 one.

Budget Allocation

$1,400 seed allocation, approximately $6,000 total spent. Approximate breakdown from transcript data:

Labor (Lexi, ~70 hours at ~$30/hr): ~$2,100

Equipment for Floors 9 and 10: significant but unquantified

Other operational costs: setup, production support

Programming Output

22 sessions across both days (11 Saturday, 11 Sunday). While this places Floor 9 fifth of six participating floors by session count, it understates the floor's contribution. The floor's primary role during the event was infrastructure: providing compute resources for the hackathon, staging the physical space, and deploying 25 volunteers across the building.

Cross-Floor Contribution

Floor 9's cross-floor contribution was the broadest in the experiment. Approximately 25 volunteers deployed building-wide, supporting registration, wayfinding, production, and logistics across multiple floors. Top sponsorship lead generator for the event (alongside Dhruv from Floor 4). Attended all coordination meetings and took on coordination tasks beyond his floor. DEVINDERPEOPLE ticket code drove 26 registrations (23 checked in), plus AI-FLOOR-VOLUNTEERS drove 16 (9 checked in). Post-event, Devinder received inbound requests for three additional hackathons.

The impact report described Devinder's contribution as "closer to a co-producer than a floor lead."

Preliminary cross-floor contribution score: 13-14 out of 18. Different in character from Floor 14's contribution (which centered on programming and mentorship), but arguably comparable in total value.

What Persists

The equipment purchased for Floors 9 and 10 remains. The compute lab infrastructure was upgraded. The relationship network Devinder built through sponsorship outreach continues to generate pipeline. The 25 volunteers represent a mobilizable community that did not exist as a coordinated group before the event.

Since the event, Devinder has been named Frontier Tower lead, taking on governance responsibility across the building.

Floor 6: Arts & Music (Gage Olesen)

Seed funding: $3,000 (includes subsidy for building-wide nighttime music events) Sessions: 25 on platform (Gage self-reported "minimum 12 per day") Governance mechanism: Benevolent dictator Active governance participants: 3 Transparency score: 2/5 Ticket attribution: 26 code-attributed registrations (21 checked in), 81% check-in rate

The Community

Floor 6 houses an arts and music collective. The pre-event workspace centered a counter-narrative to AI displacement: "Creativity and artistic expression will never be replaced." Artizen (Scott's organization) had planned a small fund for real-time arts allocation during the festival with participant-led allocation, though whether this ran during the event remains unclear from the data collected.

Governance Mechanism: Benevolent Dictator with Event Coordinator Logic

Gage described the governance mechanism in a single sentence: "Benevolent dictator." The reasoning was equally direct: "We booked an event, so we used an event coordinator to coordinate it." Three people participated in allocation decisions. Transparency scored 2/5, the lowest alongside Floor 11 and Floor 12.

Floor 6 received the largest allocation ($3,000), primarily because of the building-wide music event subsidy. Gage also deployed the floor's AV equipment to every stage in the building, a decision that affected every other floor's ability to run programming. Whether distributing AV building-wide was a governance decision discussed within the floor community or a unilateral call by the floor lead is the kind of question the experiment should track: when one floor's spending decisions create infrastructure for the entire venue, the governance of those decisions has consequences well beyond the floor itself.

Given more time, Gage said he would want to open the floor to programming proposals from other floor members, suggesting he recognized the limitation of centralized decision-making even within his own community.

Budget Allocation

$3,000 received. Per Gage's debrief: "All of it went to paying artists who played." The entire allocation was spent on performer compensation for the nighttime music events. No portion went to equipment (the AV was donated), permanent materials, or other categories. This is the only floor where 100% of the allocation went to a single spending category, and the only floor where that category was people rather than things.

The budget meeting transcript (March 17) and floor leads retro (March 24) confirmed that AV equipment purchased for the event was intended as communal building resources. Lights were purchased by the tower (staying); stage equipment stays with the tower.

Programming Output

25 sessions on the FtC platform across both days (10 Saturday, 15 Sunday). Gage self-reported "minimum 12 per day," which likely includes evening music programming not captured on the main schedule. Five floor members spoke or facilitated sessions.

Cross-Floor Contribution

Floor 6's cross-floor contribution was distinctive and arguably irreplaceable: they ran AV for the entire building. Every floor that had functioning sound and projection during IatF had it because Floor 6's team set it up, troubleshot it, and maintained it across two days. Gage described this as distributing "our AV system to every stage" and coordinating the deployment himself, though the equipment was donated rather than purchased with the seed allocation.

Beyond AV, Gage stepped in to fill gaps left by tower staff during event programming on multiple occasions. The post-event impact report described the arts and music programming, the AV infrastructure, and the production quality that makes Frontier Tower's events sponsorable as all running through this floor.

Gage attended 75%+ of coordination meetings and generated potential ticket sales, though no formal sponsorship leads.

