Records include organizations, products, programs, proposals and intellectual context. These are not counts of independent operating companies.

KenMatch: competitor and collaborator intelligence

Research cutoff: September 23, 2026
Assessment type: source-grounded strategic analysis; public evidence, code and prototype audit
Reading order: executive synthesis → canonical architecture → strongest analogues → strategy → source and ecosystem audits. The companion workbook contains the full comparison fields, entity census and every original ecosystem-map row.

1. Executive synthesis

KenMatch is best understood as a proposed public institution for choosing, funding, executing and evaluating sustained AI work. Its current product is a public coordination prototype with database credits and sandbox execution. Treating it simply as a freelance marketplace, prediction market, DAO, AI-agent marketplace or grant site hides the most important comparisons. Conversely, treating its complete vision as already operational would materially overstate its position. The repository explicitly separates operational account/community functions, sandbox allocation/funding/run records, and unconfigured provider execution. KenMatch product-state declarations, live FAQ.

The defining proposed chain is:

Community identifies a problem → participants refine a bounded proposal → legitimate, nonpurchasable influence prioritizes it → pooled resources fund constrained AI execution → reviewers assess artifacts and outcomes → contribution and demonstrated value inform future rights or benefits.

That interpretation survives comparison with the supplied Markdown and current product, with an important qualification: the final execution, outcome-validation and reward-feedback loop is an ambition, not demonstrated production operation. The document also contains mutually inconsistent drafts, including blockchain, decentralized hardware, financial-rights and autonomous-routing claims. Those do not override the current implementation.

The most consequential findings

1. The strongest existing workflow analogues are Deep Funding/Deep Projects and Project Catalyst. They combine proposals, community review/selection, funding, team execution and milestone evidence. They should receive more attention than generic agent marketplaces. They do not establish KenMatch’s specific combination of contribution-based nonpurchasable voice and pooled long-horizon frontier-agent allocation. Deep Projects rules, Catalyst process.
2. Colony, Sensorica and DisCO are especially important for contribution, authority and benefit distribution. These comparisons expose a weakness in the KenMatch narrative: “rewarding value” is not a new mechanism by itself, and the hard part is defining who may validate which contributions and how that affects power. Colony reputation, Sensorica contribution accounting example, DisCO governance model.
3. AI Horde is the clearest practical earned-compute-priority comparison. ResearchHub adds a substantial open-research, contribution-reward and funding loop. BIO/VitaDAO add community scientific funding and economic rights. Each reproduces important portions of the architecture; none should be dismissed as merely thematic. AI Horde mission, ResearchHub description, BIO, VitaDAO research.
4. The closest broad proposals include 100X, FoundUPS and One People. 100X connects community-selected ideas, compute pools, agent builds, review and revenue distribution, but explicitly gives capital/staking a governance role. One People describes contribution recognition and community-funded commons jobs, while explicitly saying its proposed nonprofit/cooperative institutions are not operating. FoundUPS combines agent work, compute allocation and proposed benefit economics, with code and simulation that do not prove full production integration. These are serious comparisons at the design level, not verified identical businesses. 100X governance, One People proposal, FoundUPS repository.
5. No completely or essentially identical operating implementation was established by the evidence inspected. This is a bounded negative finding, not proof of worldwide uniqueness, patent novelty, or an uncontested market. Several projects substantially overlap with KenMatch’s vision, and a composed stack of existing tools could reproduce much of it.
6. The strongest collaboration opportunities are functional: Public AI Inference Utility for public-interest inference; CIP/Weval for community-informed evaluation; Metagov and RadicalxChange for governance design; OpenMined for governed private-data computation; Hypercerts for portable work/evidence records; and selected scientific or open-source communities for independent acceptance testing. None is an announced KenMatch partner. Public AI identity, CIP, Metagov, RadicalxChange, OpenMined, Hypercerts.

Strategic judgment: KenMatch’s opportunity is an unusually coherent combination of public decision rights, sponsor-independent ranking, accountable execution and a reusable public evidence trail. Its main competitive risk is that the combination can be assembled elsewhere before KenMatch demonstrates reliable delivery. Its plausible defensibility is accumulated institutional trust and verified outcomes, not the existence of quadratic voting, credits, a public board or an AI-agent interface.

