Where can MCIv1-9 rev5 be used in the real world?

 

Where MCI rev5 actually has bite

MCI looks abstract, but each layer maps onto real problems people are actively trying to solve. Here are the strongest places it could earn its keep.

AI safety and alignment (its home turf)

This is what MCI was built for. The current alignment conversation is mostly "how do we get AI to do what humans want?" MCI reframes it: what kind of system is worth aligning with?

• V1–V4 gives evaluators something better than vibes. Instead of "this model seems safe," you can test: does its reasoning gain from stress (antifragility), does it route decisions to the right scale (subsidiarity), is its power contestable, do its justifications survive real scrutiny, does its self-monitoring fire proportionally?
• V5 is the diagnostic for the worst trap in alignment — a model that's been trained on safety literature and produces perfect safety-flavoured outputs without actually being safe. The "constitutional fluency" concept names something the field talks around but hasn't crisply defined.
• V6 maps directly onto the problem of how to update an AI's values without value drift.
• V7–V8 apply to multi-agent AI systems and AI-AI coordination, which is becoming a live concern as agentic systems proliferate.
• V9 Stewardship is the missing piece in most alignment frameworks: what does an AI owe to populations who can't speak for themselves (future generations, ecosystems, people without internet access)?

Institutional and organisational design

Strip out the AI framing and MCI is a theory of how institutions stay legitimate over time. That's a real problem.

• Regulatory bodies (FDA, FCA, central banks) — V1's five virtues plus V6's revision pipeline give a clean structure for how a regulator should update its rules without being captured by the industry it regulates. The "adaptive capture" failure mode is exactly what regulatory capture looks like.
• Tech platform governance — Meta's Oversight Board, Wikipedia's arbitration, Reddit's mod structure. V7's four Compact provisions (nested authority, mutual contestability, discursive justification, cross-system monitoring) are a checklist for whether a governance body is real or theatre.
• Corporate ethics functions — Distinguishes companies with ethics committees that perform ethics from ones where ethical commitments are structural.

International agreements and treaties

V7 (Compacts) and V8 (Initiative) are basically a theory of legitimate treaty-making.

• The "compact hegemony" failure mode describes what critics have said about Bretton Woods institutions, WTO, and some climate negotiations: the form is constitutional, the substance is captured by the most powerful party.
• V8's four initiative checks (especially the structural veto for affected parties) cut hard against treaties drafted by powerful states and presented to weaker ones as fait accompli.
• Climate negotiations would look very different if framed as a V8 origination process with operative vetoes for affected populations rather than just participating governments.

Commons governance

This is where Ostrom's influence shows up most directly. MCI is partly a generalisation of her design principles.

Fisheries, irrigation systems, open-source projects, scientific peer review, shared computational infrastructure — anywhere multiple parties depend on a shared substrate. V7's compact provisions and V9's stewardship duties translate almost directly.

Personal use — life decisions and self-understanding

This is where teenagers and adults actually meet the framework day-to-day.

• As a lens on character — V5's pivot is a question worth asking yourself: is honesty (or kindness, or courage) something you do, or something you are? The four tests work on people.
• Decision-making — V4's goal vector (content, scope, affected, power, reversibility, stress) is a checklist for big life decisions. Should I take this job? Move countries? Have this conversation? The vector forces you to be honest about who else is affected and what you can undo.
• Spotting bad-faith institutions — Once you can name "consultation theatre," "ritual justification," and "contestation by strawman," you see them everywhere. Your workplace, your government, your relationship.

Professional practice

• Journalism and editing — V2's eight-stage pipeline is a strong frame for editorial process, especially Contestation and Discursive Justification.
• Medical ethics committees — V4's alignment check applied to treatment decisions, especially around irreversibility and affected-party voice.
• Legal reasoning — Pettit's non-arbitrariness has direct purchase on judicial review and administrative law.
• Research ethics (IRBs) — V9 stewardship duties for research populations who can't fully consent.

Where it's weakest

Honest answer: MCI is strongest as a diagnostic tool and weakest as a building specification. It's much easier to use it to identify what's wrong with an existing system than to use it to build a new one from scratch. The mechanisms are specified at a level of abstraction that requires a lot of translation to become engineering.

Also, parts of it (especially V5's pivot and V9's Ground) make claims that are very hard to test empirically — which the framework itself acknowledges. They're useful concepts but not falsifiable in the strict sense.

The honest one-line summary

It's most useful right now as a vocabulary — names for failure modes that are real but usually go unnamed, which is half the battle in fixing them.