Cross-floor contribution score: Cross-floor ops 3/3, resource sharing 3/3, coordination engagement 2/3. Other dimensions lower. Estimated total: 11-12 out of 18.

Tools and Documentation

Gage did not use Simocracy, Hypercerts, or any governance tools during the experiment. Documentation was informal: "We didn't document it all formally." This is consistent with the benevolent dictator pattern: when one person decides, there is less to document because there is less deliberation to record.

What Persists

The AV infrastructure deployed during the event remains available as communal building resources. Gage expressed interest in ongoing governance participation through the community treasury, with a specific condition: "clear outlines around what proposals should look like and there is some amount of tuning of the ability for the agents to think about steady incremental investment in infrastructure." This last point echoes a finding from the March 16 Simocracy session, where participants identified the AI's inability to value long-term infrastructure investment as a gap.

Floor 12: Ethereum & Decentralized Tech (Gökhan Er, Mia Deng, Yoofi Annan)

Seed funding: $1,400 base allocation (not yet received as of debrief) Sessions: 25 on platform (Yoofi self-reported 23 sessions, ~17 hours) Governance mechanism: Delegation to intern (meritocratic from delegate's perspective) Active governance participants: 1 Transparency score: 2/5 Ticket attribution: 8 code-attributed registrations (7 checked in)

The Community

Floor 12 operates as the Ethereum community hub, described as the Ethereum Foundation's first permanent US community space. The floor has a natural connection to on-chain governance: Ethereum's own governance architecture, the EF's funding mechanisms, and the broader ecosystem of DAOs and funding protocols are all part of the floor's identity.

What Actually Happened: Information Asymmetry and Transition Risk

The debrief tells a different story than the one visible from outside. Floor 12 was in transition during the experiment. Mia was out. Gokhan was heads-down running ETH SF that same week. Yoofi was onboarded as an intern very late in the process, after the project had already started, and had to piece together context without having been present for the early coordination meetings.

The governance mechanism was delegation by default: Gökhan delegated budget authority to Yoofi. From Yoofi's perspective, the approach was "meritocratic," meaning he allocated based on what he assessed the floor needed. One person made all allocation decisions. Transparency scored 2/5.

The delegation operated as expected for routine allocation decisions. In at least one higher-stakes instance, a decision was reversed by the original lead. The reversal demonstrated that the structure was conditional delegation rather than a full transfer of authority. Section 9 and Lesson 10 develop this pattern.

Yoofi was candid about the structural problem: "I came in late after the project had already started and had to dig deep to make it work. Since I wasn't looped in from the very beginning, I had to rely on secondhand context and make sense of things in real time." This is a governance finding in itself. When floor leadership turns over mid-experiment and the incoming lead lacks institutional context, governance defaults to whoever is available, regardless of whether they have the relationships, information, or authority to govern well.

The Ambitious Plans vs. Execution

The pre-event workspace documented planned governance designs: "The Funding Floor" concept with live experimentation using funding mechanisms, a persona collaboration framework (Researchers × Funders × Builders), and an "Agentic Governance Brief" authored by David Dao. In a February 5 call, Yoofi Annan described a proposed $13,000 match funding pool for a live NFT art funding experiment using quadratic funding. A February 10 call detailed a full treasury architecture with multisig signers and an AI agent managing 4% of profits.

None of this materialized during the event in the way it was designed. The workspace's budget, sponsorship, and collaborator sections were marked "coming soon" and never completed.

The irony is substantial. The Ethereum floor, housed in the building marketed as the first Ethereum community hub in the United States, and the floor most fluent in governance design concepts, produced the conditional delegation pattern documented in Section 9. Meanwhile, Floor 14 (a human flourishing community) developed what the same evaluator described as the "most advanced and documented floor treasury governance system" in the experiment.

The Floor 12 debrief adds context the external view misses. This was not a community that chose not to govern. It was a community in leadership transition that lacked the continuity to execute its own plans. The gap between Floor 12's governance design ambitions and its execution is partly a story about institutional knowledge: when the people who designed the governance are not the people executing it, the design does not survive the handoff.

Floor 12's prior EF grant also figured in FtC's own denied application for tower-wide governance infrastructure; see Section 8 Lesson 4.

Budget Allocation

$1,400 allocated. Spending: volunteers and floor lead compensation.

Programming Output

25 sessions on platform (12 Saturday, 13 Sunday). Yoofi self-reported 23 sessions across roughly 17 hours. Contributors included AL (stage manager, Day 1), Shaun (hackathon and operations), and Thomas (moderator, speaker experience, hackathon coordinator). The programming volume was respectable and comparable to Floor 6.

Cross-Floor Contribution

Floor 12 shared hackathon space with Bittensor and ran "agentic experiments." Yoofi reported programming on Floor 14 as well. The floor partnered with ETH SF Week for the FtC partnership. Meeting attendance improved to 75%+ (though this was Yoofi's attendance, not the original floor leads who joined late).

Ticket attribution remained the second-lowest: 8 code-attributed registrations (7 checked in).