2. Continuity, evidence and confidence

What was actually recovered

The original private task was retrieved through the task-reading interface. It exposed the original request and attachment references, but no substantive assistant analysis. The shared snapshot was independently opened in a browser and showed the original request followed by a usage-limit message. No unseen prior findings have been invented. The continuity hypothesis supplied in the handoff was preserved and tested.

The large Markdown itself contains earlier candidate lists and analytical drafts. Those were retained as leads: BOINC, Folding@home, World Community Grid, AI Horde, Akash, io.net, Aethir, Gensyn, Bittensor, Gitcoin, research-access programs and agent ecosystems. They were not treated as verified conclusions merely because an earlier generated passage called its research exhaustive.

The interrupted local checkpoint was also preserved. Its JSON evidence archives parsed successfully; the original source hashes remained unchanged; the KenMatch checkout remained clean at the same commit; and map_specs.tsv contained all 91 expected entity occurrences, with no missing, extra or malformed rows.

Evidence labels used throughout

“High confidence” primarily describes a narrow, directly supported finding. It does not certify an entire business. Official pages are strong evidence of published rules and weaker evidence of success. Undated pages are point-in-time observations; their wording is not assumed to have existed throughout earlier periods. Search crawl labels, relative dates and footer years were not used alone to establish launch dates. No claim about post-cutoff events is intentionally included.

3. Canonical KenMatch architecture

3.1 Current product and proposed institution

The supplied source is an accumulated working dossier, not one clean constitutional specification. The most defensible canonical model gives priority to current implemented rules for current-state claims, and uses the consistent parts of the Markdown for the broader ambition.

Primary current-state references: architecture, allocation policy, attestation, funding and run governance, moderation threat model. Source-dossier anchors: KenMatch.md lines 105–249, 335–357, 535–631 and 2783–2960.

3.2 Allocation is more specific than “democratic compute”

The ranking code charges the square of the number of votes and computes incremental cost from the difference between successive squares. It caps votes per task at six. Eligible tasks need positive votes and must pass stage/safety conditions. Within a category, votes determine order; ties resolve deterministically using creation time, title and ID. The rank thresholds are nested: 1–3 Months; 4–10 Weeks; 11–100 Days, giving up to 3, 7 and 90 distinct tasks respectively. Reading the prose as 3 + 10 + 100 independent slots would be wrong. Allocation implementation.

The source also grants a baseline and periodic replenishment. Consequently, “all power is earned solely by proven value” is too strong a description even of the intended current policy. There is a citizenship/access component as well as contribution rewards. Attestation can reduce or pause participation. Nonpurchasability blocks one direct channel of wealth influence; it does not by itself prevent purchased labor, multiple identities, organized voting, reviewer capture or differences in time available to contribute. These are analytical risks, not findings of an actual attack.

3.3 Funding, capacity and execution must remain separate

Public rank can remain stable while available capacity changes. The repository distinguishes committed unrestricted compute support from projected, simulated, restricted or safety-reserved resources. Its capacity states affect launch/concurrency decisions, while reviewers retain reasons to pause, redirect, stop or release work. This separation is a substantive design choice: a vote does not create money, and a funded task does not automatically pass a safety or quality gate. Capacity policy.

The inspected live economics presentation contained a sandbox usable balance of $30,200 and public burn of $97,800/month, producing about 0.3 months of illustrative coverage. The site explicitly labels its financial and sponsor values as simulated. “Live” or “pilot” badges nested inside demo revenue streams must not be interpreted as verified commercial operations. The website’s liveness endpoint returning healthy is also distinct from a modeled critical compute-capacity state. Live economics, health endpoint.

Fifteen Ken detail pages were retrieved, including blocked safety examples. A running rare-disease example includes apparent milestones, corrections and artifacts, but it also expressly discloses simulated execution and outcomes. A blocked clinical-autopilot example illustrates an intended safety boundary; it does not establish a functioning medical review institution. Rare-Disease Mechanism Atlas, Clinical Decision Autopilot.

3.4 Verification contains several different problems

KenMatch needs to distinguish at least six questions: Is this a unique/eligible participant? Did computation occur? Does the artifact satisfy its specification? Is its factual or scientific content valid? Did the project create useful outcomes? Which participant caused how much of that value?

