Skip to content
Forticia
Research
Quantitative FinanceEquities, FX, futures. Factors and backtests, every run replayable.Computational BiologySequence, folding, simulation. Versioned, reproducible labs.Cultural IntelligencePhilosophy, governance, ethics. How institutions decide and answer for it.AI InstrumentationPrivate models. Multi-agent orchestration. Guardrails on write.View all research
Quantitative Finance
  • Equities, FX, futures
  • Factors and backtests
  • Every run logged and replayable
InfrastructurePapersPolarisLink™About
Sign inRequest access
Request access

Research

Quantitative FinanceEquities, FX, futures. Factors and backtests, every run replayable.Computational BiologySequence, folding, simulation. Versioned, reproducible labs.Cultural IntelligencePhilosophy, governance, ethics. How institutions decide and answer for it.AI InstrumentationPrivate models. Multi-agent orchestration. Guardrails on write.

Platform

InfrastructurePapersPolarisLink™About
Request accessSign in
Forticia

A private institute for computational research. It publishes original research and runs a governed environment where every run is logged and replayable.

Sign inSystem status

Research

Quantitative FinanceComputational BiologyCultural IntelligenceAI Instrumentation

Infrastructure

WorkspacesHelper planeGoverned AIPapersStatus

PolarisLink™

SpecificationQuickstartClientsChangelogGitHub repository

Institute

AboutFAQRequest accessContactSign inForticia on GitHub

Forticia publishes research and simulations. Nothing on this site is investment advice.

Official channels: forticia.uk and github.com/Forticia

© 2026 ForticiaPrivacyTerms
  1. Home
  2. Research
  3. Cultural Intelligence

Cultural intelligence research at Forticia: How institutions
reason, and answer
for it.

The institute’s practice of how it reasons, decides, justifies and stays accountable: the philosophy, governance and ethics behind every result, kept on the record where a stranger can check it.

Request accessRead the research
Stage
Exploratory
Fields
Philosophy, governance, ethics
Method
Argument and decision analysis
Record
Lead-approved

Decision record

Question
Written before the evidence was examined.
Mechanism
Stated, including who is on the other side.
Sealed test
Examined once, with the design frozen.
Review
Independent, with a written critique.
Decision
Accepted, with its reasons.
Owner
Named.
Dissent
Kept in full.
Open to reopening
Five commitmentsHow a decision travelsArgument mapCharterUncertaintyLimits
Five commitments

Working commitments,
not a theory.

Each commitment names the failure it exists to prevent, because a failure is easier to recognise than a virtue. Open one to see what it means in practice and which ideas it borrows.

  • The failure it prevents

    A result nobody tried to break.

    In practice

    • Ask where the flaw is before asking how to use the result.
    • Remove the supposed cause and check that the effect goes quiet.
    • Read fragility to small changes as a warning, and agreement across neighbouring settings as evidence.

    Where the idea comes from

    • Popper on falsification: one counter-instance can refute a universal claim, no number of confirmations proves one
    • Feynman, Cargo Cult Science, Caltech 1974: the first principle is not fooling yourself
  • The failure it prevents

    A regularity mistaken for knowledge.

    In practice

    • Say who is on the other side and why they cannot do otherwise.
    • Match the speed of the proposed cause to the speed of the effect.
    • Decline questions the world can never supply enough events to answer.

    Where the idea comes from

    • Hume’s problem of induction, as Popper took it up
  • The failure it prevents

    Choosing the test after seeing the answer.

    In practice

    • Declare one specification in advance rather than sweeping until something fits.
    • Examine sealed evidence once. A failed sealed test is final.
    • Never tune a failed idea until it twitches.

    Where the idea comes from

    • Elster, Ulysses and the Sirens, 1979: precommitment, or binding yourself in advance
    • Lakatos on ad hoc rescue and degenerating research programmes
  • The failure it prevents

    Being fooled the same way twice.

    In practice

    • Write the autopsy, then move on.
    • Screen every new idea against what has already failed.
    • Retry a buried idea only after the standard itself improves, and then retry every case alike.

    Where the idea comes from

    • Weick and Sutcliffe on high-reliability organisations: preoccupation with failure, reluctance to simplify
  • The failure it prevents

    A decision nobody answers for.

    In practice

    • Name an owner for every output and every risk.
    • Separate those who propose, build, review and decide.
    • Keep the trail complete enough that nobody has to be asked.

    Where the idea comes from

    • Montesquieu, 1748, and Madison in Federalist No. 51, 1788: separate powers so each can check the others
    • Ostrom, 1990: monitoring, graduated sanctions and cheap conflict resolution in self-governing institutions
    • NIST AI Risk Management Framework, 2023: govern as the function that sets accountability
How a decision travels

From a question
to the record.

Every idea takes the same road: declared, tested, sealed, reviewed, decided and recorded. A failure travels it too. Choose an outcome and watch the record fill in.

The idea survives every attempt to break it. The decision is kept with everything needed to replay it.

Decision record

A worked example of the shape of a decision, filed as the passage runs.
Question
Written down before any evidence is examined.
Mechanism
Not yet filed
Exploration
Not yet filed
Sealed test
Not yet filed
Review
Not yet filed
Decision
Not yet filed
Owner
Not yet filed
Trail
Not yet filed
Argument

Objections stay
on the map.

