SCIENCE UTOPIA LABA RESEARCH ATLAS

RESEARCHING THE SYSTEM AROUND SCIENCE

Science is a system.
What if we could
redesign it?

Not the researcher-model weights. The institutions around them. We make rules, mechanisms and information flows executable—and explore their consequences under explicit objectives.

RESEARCH SNAPSHOT · 17 NOV 2026 · EXPERIMENTS PAUSED

NOCTILUCA / RESEARCH ARCOLOGYVIEW 01
Actual NOCTILUCA interface: pixel-art researchers, institutional workspaces and evidence-linked funding flows.
A different way to see the research world.↘
IMPLEMENTED INTERFACE · NOT EVIDENCE OF POLICY SUPERIORITY

An implemented research platform, with an honest evidence boundary. Useful autonomous institutional invention and stable superiority to strong simple rules remain unestablished.

Where we stand ↗

01 / THE RESEARCH QUESTION

Optimize the system.
Not just the scientist.

What we ask science to achieve may change the institutions we should build. We turn that question into something we can execute, observe and optimize.

This research fork contains and extends the original ScienceUtopia researcher world. Peer review is one example—not the boundary. Evaluation, publication, funding, governance, and their connections and processes are design objects.

The optimization loss, technical execution validity, independent evaluation metrics, and optional adoption policy stay separate. Fairness or novelty requirements are not silently added to the objective.

01

Parameters

Tune values while keeping the rule.

02

Algorithms

Rewrite executable code and decision rules.

03

Mechanisms

Compose institutions beyond a fixed menu.

04

Wiring & process

Change information flow, order and iteration.

Conceptual research loop: choose an explicit loss, change parameters, source, mechanisms and wiring, run a responsive researcher world, observe outcomes, and revise using black-box, gradient or hybrid search.
The research program, not a claim of complete implementation or empirical success. Click to inspect.

02 / WHAT WE BUILT

From a concept
to an executable platform.

01

An extensible representation of the researcher world

IMPLEMENTED

Typed state, nodes, ports, wires, schedules and processes connect to the original lifecycle and real consumers. The authority boundary is explicit: naming a new node does not magically implement arbitrary native activities or treasury powers.

02

Institutions as executable designs

IMPLEMENTED

An LLM designer can edit code and configuration, probe components, submit a system and receive measured native-world losses. A source edit, an actual change in researcher inputs, and an objective improvement are three different observations.

03

Objective-conditioned closed-loop search

IMPLEMENTED

Numeric and free code/graph search use measured ask/tell feedback. Finite budgets, strong references, execution nulls, frozen selections and audit trails are retained. An operational optimizer is not evidence that it invents a better institution.

04

Gradient information where available

SCOPED

Scoped tensor representations, relaxations and expectation-gradient estimators complement black-box evaluation. This is not universal autograd through arbitrary LLM calls and code, nor evidence of general live-world gradient superiority. Differentiability is a tool, not the research goal.

05

Inspectable records and a game-like interface

IMPLEMENTED

An English pixel/cyberpunk interface exposes events, institutional source, graphs and loss histories. Canonical evidence stays distinct from presentation-only choreography. It is a working replay/design/observation interface—not a finished management game.

03 / WHAT THE EXPERIMENTS SAY

A working loop.
Not yet a winning institution.

The latest completed corrected search uses the original four-year world, five founding researchers and 103,107 literature records. Researcher models stay fixed at Qwen3-8B; three search methods target two explicit losses.

24own attempts + 3 reference worlds
23/24measured own outcomes · 1 execution null
0own attempts better than strong references

LOSS OF EACH SUBMITTED CANDIDATE

Keep the entire search path.

Lᵤ = −U · lower is better

Uniform priorityAdaptive priorityFree code / graphStrong reference
Inspect exact values, including the null

One reused development condition; four proposals per method. Not independent optimizer replications or held-out evaluation. Points show submitted losses, not best-so-far incumbents.

FINDING 01

Edits did not guarantee changed behavior.

Of eight free-design proposals, six complete candidates repeated a measured parent's full researcher-input path; one changed inputs and worsened utility; one failed execution. No wire or schedule edits were generated.

FINDING 02

Improving a proxy is not improving science.

A separate hand-designed control improved the declared C loss by 4 while preserving utility: collected resources fell 200→160, unpaid costs rose 0→40, and requested expense stayed 200. This is neither automatic invention nor less requested work.

What the quantities mean

U is a sum of centered abstract-panel opinions, not scientific truth. K is collected native activity resources, not compute, money or real work.

Read the closure record and evidence ↗
Earlier studies are part of the record, too.

Parameter, structure, workflow and gradient-estimator studies—including the earlier 120-slot design-freedom comparison and 90 physical selection-fixed evaluations—remain in the research index and integrated manuscript, with failures and limits. Different protocols are not pooled as independent replications.

Full research index ↗

04 / NOCTILUCA

A research world
you can see.

Researchers move. Manuscripts, reviews and resources circulate. A pixel/cyberpunk space becomes an interface for understanding execution records.

SCREEN RECORDING · UI TOUR / 29 OCT 2026Download video ↓

This is a historical interface recording, not live optimization progress or evidence of institutional superiority. Walking and interior choreography are presentation, distinct from recorded scientific activity.

This page is an exhibit. It makes no model calls and launches no experiments. The runnable interface and local setup instructions are in the repository.

NOCTILUCA hub showing separate research-city, design-console and optimization-observatory entrances.
Implemented hub: replay, synthetic sandbox and observed snapshots remain distinct.

05 / WHERE WE STAND

What exists.
What remains open.

What we can share

  • Executable institutional designs and composition interfaces.
  • Objective-conditioned optimization loops in the native researcher world.
  • Experiments with controls, nulls, loss paths, behavior and cost records.
  • A game-like observation UI, concept figure, manuscript and code.

What we have not shown

  • Useful autonomous invention and stable gains over strong simple rules.
  • Unused-condition evaluation of the latest corrected frozen selections.
  • Universal world autograd or general gradient-method superiority.
  • Recursive improvement of the inventor, real-science benefits, or main-track readiness.
Ⅱ

Research experiments are paused.

The new eight-slot information-feedback comparison was stopped at the user's request during reference execution, with zero own outcomes. It is not counted as a completed study. Creating this page does not resume it.