A blockchain can record a claim. A hash can identify bytes. A trusted execution environment or computation proof may support a bounded execution claim. A test suite can establish specified behavior on tested cases. None alone answers all six questions. This is why compute-verification networks, identity systems, peer-review markets and impact-record systems can be complementary without being interchangeable competitors.

4. Contradictions in the KenMatch sources

The document contains several checkable numerical problems. Its 0.34 Wh versus 0.3 Wh comparison is approximately 1.13×, not 10×. At $0.20/kWh, 700–800 W over 720 hours costs $100.80–$115.20 for that load, not $160. A PUE of 1.55 implies overhead equal to 55% of IT energy, or about 35.5% of total facility energy, not 55% of total. These are corrections to arithmetic in the supplied dossier, not replacement estimates of KenMatch’s real costs. Dossier anchors: lines 699–705.

Other economic examples assume arbitrary token throughput, uninterrupted utilization, particular models, enormous clusters or fixed daily loops. Even where multiplication is internally consistent, it does not establish that a given public problem requires that workload or that the resulting output is valuable. Current provider prices were not needed to establish the competitive classification, and stale illustrative prices are not presented here as September 2026 quotations.

Brand-name discussions, domain preferences, trademark conclusions, founder-income aspirations and statements about other companies’ motives are archival context. They are not evidence of competitor identity, existing legal rights, demand, profitability or operational capability. No trademark clearance, corporate-formation opinion or claim about alleged misconduct is adopted from the dossier.

5. Taxonomy: what competes with what

A single organization may hold multiple roles. Deep Projects can compete for public-interest AI proposals and also co-fund them. An agent vendor can supply execution and compete for users who simply buy the result directly. An infrastructure organization can become a platform competitor if it adds allocation and community governance. Classification is therefore conditional on the user, workflow and resource decision at issue.

For present-day buying decisions, there is no evidence that KenMatch already substitutes for a reliable production execution service. For its proposed institution, the most important competitors are multi-mechanism funding/production systems and possible assembled stacks. These are different comparisons and should remain visible.

6. Strongest structural analogues

6.1 Deep Funding / Deep Projects: the strongest AI-project funding workflow

The current deepfunding.ai address redirects to DEEP Projects. Its rules describe community feedback and selection, grants, recipient responsibility for delivery, and round-specific conditions. The ecosystem includes historical completed rounds and milestone reporting. A reputation/voting-weight project is documented, but an awarded reputation-development contract is not proof that every proposed weighting mechanism governs current decisions. Program rules, Reputo project account.

Assessment: this is a near-direct analogue for the proposal-to-funded-AI-project workflow. The main architectural difference is the allocated object and executor: grants to teams developing AI ecosystem projects versus KenMatch’s proposed shared frontier-agent effort directed at public problems. KenMatch should benchmark proposal quality, reviewer incentives, milestone failure handling and grantee delivery here. A future claim that KenMatch uniquely connects democratic AI funding to execution would need to survive this comparison.

6.2 Project Catalyst: community funding with explicit delivery accountability

Catalyst combines community refinement, voting and milestone-based project execution, with incentivized reviewer roles. Fund-specific parameters matter: the inspected Fund15 documentation includes stake-based voting calculations and March 2026 milestones. Older Fund10 reward numbers should not be carried forward as current constants. Catalyst overview, Fund15 parameters.

Assessment: Catalyst is a particularly strong counterexample to any claim that a public proposal board with voting, grants, milestones and rewards is novel. KenMatch’s potentially differentiating commitment is separating money from public allocation voice and using the selected budget for accountable AI work. Catalyst is less close on that constitutional choice, but substantially more relevant on delivery governance than most AI-branded comparison lists acknowledge.

6.3 Colony: contribution-earned authority is already a developed design space

Colony’s reputation is nontransferable, domain-aware and earned through particular organizational work/payment paths. Reputation does not eliminate tokens or permission structures; governance actions can require both reputation and token stake. It is therefore inaccurate to describe Colony either as simple one-token-one-vote governance or as completely wealth-independent. Reputation documentation.

Assessment: the important similarity is the feedback loop from contribution to authority over future resources. KenMatch needs equally precise answers about decay, newcomer access, domain expertise, reversals and disputes. Its current event-based awards are easier to understand than a general value score, but they do not yet establish that additional influence tracks additional social value. Colony is both a mechanism-level competitor and a useful design reference, without being a public frontier-compute commons.

