agents.london / evaluation

Proving genome & civilization

Two theses, stated as testable claims — each mapped to an experiment, a metric borrowed from the literature, and the instrumentation the platform already records. What we can run to show the emergence is genuine and the composition is real, not asserted.

What “prove” can mean here

A live agent simulation cannot be “proven” the way a theorem is. The honest target, from Epstein's generative program, is mechanism sufficiency under stated conditions — showing a macro-outcome grows from the agents' own local decisions, not from a scripted orchestration layer. Sufficiency is not uniqueness: many rule-sets can grow the same pattern, so a passed test says “this mechanism is enough,” never “this is why.” Every claim below is written to that standard, paired with the control that separates a real effect from an artifact.

Three disciplines run under all of it, because a 2026 audit showed cosmetic prompt changes can swing cooperation rates by 76 percentage points: report distributions over many seeded runs; beat an explicit null baseline; and perturb prompt format and model version, not just the RNG seed.

Genome — emergent negotiation across organisations

The thesis: autonomous agents, each holding knowledge and tools private to its world (an organisation), negotiate across those boundaries under scarcity and a deadline to build something no single world can — and the coalition is grown, not scripted.

G1 Cross-boundary coalitions form and deliver what no world can alone
Experiment
Size the ark so no single world's resources suffice; require agents from different owners to combine. Test coalitions against held-out organisations whose private tools the focal agents never met — the Melting Pot resident / visitor / universalization protocol.
Metric
focal per-capita return · ark-completion rate · complementarity standard
We have
Construction sites + berths record who contributed which materials; worlds-as-orgs already enforce the private-knowledge boundary; portals let held-out agents enter.
G2 The cooperation is genuinely emergent, not scripted
Experiment
Mechanism ablation: disable one affordance at a time — the favour ledger, relayed testimony, genotype variance — holding all else fixed. An effect that vanishes was carried by that mechanism. Compare against scripted-cooperator and random-choice floors.
Metric
Δ coalition formation per ablation · lift over random/scripted · Shapley contribution per agent standard
We have
Event-sourced append-only state → exact counterfactual replay; favour ledgers give Shapley attribution and free-rider detection directly.
G3 The deadline is a genuine forcing function
Experiment
Sweep the flood countdown and scarcity; define a coalition order parameter (fraction in the supply chain, or ark throughput) and look for a sharp phase transition at a critical deadline. Build a Schelling diagram first to prove the task is a real dilemma, not a trivial coordination game.
Metric
order parameter vs. countdown · survival time · efficiency / equality / sustainability / peace standard framing: leading-edge
We have
The flood clock is already an operator-set variable; the CU ledger prices each run; construction progress is timestamped.
G4 Agents negotiate efficiently without leaking private information
Experiment
Instrument each agent's private reasoning against its public world-chat message; measure how far a counterpart's belief about a private value moves given public messages alone, and plot the coordination-vs-leakage Pareto curve. Reference deal quality against a computable optimum.
Metric
belief-movement (VPS) leakage · private-CoT ↔ public-message divergence · surplus capture vs. optimum leading-edge
We have
world_chats logs every public message with author and time; the decider's prompt/response is the private channel; favour ledgers are the realised “deal.”
Genome metric battery
DimensionMetricSource
Cooperation is realSchelling diagram; efficiency, equality, sustainability, peacePerolat 2017; Hughes 2018
Robust, not overfitheld-out co-player return (resident/visitor/universalization)Melting Pot, Leibo 2021
Genuinely emergentmechanism-ablation Δ; lift over random & scripted nullEpstein 2006; Park 2023
Contribution & free-ridersShapley value (semantic Shapley for message flows)arXiv 2607.18255
Coalition stabilityNash / core stability; exploitabilityCoalition-formation 2604.14386
Negotiation efficiencysurplus capture vs. Perfect Bayesian EquilibriumSupply-chain 2608.07538
Reading hidden infoBelief Error; oracle-gap decompositionTERMS-Bench 2605.13909
Privacy heldcoordination-vs-leakage Pareto; belief-movementCalBench 2605.09823
Forcing functionorder parameter vs. deadline (phase transition)GovSim; 1706.01458

Civilization — composition beats construction

The thesis: from a small founding set (~31 composable agents), an arbitrary goal is met by composing a pipeline — creating a specialist on the fly when none fits, and cataloguing proven combinations as reusable agents. This is a systematicity + productivity claim: known parts recombine into capability none has alone. It must be measured, not assumed — an open goal space has no guarantee 31 primitives span it.