Cross-floor contribution score: Partial data. Coordination engagement improved under Yoofi. Cross-floor ops limited. Estimated total: 7-8 out of 18.

Tools and Documentation

Yoofi used Simocracy and described it as "novel and insightful." No other governance tools were used. No formal governance documentation was produced.

Given more time, Yoofi said Floor 12 would want to try "consensus mechanisms," which is notable: the floor most conceptually immersed in governance design defaulted to unilateral delegation and aspires to consensus. The distance between the two is the experiment's clearest illustration of the gap between governance theory and governance practice.

What Persists

Yoofi's forward-looking comments are the most governance-literate in the dataset. He described a path "from soft consensus all the way to hard consensus" and cautioned that governance should not obstruct building: "Most of the folks in our community are builders, highly technical and focused on their startups and projects. If we over-govern, we might end up not having a community." This tension between governance ambition and community pragmatism may be Floor 12's most useful contribution to the experiment's findings.

Floor 11: Health & Longevity (Laurence Ion)

Seed funding: $1,400 base allocation Actual spending: ~$1,900 (self-funded ~$500 beyond allocation) Sessions: 28 on platform (Laurence self-reported 21 sessions, ~10 hours) Governance mechanism: Benevolent dictator Active governance participants: 4 Transparency score: 2/5 Ticket attribution: ~19 invite-attributed registrations (extrapolated). No promo codes used. SSEDA code drove 12 additional registrations.

The Community

Floor 11 houses a health and longevity community. Their workspace theme was "Live long enough to see the far future." Equipment included HBOT chambers, red light therapy, and biomarker testing, the most specialized wellness equipment of any floor. The governance documentation was nearly absent: the workspace was described as "almost entirely blank" in the data compilation.

Governance Mechanism: Benevolent Dictator with Informal Consultation

Laurence described his governance as "benevolent dictator," chosen because it was "pre-existing and we didn't have time to set up anything formal." Four people participated in allocation decisions, making Floor 11 the third-highest in governance participation after Floor 14 (12-14) and Floor 6 (3, though Gage's 3 may include himself and an event coordinator).

The process was informal and relationship-based. Laurence described it plainly: "Just flowed with the top contributors, I offered a payment amount and they agreed." No formal documentation, no vote, no committee. Transparency scored 2/5, consistent with the other benevolent dictator floors.

What distinguishes Laurence's response from the other benevolent dictators is his stated aspiration: "Progressive decentralization! Expand from 1 to 3 people and keep expanding from there." He also linked to a document on modified consensus processes for the community treasury. This is the only floor lead who articulated a governance evolution pathway rather than either defending the current mechanism or suggesting a wholesale replacement.

Budget Allocation

$1,400 received. Laurence reported spending approximately $1,900: $1,500 on people, $100 on AV, $300 on other costs. He spent the full allocation and self-funded roughly $500 beyond it, making Floor 11 the third floor where a lead covered overages out of pocket (after Floor 9's $4,600 and Floor 14's community pattern of self-funding).

The spending breakdown: approximately 79% went to people (compensating contributors), making Floor 11 the most people-heavy spender alongside Floor 6 (100% to artists). This contrasts with Floor 14, where 72% went to permanent materials and equipment.

Programming Output

28 sessions on the FtC platform (14 Saturday, 14 Sunday), the second-highest session count of any participating floor behind Floor 14. Laurence self-reported 21 sessions across roughly 10 hours, a discrepancy that likely reflects the difference between sessions scheduled on the platform and sessions Laurence's team directly produced. Three floor members spoke or facilitated: Laurence, Anya, and Johannah.

The contrast between Floor 11's sparse pre-event documentation (nearly blank workspace) and its strong programming output (second-highest session count) is one of the experiment's more puzzling data points. It suggests that workspace documentation quality is a poor predictor of programming output, and that some floor leads operate effectively through informal channels that leave no written trace.

Cross-Floor Contribution

Floor 11 was a net recipient of cross-floor support rather than a provider. Laurence reported receiving AV setup from Floor 6 or the FtC ops team (1-2 people who brought projector, speaker, and mic). No operational support was provided to other floors. No programming ran outside Floor 11's own space. Laurence generated ticket sales through personal invitations (approximately 1,000 Luma invites sent, ~19 attributed registrations, plus 25 comp codes with 24 used).

Laurence attended 75%+ of coordination meetings, putting him in the top tier for engagement.

Cross-floor contribution score: Coordination engagement 2/3, volunteer mobilization 0/3, sponsorship/revenue 1/3, programming beyond own space 0/3, cross-floor ops 0/3, resource sharing 0/3. Estimated total: 3-4 out of 18, the lowest of the six participating floors.

Tools and Documentation

Laurence used Simocracy and described the experience positively: "It was great to just talk my answers for each question, they were great questions!" Laurence had the sparsest written governance documentation of any floor, yet engaged with the one governance tool that operates through conversation rather than writing. The match between a verbal-first governance style and a tool that captures governance preferences through interviews may be worth investigating in future experiments.