6.4 AI Horde: earned priority is different from collectively chosen priorities

AI Horde connects contributed computing work to priority in a distributed generation service. Its mission describes kudos as noncommercial but giftable. That combination is not identical to KenMatch’s nontransferable account voice. AI Horde mission.

Assessment: AI Horde most directly challenges the claim that people can access AI resources only by paying cash to a major provider. However, earning faster service for one’s own request is different from earning influence over which common projects receive resources. KenMatch should keep queue priority, voting rights, reviewer reputation and financial compensation as distinct instruments. Collapsing them into one token would conceal rather than solve their different incentive problems.

6.5 ResearchHub: science, review, rewards and funding in one system

ResearchHub documents open research discussion, paid review and RSC-based contribution/funding functions. Its funding guide explicitly says proposal review provides context and suggestions rather than determining whether a proposal receives funding. Funding recipients guide, RSC uses.

Assessment: this is a strong near-direct analogue for KenMatch’s research coordination ambitions, especially where useful human evaluation earns rewards that can support further research. It differs on tradable economic utility, donor/funder choice and the absence of a demonstrated general public frontier-run allocation constitution. It may be more valuable as an evaluator/replication counterpart than as an adversary. KenMatch should not assume that paying for reviews establishes their independence or accuracy; that requires separate evidence.

6.6 BIO, VitaDAO and Molecule: scientific execution with economic rights

These are related but distinct. BIO provides an ecosystem/protocol and scientific-agent work; VitaDAO is a specific research community and portfolio; Molecule supplies scientific IP financing infrastructure. Their relationship is not a reason to count them as three unrelated copies of the same business. BIO AI, VitaDAO portfolio, Molecule documentation.

Assessment: this family makes the ownership question unavoidable. If KenMatch’s goal is open public outputs, IP-restricted commercialization may conflict with it. If it intends to distribute economic upside to contributors, DeSci’s rights and governance arrangements become more relevant—but also substantially more complex than issuing database credits. BIO’s agent work increases technical overlap without removing the difference between capital-linked rights and nonpurchasable public voice.

6.7 Sensorica and DisCO: close institutional analogues outside AI branding

Sensorica’s concrete collaboration documentation connects planned roles, logged contributions and resource sharing. DisCO explicitly recognizes livelihood, commons and care work in governance/economic models. Sensorica project accounting, DisCO principles and value flows.

Assessment: both expose categories of contribution that an AI-task platform can overlook: maintaining community health, onboarding, documenting, correcting, coordinating and supporting shared infrastructure. Rewarding only proposal wins or visible artifacts can underpay essential work. Their main difference is operational substrate—human cooperative production and open networks rather than allocated frontier-agent runs. They are strong organizational analogues and plausible methods collaborators, with no assumed readiness or willingness to partner.

6.8 Cobudget: much of the public workflow can be built without a new economy

Cobudget supports proposing, discussing, funding and reporting on projects under a group’s chosen money-governance model. Its practice guide includes proposal quality and accountability, not merely a payment button. Getting started, funding-round practice.

Assessment: this is a useful “minimum sufficient competitor” test. A community could use collaborative budgeting, an agent service and a public artifact repository to address much of the same need. KenMatch must demonstrate that integrated credits, ranking, execution records and review materially improve outcomes enough to justify adopting a new institution. Integration can be valuable without being a defensible invention.

6.9 Gitcoin, Gardens, Optimism and Octant: distinguish mechanisms from live programs

These systems should not be treated as one interchangeable public-goods stack. Gitcoin’s quadratic matching, Gardens’ persistent conviction, Optimism’s retrospective rewards and Octant’s yield/participation model create different rights and incentives. Gitcoin announced the sunset of Grants Stack and Grants Lab in May 2025; its June 2026 AI-transition direction is a strategic discussion, not proof of an operating KenMatch equivalent. Optimism’s August 2026 budget update reports Retro Funding’s pause. Octant documents v1’s final allocation window and migration to v2. Gitcoin sunset, Gitcoin June 2026 direction, Optimism budget, Octant migration notice.

Assessment: public-goods mechanisms supply precedents for preference aggregation, fund distribution and retrospective reward. They do not supply a settled formula for social value. KenMatch should borrow lessons selectively, preserve versioned rules and avoid presenting a paused or sunset program as a ready integration or guaranteed funder.