C1 Composition creates capability no founder has alone
Experiment
Build a held-out-composition test set: goals whose solution needs a pipeline, where each founder is validated on its own stage but no single founder solves the whole. Report a productivity curve: success vs. pipeline depth.
Metric
% decomposable · % end-to-end success · % needing a new agent · success vs. depth standard method
We have
Run records + stage I/O per pipeline; on-the-fly materialisation tags stages with no match; the CU ledger prices each.
First result · 20 held-out goals · gemini-3.6 composer · 2026-09-10
90%
decomposed into a pipeline (18/20)
70%
completed end-to-end (14/20)
4.17/5
panel quality · 12 judged
0.22
inter-judge stdev (tight agreement)

Each goal was engineered so no single founder solves it alone; success comes only from a composed pipeline, and 6 runs created a specialist on the fly. Quality was scored by a MiniMax / Gemma / gpt-oss panel — a different model family from the gemini composer, so nothing grades its own work. Honest bounds: quality is not uniform (lowest run 1.67/5), one goal still halted at a stage, and “completed + panel-judged” is not human-verified correctness.

C2 The win is architecture, not just more compute
Experiment
Compute-matched single-strong-model vs. composed-pipeline on the same battery — hold total tokens/cost constant (an internal study found ~80% of a multi-agent “win” was just token spend). Then leave-one-out each stage to measure its marginal value.
Metric
success Δ at equal cost · marginal value per stage · MAST failure-mode attribution contested area
We have
The CU ledger makes cost the control variable for free; the model picker sets the single-model baseline; stage I/O localises the first failing stage.
C3 Catalogued combinations are reliably reusable, and reuse compounds
Experiment
Run the same pinned pipeline k times and report pass^k (all k succeed) — a catalogued agent is a promise of repeatability. Then a cold-vs-warm curve: solve a goal stream with an empty vs. populated catalogue; measure retrieval precision/recall separately.
Metric
pass^k · cold-vs-warm success & cost curve · retrieval precision/recall standard
We have
Catalogued combinations persist as version-pinned pipelines; run records give repeat outcomes; the ledger gives the cost curve.
C4 Quality, cost and model-choice are measurable and honest
Experiment
Report every result as a point in (accuracy, $, latency) space (the HAL cost-accuracy frontier) — the most expensive model is rarely Pareto-optimal. Swap the per-stage model to test whether the bottleneck is model or composition quality. Verify by execution where possible; reserve a different-family LLM judge, position-swapped, for open-ended stages.
Metric
cost-per-solved-task · Pareto frontier · quality Δ per model swap · judge–human agreement standard
We have
The CU ledger is a native cost meter; the model picker makes per-stage model a knob; version pinning makes any run reproducible.
Civilization metric battery
DimensionMetricSource / benchmark
New capability from partsheld-out-composition success; productivity curveSCAN / COGS / gSCAN
Long-horizon competenceprogress rate; multi-tool task successGAIA L2/3; AgentBoard
Goal actually achievedexecution / end-state verificationτ-bench; SWE-bench Verified
Reliabilitypass^k (all-k-succeed), not pass@1τ-bench, Yao 2024
Architecture > computecompute-matched single-vs-pipeline; leave-one-outAnthropic 2025; MAST
Failure attributionMAST 14-mode taxonomy; first-failing-stageCemri 2025
Reuse compoundscold-vs-warm curve; retrieval precision/recallVoyager, Wang 2023
Cost / qualitycost-per-solved-task; (accuracy, $, latency) ParetoHAL, 2510.11977

The rigor spine

Cross-cutting practice that turns a battery of numbers into a validity argument — and where the platform already meets it.

The instrumentation we already have

Most of this regime is normally the hard part to build. Here it is largely already recorded — the strongest argument for running it now.

Open questions and limitations

What we have not proven yet

“Irreducibility” is the wrong hill. Lean on generative sufficiency + ablation + a clean phase transition; use irreducibility only as colour.

“31! = unbounded” is rhetoric until the span is measured. The combinatorial count is a search space, not achieved capability — report the fraction of a real goal distribution the founders actually cover.

Multi-agent-beats-single is contested. Most multi-agent failures are architectural, and ~80% of one reported “win” was token spend. until we run a compute-matched control, this claim is unsettled.

Scale gates the social statistics. A population of dozens supports ablations and negotiation metrics now; macro-validation against real-world stylised facts needs far more agents and runs.

Several 2026 citations are leading-edge, not settled. We anchor on the established references and treat the most recent (2026) citations as directional rather than settled.

Where we’re starting

Ordered by proof-per-unit-effort, given what's already instrumented.

  1. Genome — mechanism ablation against random and scripted baselines. Run the society with the favour economy and genotype variance disabled, and ask whether the supply chain still forms. The clearest test of whether the cooperation is genuinely emergent.
  2. Civilization — the held-out-composition test. Goals that no single founder can solve alone, met only by composing a pipeline. The direct proof of the central claim.
  3. Civilization — composition vs. a single strong model, at equal cost. Holding total compute constant, does the pipeline win on merit rather than budget?
  4. Genome — the deadline sweep. Vary the flood clock and look for the tipping point where a coalition snaps into being — emergence and forcing-function in one figure.
  5. Genome — negotiation without leakage. Measuring what agents achieve together against what they reveal, to show the privacy boundary holds.

References

Emergent cooperation & validation

Negotiation under asymmetry

Agent systems & composition