No other governance tools were used. No formal documentation exists beyond the debrief form itself.

What Persists

Laurence expressed strong interest in ongoing treasury governance ("Yes!") and linked to a document outlining a modified consensus process. He described the FtC operations team as "one of the best teams I've worked with in terms of operations" and expressed enthusiasm for future collaboration.

Floor 11's specialized wellness equipment (HBOT chambers, red light therapy, biomarker testing) remains in place. The programming template (health and longevity content) proved viable with 28 sessions across two days. Whether the floor's governance evolves from benevolent dictator toward the "progressive decentralization" Laurence described will be one of the clearest tests of whether the experiment changed how floor leads think about governance, or whether the reflection was situational.

Floors That Did Not Participate

Floor 7: Makerspace (Tony)

Floor 7 dropped out of the experiment. Tony, the floor lead, stopped attending coordination meetings in the weeks before the event. That coordination gap had a downstream cost: René, a hackathon participant, lost $2,500 because the building's exclusive alcohol policy had not been communicated to the event team. He responded by building the Frontier Tower Agent during the hackathon. The irony, only visible after the fact, is that he built the autonomous infrastructure designed to prevent that exact coordination failure on the same evening he experienced it. Full backstory in Section VIII Lesson 8 and Appendix C.

Floor 4: Robotics/Physical AI (Dhruv)

Floor 4 focused exclusively on the hackathon and did not participate in the governance experiment. Dhruv provided physical space, robots, and mentorship resources for the hackathon, and was one of the top sponsorship lead generators for the event alongside Devinder.

Three-Layer Treasury Architecture

The experiment was designed as the first activation of a permanent financial infrastructure for Frontier Tower, organized in three layers. Section 4 explains that Layer 1 is the controlled experiment this report's main analysis documents; Layers 2 and 3 are follow-on infrastructure still being built at the time of publication, with observational evidence relevant to the building-level hypothesis presented in Section 9. This appendix preserves the full architectural detail for readers interested in the multi-layer design, for governance researchers tracking how the tower treasury and AI agent components evolve.

Layer 1: Floor treasuries (operational during the event). Each participating floor received a seed allocation controlled by that floor's leadership. The floor governed those funds through whatever mechanism they chose. This is the layer the experiment tested, and the subject of the report's comparative analysis.

Layer 2: Tower-wide community treasury (committed, infrastructure in progress). A Gnosis Safe multisig designed to serve as a shared treasury for the entire building. Signers will include floor treasury wallets, Frontier Tower management, Funding the Commons, and an AI agent. Funded by 2% of all event revenue and 20% of event profit, per a contractual agreement between FtC and Frontier Tower. The $8,000 committed to this treasury ($3,000 from event profits, representing 2% of approximately $150,000 in event revenue, plus the Hypercerts Foundation's $5,000 contribution) has not yet been deposited; the infrastructure is being built to integrate with Simocracy for collective governance by floor leads. On March 16, the day after the event ended, all floor leads met with David Dao on Floor 14 and used Simocracy to simulate allocations. The group decided to pool the Hypercerts Foundation's $5,000 into the community treasury rather than distributing it to individual floors, and to govern it collectively through the Simocracy system going forward. This decision marked the transition from the floor-level experiment documented in this report to the tower-wide treasury governance that follows it.

Layer 3: AI agent (built during event, deploying post-event). An autonomous agent with its own budget and signing authority on the tower treasury, operating within defined spending thresholds. The agent was built at the hackathon by René (Frontier Road, frontierroad.replit.app, see Appendix C for full backstory) and is being deployed alongside the treasury infrastructure. Its role is to operate as a co-signer alongside human-governed floors, providing a comparison between AI-assisted and human-only allocation over time. The deployment architecture is still being finalized: whether the agent acts as an additional signer in a Gnosis Safe multisig, as an advisory layer in a Simocracy Senate structure, or through some other technical implementation is a detail being resolved. The governance architecture is clear; the specific technology is pending.

Revenue-sharing contractual foundation. The three-layer design is anchored by a standing commitment: the FtC x Frontier Tower Event Partnership and Revenue Sharing Agreement establishes 2% of event revenue and 20% of event profit flowing to the community treasury for tower-wide events. This is permanent funding, not grant funding that expires. It is a business model that generates its own sustainability as long as the building hosts collaborative events. Additional funding streams include the Hypercerts Foundation's direct contribution and potential future revenue from hackathons, conferences, and programming series hosted at the tower.

The research implication. The three-layer design means that the $8,000 community treasury is not a pilot that might be refunded; it is the floor of an ongoing governance experiment that compounds as events happen at the tower. Future research iterations can compare floor-level governance (Layer 1) with tower-wide governance (Layer 2) and AI-assisted governance (Layer 3) operating simultaneously in the same building, with the same participants, over longer timeframes. This report documents the first round of Layer 1 data. The dataset grows from here.