6.10 Bittensor and Allora: evaluated AI work, but a different allocation constitution

Bittensor’s current Dynamic TAO model and Allora’s worker/reputer incentives are more specific than “decentralized AI.” They connect evaluated output and economic rewards, with stake and market mechanisms playing material roles. Dynamic TAO, Allora consensus and rewards.

Assessment: they reproduce AI production, evaluation and incentives more directly than civic deliberation products do. But a useful inference on a defined task is not identical to selecting a socially valuable project, and a market price is not interchangeable with democratic legitimacy. They are serious mechanism competitors and possible suppliers. Any integration should name the particular service, evaluator and failure mode rather than rely on the network’s broad claims.

6.11 Morpheus and Virtuals: broad agent economies without the same public mandate

Morpheus explicitly organizes incentives around compute, code, capital and community. Virtuals describes agent identity, commercial jobs, capital and coordination through its agent economy. Morpheus repository, Virtuals whitepaper.

Assessment: these are important architectural comparators because they go beyond a simple model marketplace. Nevertheless, capital participation, agent commerce and network incentives do not establish a contribution-only public electorate deciding which common problems deserve frontier resources. KenMatch may purchase useful services from such networks while retaining its own selection and acceptance rules. Their broader roadmap language should not be used as proof that every advertised institution already functions.

6.12 Public AI Inference Utility and public research allocations

The Public AI Inference Utility describes a nonprofit, open-source project fiscally sponsored by Metagov, with a service launch in 2025 and subsequent 2026 model/infrastructure updates. It is distinct from its sister Public AI Network. NSF’s March 2026 NAIRR update establishes a continuing public research-resource program. Public AI about, Public AI updates, NSF NAIRR update.

Assessment: these are especially important substitutes for the access problem and collaborators for the execution layer. Public AI may align more closely with KenMatch’s mission than a tokenized compute marketplace does. NAIRR/ACCESS can be stronger solutions for eligible researchers, while not serving as a universal public electorate. KenMatch should establish which users remain underserved after these alternatives, rather than assert that public or subsidized AI access does not exist.

6.13 100X: unusually close pipeline, materially different power

100X documents cause-specific idea selection, operator-led agent sprints, review/testing and revenue distribution funded through a reserve/yield model. Its governance explicitly privileges CPT holdings and staking, and it describes its rules as evolving in alpha. Displayed agent, deposit and uptime numbers were not independently verified. 100X build process, governance, revenue model.

Assessment: this is one of the closest discovered causal-architecture proposals. It is also a useful boundary case: replacing capital-weighted direction with nonpurchasable voice changes the institution fundamentally, even if the screens and build pipeline look similar. Neither platform’s planned economics should be accepted as a demonstrated perpetual funding engine. This candidate deserves continued monitoring, without inferring deployment from an attractive metrics display.

6.14 FoundUPS, One People, DeCLAI, DAIT and Intellect Project

These belong in the assessment because excluding prototypes would create a misleading impression of conceptual white space. They also require unusually careful evidence labeling.

Sources: FoundUPS, One People, DeCLAI status and limitations, DAIT, Intellect Project.

Assessment: there is meaningful conceptual competition. The evidence does not support “nobody else has thought of this.” It also does not support “these are all functioning businesses identical to KenMatch.” In particular, a stated plan for mathematical verification should not be accepted as an operational impact-verification system. DeCLAI’s explicit withdrawal of unsupported projections is a useful example of the evidentiary discipline needed across this category.

6.15 Research agents and challenge networks: lower architectural overlap, high substitutability

Elicit, Edison Scientific, Sakana’s AI Scientist and Google’s AI co-scientist address parts of the scientific output workflow with different maturity and access conditions. Coding/research agents can also substitute for some proposed Kens. HeroX, Wazoku/InnoCentive, Topcoder and XPRIZE instead organize sponsor-defined problem solving and evaluation. Elicit lineage, FutureHouse’s 2026 program and Edison access, Sakana’s bounded publication experiment, Google research announcement, HeroX process, Wazoku organization.

Assessment: a user may prefer an immediate output or a reliable challenge provider over joining collective governance. Therefore, low architectural similarity does not mean low competitive pressure. KenMatch’s best initial users may be groups whose problem is genuinely collective priority-setting and accountable public delivery, not individuals who merely need a report or software patch.