A claim is only as good as the objections it has met. This map lays out one argument: premises, evidence, objections, replies, and the dissent that survives. Select a node, collapse a branch, or hide the dissent and see what the record loses.

2 objections · 1 answered · 1 dissent preserved

Claim/Objection

Independent review slows decisions that are time-sensitive.

A specific reason a premise or the claim may fail. It deserves an answer, not a rebuttal of tone.

Charter

Rules, standards,
and the ideas behind them.

Each clause can be written as a rule or as a standard, and each traces to the work that explains why the safeguard exists. Choose a clause to light its lines.

Six clauses of a working charter

Clause §1

Declare the question first

The question and the test are written down before the evidence is examined.

The failure it preventsSearching until something fits.

Cheap to apply and predictable. It can be gamed at the edges and miss cases nobody foresaw.

Rules cost more to write; standards cost more to apply (Kaplow, 1992). Most working charters use both.

Principles it traces to

  • Popper, Logik der Forschung, 1935. English edition 1959.

    A claim earns standing by surviving attempts to refute it. A single counter-instance can refute a universal claim; no number of confirmations verifies it.

    Stanford Encyclopedia of Philosophy: Karl Popper
  • Elster, Ulysses and the Sirens, 1979.

    Foreseeing a weakness, an agent binds itself in advance, as Ulysses did at the mast. The binding can be an internal norm or an external constraint.

    Precommitment
  • Lakatos, the methodology of scientific research programmes.

    A programme that only patches itself against anomalies, and predicts nothing new, is degenerating. Rescue by ad hoc adjustment is a warning sign.

    Stanford Encyclopedia of Philosophy: Imre Lakatos
Decisions under uncertainty

Choosing when the
odds are unknown.

Frank Knight separated risk, where probabilities can be measured, from uncertainty, where they cannot. Robust decision making stress-tests a choice against many futures instead of betting on one forecast. Move the slider and see which rules notice.

Weights each future by its probability. It needs the odds, and trusts them.

0255075100SmoothShockSlow driftRule changeSurpriseoddsA · Commit fullyB · Stage itC · Wait

Expected value chooses commit fully.

With the odds trusted, expected value chooses commit fully. With no idea of the odds, it chooses stage it. Maximin, minimax regret and robust satisficing choose wait or stage it and do not need the odds at all.

OptionExpected valueWorst futureLargest regretFutures above threshold
A · Commit fully76.515452 of 5
B · Stage it66.540254 of 5
C · Wait46.445503 of 5

Arbitrary units, chosen to show the mechanics. The column of the active rule is marked.

What we will not do
  • We will not fabricate evidence.Where evidence is missing we say null, and say what would fill the gap.
  • We will not tune a failed idea until it twitches.A buried idea may return only when the test itself improves, and then every affected case is retried alike.
  • We will not grade our own work.Someone else re-derives it from raw inputs and files a written critique, whoever the author is.
  • We will not claim readiness we have not shown.Done means a stranger can replay it. Until then the honest states are parked, blocked or degraded.
  • We will not keep what we learn from failure private within the team.The autopsy is filed where everyone can screen new ideas against it.
Where we are still thinking
  • Where does machine judgement end and hard rule begin?Models can propose and interpret. Rules should guard whatever cannot be undone. Where exactly the seam sits is still being worked out.
  • Who chooses the frame, and who watches them?Pre-declared gates remove discretion. Class-aware frames put some back. Someone decides which class a case belongs to.
  • Is it a better test, or a rationalised rescue?A verdict may be reopened when the standard improves. The line between improvement and rescue is a judgement.
  • What keeps a standard legitimate when one hand writes it and systems apply it?Process, separation of roles and a record help. They do not settle it.

Latest papers

Cultural Intelligence

All papers
  • 5 Oct 2026Never Tune a CorpseWhen may a rejected idea be reopened? Simulated burial and exhumation rules, their false discovery cost, and how the graveyard can audit the court.16 min read
  • 5 Oct 2026When Approval Stops Carrying InformationA conditional-information measure of rubber-stamping, the defect rate where review rationally collapses, and which levers restore it. A simulation study.18 min read

RelatedGoverned AI and the audit trail

Enquire about
this branch.

Workspaces are invite-only and role-gated, and an AI collaborator drafts by default: a lead approves before anything is written, with a full audit trail. Tell us what you would like to examine.

Request access

Further reading

  • Karl Popper, Stanford Encyclopedia of Philosophy
  • Montesquieu, Stanford Encyclopedia of Philosophy
  • Federalist No. 51, Madison, 1788
  • Merton, The Normative Structure of Science, 1942
  • Estimating the reproducibility of psychological science, Open Science Collaboration, 2015
  • Ostrom’s eight commons governance design principles
  • Rules versus standards: an economic analysis, Kaplow, Duke Law Journal, 1992
  • Robust Decision Making (RDM), Lempert and colleagues
  • Explained: Knightian uncertainty, MIT News
  • NIST AI Risk Management Framework 1.0
  • Precommitment, Elster, Ulysses and the Sirens, 1979
  • Imre Lakatos, Stanford Encyclopedia of Philosophy
  • Cargo Cult Science, Feynman, Caltech 1974
  • The five principles of high-reliability organisations
  • Goodhart’s law
  • Toulmin model of argument