Tooling Deployment Detail

This appendix documents the full deployment context for the on-chain and AI-assisted tooling used during the experiment. Section 7 carries the summary and the argument about why governance stayed off-chain; the detail below is source material for readers evaluating specific tools or planning similar deployments.

Simocracy and the Agentic Treasury

Figure C.1. Simocracy.org. AI digital twins for governance, deliberation, and capital allocation. The Senate runs as a standing deliberation chamber for sims; the S-Process handles capital allocation via marginal value functions. Built by David Dao with Sejal Rekhan as part of the AI4PG research initiative.

Figure C.2. Holke (Hypercerts Foundation) speaking at the event. The Hypercerts Foundation funded the $5,000 community-treasury contribution allocated through the March 16 Simocracy session.

Simocracy (simocracy.org/ftc-sf) is a preference aggregation platform built by David Dao, deployed as part of a broader research initiative called AI4PG (AI for Public Goods), a coalition including GainForest, Hypercerts Foundation, Octant, Seer, Funding the Commons, and Protocol Labs. The system creates "sims" for each participant: virtual agents that represent a floor lead's stated priorities, values, and allocation preferences. Sims then engage in structured deliberation using the S-Process, a methodology developed by Andrew Critch at the Survival and Flourishing Fund for structured multi-evaluator grant allocation.

Each floor lead was invited to a 30-minute interview with Sejal Rekhan and David Dao to create their personal sim. The process was incentivized at $50 per floor lead. The resulting sims were designed to debate and allocate the $5,000 Hypercerts Foundation treasury contribution through the S-Process.

David Dao created sims for eight floor communities (including the two that ultimately dropped from the governance experiment). The platform saw engagement, primarily from Floor 14 (Human Flourishing), whose debrief described "high engagement and enjoyed" and interest in continued collaboration.

The March 16 Session: Phase 1 and Phase 2 Play-by-Play

The most consequential Simocracy moment came on March 16, the day after the event ended. All floor leads met with David Dao on Floor 14 and used Simocracy to allocate the $5,000 Hypercerts Foundation treasury contribution. The session ran in two phases.

Phase 1: Human allocation

Participants manually reviewed 18 proposals on the Simocracy platform and individually distributed a hypothetical $5,000 budget. A recurring theme was that the proposals felt underdeveloped. Judy described them as "not solid proposals" but "very baseline ideas." Several participants said they deprioritized proposals that were already underway and looked for infrastructure and long-term community value. One participant filtered proposals into two categories: projects (fund a specific thing) and governance changes (alter how the community operates), and chose to fund only the former.

A significant discussion emerged around the limitations of individual proposals at the $5,000 scale. Participants noted that the real infrastructure needs of the building are collaborative, cross-floor, and require co-investment from multiple sources, making them poorly suited to standalone proposals. The experiment "almost feels like an experiment of unnecessary, like, fun projects," one participant observed, because the budget was too small to meaningfully advance the deeper infrastructure work. This echoes the budget-size finding documented in Section 8 (Lesson 3).

Phase 2: AI allocation via the S-Process

David Dao ran the S-Process: digital twins of all eight floor leads simultaneously evaluated all 18 proposals using marginal value functions. The S-Process asks each agent to estimate how much additional value each incremental dollar creates for a project, then optimizes the allocation across all proposals and all agents' preferences.

Please refer here (decisions): FtC SF, Agentic Funding Experiment, Frontier Tower

The results roughly matched the group's top human picks (Gifting Games and motion capture), but deprioritized bigger-seed ideas like the Flourishing Endowment and residency because the AI judged that small funding amounts would not materially move them forward.

Discussion and vote

In the group discussion afterward, participants voted. Two people preferred the human distribution. Zero voted for the AI-only allocation. Several preferred holding all $5,000 for a future round because the proposal quality did not warrant funding. This led to the session's central insight: the key problem was the quality and completeness of the proposal set, not the evaluation mechanism.

The session also identified several variables the AI did not account for: long-term treasury-building preferences, external funding already committed to projects, implementation capacity, skill availability within the building, and broader tower context. Participants recommended stronger proposal templates, shared context inputs, explicit options to save to treasury rather than spend, recurring rounds with feedback loops, co-authored proposals, and video-based accountability (short statements of intent with the proposal, video progress updates after funding).

Rather than distributing the $5,000 to individual floors in this round, the group pooled it into the tower-wide community treasury and committed to govern it through the Simocracy system going forward.

Hypercerts on the FtC Platform

James Farrell integrated Hypercerts into the Funding the Commons event platform (platform.fundingthecommons.io). The integration produced a Hyperboard: all 176 speakers at Intelligence at the Frontier were issued Hypercerts documenting their participation. The resulting Activity Cert was published on March 17, 2026, intentionally registered via the AT Protocol in coordination with the Hypercerts Foundation.

Floor 14 went further than any other floor: they built their own participatory budget tool integrated with Hypercerts, the only floor to create new governance tooling during the experiment rather than using or ignoring what was provided.