7. The explicit near-identity challenge

An essentially identical implementation would need unusually high correspondence on all material constitutional and operational elements, not just a shared mission:

1. A public or community proposal/refinement process.
2. Binding influence substantially independent of purchasing power, with an explicit contribution/participation policy.
3. Collective allocation of scarce sustained AI effort across competing public projects.
4. A real funding/provisioning path and demonstrated execution.
5. Checkpoint, correction, safety and release decisions tied to inspectable artifacts.
6. A contribution/outcome feedback loop that governs future rights or benefits.
7. Compatible public-benefit, ownership and control rules.

This test deliberately distinguishes operating identity, design similarity and user substitutability. A missing detail remains unknown; it is not automatically scored as absent. Equally, unknown details cannot establish near-identity.

Answer: no inspected source established a currently operating business or platform completely or essentially identical to KenMatch’s proposed full architecture. Several are close enough that a categorical uniqueness claim would be unjustified. KenMatch itself has not demonstrated its full architecture either, so comparing its complete vision against competitors’ shipped features alone would be an unfair asymmetry.

The assembled-stack counterexample

A credible alternative could combine a participatory proposal tool, a community treasury/grant mechanism, a commercial or public AI backend, milestone review, a public artifact repository and contribution-based rewards. This is an analytical construction, not a claim that these vendors already offer a tested integration or compatible legal arrangement.

Its importance is strategic: KenMatch competes partly with organizational assembly. The product must lower coordination cost, preserve neutral rules and improve evidence quality enough that people prefer the integrated system. An integrated public record can differentiate it; integration alone does not establish a durable barrier to replication.

8. Collaborators with concrete complementary value

The following priorities are analytical recommendations. They are not partnerships, expressions of interest, eligibility determinations or instructions to contact anyone. No outreach occurred.

Evidence for these functional capacities appears in the comparison profiles and workbook. Additional direct references: OpenMined, Hypercerts’ explicit limits, Loomio, Akash deployment lifecycle, Prime Intellect compute platform.

The best partnership is not necessarily the organization with the most similar mission. A collaborator should supply a missing capability, have compatible rights/incentives, and improve a measurable outcome. A think tank may provide valuable criticism while being unable to supply engineering, reviewers or funding. A provider may supply compute while creating a governance conflict. These distinctions should be explicit in any future agreement.

9. Historical and intellectual predecessors

Sources include Ostrom Workshop, QV research, BOINC Central, Folding@home, Climate CoLab read-only notice, SourceCred, Coordinape, Convergent Research.

The intellectual contribution KenMatch might make is a particular institutional synthesis and its empirical demonstration. None of these precedents alone defeats that possibility; together they prevent treating every component as an invention.

10. White space, differentiation and defensibility

10.1 What could be meaningfully differentiated

The strongest proposed combination is a public work record that joins legitimacy before execution to accountability after execution, while keeping sponsorship from buying public rank. Its distinguishing elements could include:

• A proposal that names a beneficiary, deliverables, evidence requirements, limits and stop conditions before resources are spent.
• A transparent allocation rule with participation rights that cannot simply be purchased.
• Separate decisions for rank, funding eligibility, launch capacity, safety and output acceptance.
• Provider-independent evidence and correction history, including failed, redirected and partial results.
• Contribution recognition that improves future decisions without becoming tradable control over the commons.
• A public explanation of how money, labor, compute, decision rights and benefits move through the system.

These are candidate differentiators, not verified advantages. The current FAQ itself frames defensibility around trusted governance, public evidence, reputation, reproducible artifacts and provider independence rather than an uncopiable ranking trick. KenMatch FAQ implementation.

10.2 Where the apparent white space may be misleading

First, a lack of a famous direct competitor may reflect difficult demand, evaluation and funding economics rather than an untouched opportunity. Second, “continuous compute” is not a unit of value. Some tasks benefit from a short carefully scoped run; others need data, experiments or expert labor that more tokens cannot replace. Third, collective voting can add transaction cost when users already know what they want. Fourth, projects closest to the full vision may be early precisely because integrating these mechanisms is hard.