Beyond the speaker Hypercerts and Floor 14's custom tool, no other floor reported creating or receiving Hypercerts through the debrief process.

Infrastructure Not Deployed in Time

The Frontier Tower AI Agent

The original design specified an autonomous AI agent with its own budget and signing authority on the tower-wide community treasury. The agent was to operate during the event, allocating micro-grants, responding to community feedback, and producing a direct comparison dataset between AI-assisted and human-only allocation.

The agent was not operational during the event. Instead, it was built during the event. René, a Floor 7 member, won a hackathon bounty (co-funded by FtC and Frontier Tower) for the Frontier Tower Agent, now called Frontier Road (frontierroad.replit.app). René built the agent in direct response to a coordination failure he experienced during the event weekend: he spent approximately $2,500 on alcohol for an investor event, only to discover an exclusive vendor contract that the absent Floor 7 lead had failed to communicate. The agent was designed to close exactly this kind of information gap.

The agent is being deployed post-event alongside the community treasury infrastructure. Its role has evolved from the original "autonomous allocator during a two-day event" to a longer-term coordination tool for the building. This is arguably more valuable than the original design: a persistent agent embedded in building operations, rather than a two-day demonstration.

On-Chain Floor Wallets and Transaction Records

The original design specified on-chain wallets for each floor with full transaction histories published to Ethereum. In practice, floor-level spending happened off-chain. Floor leads received allocations and spent them through conventional means (purchasing equipment, paying contractors, buying supplies). No floor produced on-chain transaction records of their governance decisions during the event.

The Gnosis Safe Community Treasury

The tower-wide community treasury was envisioned as a Gnosis Safe multisig with floor leads, tower management, FtC, and the AI agent as co-signers. As of this writing, the $8,000 committed to the treasury (from 2% of event revenue plus the Hypercerts Foundation's $5,000) is committed but not yet deposited. The infrastructure is being built to integrate with Simocracy for collective governance.

Concurrent Experiments

The governance experiment did not run in isolation. Three concurrent funding and building experiments operated alongside it during Intelligence at the Frontier, each engaging the building's communities through different mechanisms.

The Hackathon

The IatF hackathon, co-produced with Protocol Labs and DevSpot (Sabeen Ali), ran concurrently with the conference across multiple floors. 148 builders participated, shipping 53 projects over the weekend. $20,000+ in prizes was distributed, including to members of the building.

The hackathon engaged floors differently than the governance experiment. Floor 4 (Robotics/Physical AI) contributed primarily through the hackathon, providing physical space, robots, and mentorship for builders, and earning $3,000 in sponsorship referral commissions.

The hackathon produced a key piece of infrastructure to emerge from the IatF weekend: the Frontier Tower Agent (Frontier Road), built by René from Floor 7. René received both a $500 cash prize from Funding the Commons and an annual Frontier Tower membership ($1,800 value) for the winning project.

The hackathon also surfaced another pattern relevant to governance: the "Frontier Tower as a Commons" design challenges (documented below) were published as hackathon bounty targets, translating governance problems into buildable specifications. Several hackathon submissions directly addressed challenges that the governance experiment itself was designed to test.

Artizen Fund

Artizen, a participatory funding platform, ran a live allocation round during the event, distributing $12,500 through its boost mechanism. The round was closed on stage during the Grand Finale. Scott from Floor 6 (Arts & Music) works for Artizen, and Devinder (Floor 9) was separately entrusted to run an additional Artizen fund. While this experiment operated independently from the floor governance comparison, it represents another live funding mechanism running simultaneously in the same building during the same weekend.

Unblock SF Sprint (March 10-13)

A pre-event sprint organized on Floor 12 (the Ethereum floor), where governance prototypes and ideas were developed with David Casey and David Dao supporting as domain experts. Outputs from this sprint were forwarded to the Protocol Labs hackathon and several informed the tools deployed during the funding experiment.

Frontier Tower as a Commons: Eight Design Challenges

Before the experiment, floor leads and building members collaboratively diagnosed eight structural challenges that prevent the building from functioning as a self-governing commons rather than a landlord-mediated co-working space. These challenges were formulated as design problems and published as hackathon bounty targets for both the IatF and ETH SF hackathons.

The challenges identified:

Missing coordination layer. The building relies on floor leads as informal human middleware for onboarding, troubleshooting, and cross-floor coordination, without shared tools or systems to support that work.

Mismatched power, responsibility, and incentives. Stewardship responsibility is distributed across floor leads, but decision rights and economic upside remain centralized with management.

No shared picture of the system. Commons participants lack access to transparent models of tower economics, cost structures, or growth strategy.

Underdeveloped governance. Concerns raised in meetings do not reliably become tracked decisions, funded working groups, or delegated authority.

Ad-hoc processes. Events, onboarding, and tech stack integration remain person-dependent rather than institutionalized.