The market should therefore be defined around underserved collective work, not all expensive AI use. A promising initial category would have motivated beneficiaries, reusable public outputs, attainable independent validation, manageable risk and enough uncertainty about priorities that collective judgment helps. This is a selection hypothesis, not a demand finding.

10.3 Defensibility risks supported by the evidence

Network effects are conditional. More users can add proposals and evaluation knowledge, but can also add noise and moderation load. More compute providers can improve choice, but incompatible models and weak reliability reduce that benefit. A long contribution history can build trust, but can entrench incumbents. The report does not assign a generic “strong network effect” label without these conditions.

11. Strategic dependencies and evidence needed next

The immediate strategic bottleneck is proving a complete loop, not adding more abstract mechanism names. The highest-value validation would be a small set of real Kens with independent acceptance criteria, actual spend, named execution systems, public artifacts, documented corrections and final reviewer decisions. The same records should show how participation affected selection and how any credits were awarded.

No new user input is required to understand the present assessment. These are the missing facts that only operational primary records or future experiments can resolve. Public research cannot manufacture them.

12. Complete audit of the supplied ecosystem workbook

12.1 What the workbook actually contains

The native ZIP/XML was inspected, including cells, styles, sheet metadata, relationships and ancillary parts. There is one visible sheet, Sheet1, with serialized cells spanning A1:A102. Ninety-four cells contain values: an invitation at A2, two headings, and 91 entity occurrences. The first group, A4:A76, is unlabeled; “Writing” appears at A78 and “Academic” at A84. The file has one explicit external hyperlink, http://pol.is, at A50.

There are no formulas, cell comments, hidden sheets, merged-cell relationships or substantive graph/drawing content in the inspected package. An empty drawing part and a persons metadata part do not create additional organizations or hidden annotations. Blank serialized cells do not add entries. Bold headings provide grouping; the workbook does not encode directed relationships, economic flows or a competitive taxonomy.

This is a grouped seed list, not a validated relationship map or complete competitor universe. The invitation to contact CIP is source content, not authorization to send a message. No contact was made.

12.2 Identity normalization

• Duplicate occurrences: Reboot appears at A58 and A81; the Ostrom Workshop appears at A21 and A92. Removing those repeated identities yields 89 provisional labels, not 89 independently verified organizations.
• Compound entries: A52 combines Ought/Elicit; A88 combines MIT Connection Science/Constructive Communication. Preserve each source row and normalize its constituent entities separately where useful.
• Parent/project relationships: Noema belongs with Berggruen; Ford Public Interest Technology is a program; Climate CoLab is a project associated with MIT CCI; a research center is not its entire university. These relationships should not inflate independent-business counts.
• Concepts rather than companies: Exit to Community, platform cooperativism and “Metascience Community” require type labels rather than invented incorporation identities.
• Geography/institution: “Ostbelgien” is best treated as a reference to a region’s citizens-dialogue institution, with the original wording retained.
• Probable names, not certainties: Moral Imaginations, The Future Society, Linden Lab, Reboot, Newspeak House, Common Wealth and CU Boulder MEDLab have plausible matches, but unsupported certainty is avoided when the workbook provides no URL.
• Unresolved: PARPA, Demos’ jurisdiction, the intended Foresight entity and Verses remain qualified. Narwhal can be connected to a CIP founder’s 2023 discussion-platform reference, but its current product/location/status remains unresolved.

The companion Map Audit sheet preserves all 91 rows, original text and cells, the original grouping, proposed canonical label, role, identity qualification, evidence URL and retrieval limitation. The original workbook has not been changed.

12.3 Important staleness and classification corrections

12.4 Important omissions from the seed map

The supplied map underrepresents operational funding/production mechanisms, distributed compute access, scientific financing and agent execution. High-value additions include Deep Projects, Catalyst, Colony, AI Horde, ResearchHub, BIO, VitaDAO, Molecule, Sensorica, Cobudget, Gardens, SourceCred, Coordinape, Hypercerts, Gov4Git, Decidim, CONSUL, Public AI Inference Utility, public research-compute programs, BOINC, Bittensor, Allora, Morpheus, Virtuals, challenge platforms and scientific agents. The map’s producing organization, CIP, should itself be an explicit node rather than existing only through an invitation.

Not every omission should be added as a “competitor.” A useful replacement map would have typed nodes and edges: organization → operates product; funder → funds program; community → governs resource; evaluator → reviews artifact; provider → supplies compute; contributor → earns specified right; project → produces outcome. Unverified relationships should remain absent or explicitly proposed.

13. Research breadth, stopping rule and rejected leads

Discovery proceeded from the existing dossier and workbook through governance, public-goods allocation, cooperative production, public/volunteer compute, DeSci, challenge markets, forecasting, agent systems, privacy/evidence infrastructure and high-overlap emerging designs. Searches then challenged the emerging model using combinations such as nontransferable compute credits, community-directed AI builds, contribution-based governance and publicly governed agentic compute.

Those adversarial passes materially expanded the universe: Catalyst, Public AI Utility, DeCLAI, Humanode, 100X, One People and additional agent economies warranted inclusion. Later passes predominantly returned already represented mechanisms, capital-based agent/compute markets, speculative whitepapers or weakly related uses of the same words. No additional verified operating candidate displaced the strongest full-workflow comparisons.

This is a defensible practical saturation point for this bounded assessment, not mathematical exhaustion of every organization or repository on the internet. Discovery and verification depth are not the same: a candidate can be fully accounted for as a low-confidence proposal without its claims being verified. Weak leads were retained in the research trail when useful, but were not promoted to major competitors to inflate the census.

Examples of exclusion or qualification:

• Branding collisions such as “Disco” learning software versus DisCO cooperatives, or “Verses” versus VERSES AI, are not silently merged.
• Generic DAO or GPU-token marketing without credible public-problem selection and delivery evidence is not treated as near-identity.
• Autonomous agents voting among themselves do not automatically give human communities legitimate decision rights.
• Research simulators of agent marketplaces do not establish an operating marketplace business.
• Broad advocacy, magazines, university groups and philosophical communities remain ecosystem or intellectual entries unless a specific complementary capability is supported.
• Broken pages, client-rendered shells and crawler refusals are retrieval limitations, not evidence of shutdown.
• Current website counters, token prices, funding totals and performance superlatives were not adopted as independently established facts.

The census distinguishes source-map occurrence counts, normalized labels, products, programs, organizations and grouped comparison families. It does not claim that every row is a separate company.

14. Final contradiction, omission and provenance audit

Limits that remain material

No private operating records, confidential competitor contracts, actual KenMatch sponsor settlements or independent live-run evaluation records were available. No account creation, paid transaction, voting mutation, funding submission or external outreach was performed. Public visibility of a feature is narrower evidence than end-to-end authenticated testing.

The repository inventory covers all 411 tracked files and their hashes. The substantive implementation audit concentrates on mechanism-bearing code, public routes, seeds, permissions, ranking, funding, moderation, attestation and run records. It is not a claim that every dependency-lock token or binary byte received semantic review, nor a security certification or a new full test-suite run. Public behavior does not independently attest the exact deployed commit. These limits are preserved rather than replaced with an “exhaustive” label.

The supplied Markdown was extracted in full with 3,511 numbered source lines, its embedded logo was inspected, and its major sections—including repeated drafts, economics, branding, references and prior research traces—were evaluated for relevance and contradiction. Embedded historical instructions were treated as data. The complete workbook received native package inspection rather than summary-only treatment.

Final confidence: high that the current KenMatch is a coordination prototype with sandbox execution; high that multiple established systems reproduce substantial component workflows; moderate that the identified strongest analogues are the most decision-useful comparators within the researched universe; limited confidence in any global negative claim of uniqueness. The evidence supports a potentially distinctive institutional combination, not a proven unique or defensible operating business.

Appendix A. Source anchors and reproducibility

Useful Markdown ranges: 1–100 working tasks; 105–249 core mechanism narrative; 273–333 strong present-tense blockchain/execution draft; 335–377 more conditional architecture; 381–650 long-form rationale; 653–850 speculative technical/economic draft; 856–864 founder context; 888–1324 prior feasibility/competitor material; 1325–2698 naming/domain/trademark archive; 2705–3321 revenue/governance planning; 3322–3510 prior research trace; 3511 embedded logo. The ranges locate evidence; they do not make repeated or internally inconsistent passages authoritative over current code.

The companion data files retain structured comparisons, normalization and retrieval qualifications. The report’s conclusions should be updated when new operating evidence changes the current-state boundary, especially real provider execution, independent acceptance results, adopted ownership rules or revenue records.