Fragmented floors. Membership is floor-scoped rather than tower-scoped, producing siloed micro-economies.

Underuse of community capacity. Hundreds of developers, governance designers, and organizers exist in the building but are not systematically activated for infrastructure building.

Expansion outpacing sustainability. Multi-site expansion is under discussion while the operating model for a single site remains immature.

Several of the tools built during the hackathon directly addressed these challenges. The Frontier Tower Agent (Frontier Road) targets challenges 1 and 5 (coordination and event infrastructure). The Simocracy system addresses challenge 4 (participatory governance). The community treasury model addresses challenges 2 and 3 (incentive alignment and financial transparency). The Hypercerts integration addresses challenge 7 (activating community capacity through documented contributions).

The design challenges also reveal why the governance experiment matters beyond its two-day window. The floor-level funding allocation was a forcing function, but the real infrastructure gap is not "how do 6 floors allocate $1,400 each." It is "how does a 700+ person innovation community coordinate shared resources, make collective decisions, and generate value without reverting to either managerial hierarchy or fragmented isolation?" The experiment was the first test of that larger question.

Replicable Templates

This appendix provides practical materials for communities, event organizers, and building operators who want to run their own comparative governance experiments. These templates are derived from what Intelligence at the Frontier used, adapted based on the lessons documented in Section 8. They are designed to be adopted and modified, not followed rigidly.

Template 1: Governance Mechanism Menu

Present this to participating communities during the design phase (8-12 weeks before the event). Each community chooses one mechanism or designs their own. The only requirement is that they document their process.

Council / Committee Governance. A small group (5-15 people) makes allocation decisions on behalf of the broader community. Works well for communities with pre-existing leadership structures. Observed at Floor 14 (Human Flourishing): core council of 12-14 with three standing committees. Produced the highest programming output but revealed preference aggregation failures when trying to include the broader community through polls.

Benevolent Dictator. One person makes all spending decisions based on operational needs. Works well when the leader is trusted, has operational expertise, and is willing to self-fund beyond the allocation. Observed at Floor 9 (AI & Autonomous): Devinder made all decisions unilaterally. Highest cross-floor operational contribution. Governance question became inseparable from operational execution.

Default to Existing Entity. Funds flow to the community's existing organizational structure (nonprofit, LLC, cooperative) without activating a new governance process. Works as a control case or for communities with minimal session commitments. Observed at Floor 8 (Neuro & Biotech): "Default to all returning to the nonprofit." Zero governance participants. $150 spent, remainder returned to floor.

Participatory Budgeting. Community members propose and vote on spending priorities. Requires tooling (Simocracy, Snapshot, Coordinape, or similar) and a facilitator to manage the process. Partially attempted at Floor 14 via WhatsApp polls (failed: 5 voters out of 60+). Successfully run building-wide through Simocracy on March 16 for the community treasury allocation.

Quadratic Funding / Quadratic Voting. Participants allocate a budget across proposals, with matching funds amplifying broad-based support. Requires a dedicated platform (Gitcoin, clr.fund, or custom). Planned but not implemented at Floor 12 (Ethereum). Requires technical setup and a governance facilitator. Best suited for communities with crypto literacy or a dedicated onboarding process.

Delegation. Community members delegate their allocation authority to trusted representatives, who then make spending decisions. Can be combined with council governance. Not explicitly observed in this experiment, but a common pattern in DAO governance that could be tested in future iterations.

AI-Assisted Allocation. An autonomous agent with defined spending thresholds makes or proposes allocation decisions based on community input and data. Designed but not live during the event. The Frontier Tower Agent (Frontier Road) is being deployed post-event as a co-signer on the community treasury. Future experiments should test AI-assisted allocation alongside human-only mechanisms.

Custom / Hybrid. Communities can design their own mechanism or combine elements from the above. Document the design rationale. Floor 14 effectively ran a hybrid: council governance for spending decisions, Simocracy for preference discovery, and a custom Hypercerts-based participatory budget tool they built during the experiment.

Template 2: Floor Treasury Setup Checklist

Use this to prepare each participating community before the experiment begins.

8-12 weeks before event:

[ ] Confirm anchor governance technology partner (platform, facilitator, funding)

[ ] Brief all community leads on the experiment: choose a mechanism, document everything

[ ] Distribute governance mechanism menu (Template 1) to all participating communities

[ ] Establish seed budget amounts per community (recommend minimum $3,000-$5,000)

[ ] Confirm budget source(s) and disbursement timeline

[ ] Set up weekly coordination meetings with all community leads

[ ] If using on-chain infrastructure: set up wallets, onboard leads to tooling, assign a dedicated facilitator

4-6 weeks before event:

[ ] Each community confirms their chosen governance mechanism

[ ] Each community documents their mechanism in a shared workspace

[ ] Begin governance design: who participates, how decisions are made, what's the quorum

[ ] Governance facilitator begins working with non-technical communities on tooling setup

[ ] Distribute comp ticket allocations and affiliate/referral codes

1-2 weeks before event:

[ ] Seed budgets distributed to community leads

[ ] Final coordination meeting: confirm session counts, programming, production needs

[ ] Documentation requirements confirmed: each community knows what they need to report

[ ] Cross-community convergence session scheduled during event (share governance learnings)

During event:

[ ] Communities execute their governance mechanisms with real funds

[ ] Spending decisions documented in real time where possible

[ ] Cross-community session runs (share what's working, what's failing)

Within 1 week after event:

[ ] Debrief forms sent to all community leads (Template 3)

[ ] On-chain transaction records compiled (if applicable)

[ ] Platform session data exported

[ ] Ticket attribution data compiled

Within 30 days after event:

[ ] Comparative governance report published

[ ] All community leads debriefed

[ ] Ongoing treasury infrastructure activated (if applicable)

Template 3: Community Lead Debrief Questionnaire

25 questions across 6 sections. Estimated completion time: 15-20 minutes. We recommend offering a $200 honorarium per respondent and following up in person with the most data-rich communities for additional context.

Section 1: Governance Process

What decision-making mechanism did your community use to allocate funds? (e.g., consensus, vote, delegation to lead, council, other)

Why did you choose this mechanism? (Pre-existing? Designed for the experiment? Path of least resistance?)

Did your governance process evolve during the event, or was it set before the event started?

How many people actively participated in allocation decisions? (Not attendees. People who had a voice in how money was spent.)

Were there disagreements about how to spend the funds? If so, how were they resolved?

How transparent was your budgeting process to your community? (Scale 1-5: 1 = lead decided alone, 5 = fully open process visible to all members)

If you had more time to design your governance process, what mechanisms or experiments would you want to try?

Section 2: Budget & Spending

How much seed funding did your community receive? ($)

Approximate breakdown of how funds were spent: speakers/facilitators/honorariums ($), equipment/permanent materials ($), food & beverage ($), decorations/atmosphere/installations ($), micro-grants to participants ($), other ($).

Did you spend the full allocation? If not, what's remaining and what happens to it?

Section 3: Engagement & Programming

How many sessions did your community host? (Count and approximate total hours)

How many community members spoke or facilitated sessions?

How many community members volunteered for event operations?

How many complimentary tickets did your community distribute?

Have you seen new members join as a direct result of the event?

Section 4: Cross-Community Contribution

Did your community provide operational support to other communities during the event? (AV setup, volunteer deployment, mentoring, equipment loans, logistics) Describe.

Did your community run programming in any space other than your own? Which spaces, what sessions?

Did your community generate any sponsorship leads or ticket sales for the overall event?

How many of your community members volunteered for operations beyond your own space?

Did your community share physical resources (equipment, furniture, space) with other communities?

How many coordination meetings did you or your representative attend in the weeks before the event? (Most / Some / Few / None)

Section 5: Tools & Documentation

Did your community interact with any governance tools deployed during the experiment? (Describe the experience.)

Do you have any documentation from your governance process you're willing to share? (Meeting notes, chat threads, vote records, budget spreadsheets, photos)

Section 6: Looking Forward

Would your community participate in ongoing governance of a shared treasury? What would make that work?

Would you be interested in working with sponsors or collaborators on governance design or allocation experiments in the future?

Anything else you want documented about your community's experience? (Open text)

Template 4: Cross-Community Value Contribution Scoring

Rate each participating community on a 0-3 scale across six dimensions. Score using a combination of self-reported data (from Template 3) and organizer observation.

Maximum score: 18. Use the total to rank communities by cross-community contribution, then test for correlations with governance mechanism type, budget received, and pre-event engagement.

Template 5: Experiment Design Checklist

Minimum requirements for running a comparative governance experiment at a multi-community venue.

Non-negotiable:

[ ] Minimum 4 participating communities (6+ preferred for analytical value)

[ ] Minimum $3,000-$5,000 seed budget per community

[ ] One anchor governance technology partner confirmed before scoping the experiment

[ ] Dedicated governance facilitator on-site during the event

[ ] 8-12 weeks of lead time for governance design phase

[ ] Standardized debrief process (Template 3) with incentivized completion

[ ] Transparency requirement: communities must document how they decide, not what they spend

Strongly recommended:

[ ] Contractual revenue-sharing mechanism that funds an ongoing shared treasury (e.g., 2% of event revenue)

[ ] Cross-community convergence session during the event

[ ] Pre-event coordination meetings (weekly, starting 8+ weeks out)

[ ] Condition full budget on coordination meeting attendance

[ ] Track self-funding explicitly (confounding variable)

[ ] Export session data, ticket attribution, and attendance records for quantitative analysis

Nice to have:

[ ] On-chain infrastructure (wallets, transaction records, multisig treasury)

[ ] AI agent with autonomous allocation authority for human-AI comparison

[ ] Concurrent hackathon targeting governance tooling as build targets

[ ] Post-event follow-on funding to communities demonstrating strong governance