ELEffective LibertyWorking draft 0.1

A four-session pilot for an open inquiry

Freedom after powerful intelligence.

Effective Liberty is a working inquiry into agency, institutions, and powerful AI. Personal AI sovereignty and the other models explored here are candidates to test—not commitments participants must share.

Explore the sessionsRead the charter

Built to be argued with. Every claim here is provisional.

Our working question

What institutions and norms would allow people to retain meaningful agency in a world of increasingly powerful AI?

Where
3562 18th Street
San Francisco, CA 94110
WhenDates and times forthcoming

Each meeting is capped at 90 minutes.

On handCold brew coffee on tap

Available at each meeting.

A practical research program

From moral value to design problem.

Effective Liberty treats liberty as a practical research and design problem: to be understood clearly, measured where useful, and expanded through institutions, cultures, and technologies.

Its methodological inspiration from effective altruism is seriousness about evidence, intervention, and results—not a commitment to a single aggregate metric or centralized optimization.

Starting hypothesis · not a membership test
Effective liberty is a person’s protected, practical capacity to understand, choose, and pursue a self-directed life without being subject to arbitrary power.
01

Protected choice

Which choices must be secure from coercion or arbitrary interference?

02

Practical capacity

What resources and capabilities make formal freedom real?

03

Self-direction

What lets a person understand and author the course of their own life?

01 / The epistemic charter

A constitution for inquiry, not an ideological test.

A shared question and serious method, without a required answer.

Participants can challenge personal AI sovereignty, decentralization, competing institutional models, or the group's own framing. Effective Liberty names the inquiry, not a doctrine participants are asked to adopt.

  1. 01

    Steelman the strongest opposing case before rebutting it.

  2. 02

    Separate empirical claims, moral premises, and institutional proposals.

  3. 03

    Expose assumptions, control points, and failure modes.

  4. 04

    Treat favored ideas and founding assumptions as revisable.

  5. 05

    Require serious engagement, not agreement.

  6. 06

    End by recording what remains open and what could change minds.

02 / The pilot

Four meetings to sharpen the inquiry.

Each session is capped at 90 minutes: 15 min light framing, a maximum 60 min focused block (40 min discussion + 20 min objections and synthesis), then 15 min closing reflection. Dates and times are forthcoming.

Session 01 · Values before mechanisms

What is liberty for?

What kind of freedom should remain morally important when intelligence and material capacity change radically?

1 of 4

Questions to press

  • Is liberty non-interference, non-domination, self-authorship, effective power - or a combination?
  • When, if ever, may aggregate welfare override individual choice?
  • Which constraints should remain non-negotiable under extreme stakes?

Suggested readings

Friedrich HayekThe Use of Knowledge in SocietyStanford Encyclopedia of PhilosophyRead §1.2 only: Liberty as Non-DominationA short, open introduction to republican freedom as independence from arbitrary power.
Working output

Critique and refine the starting hypothesis, including rival definitions worth retaining and candidate non-negotiable constraints.

03 / The working vocabulary

Concepts under construction.

These are handles and candidate models for inquiry, not articles of faith. Open a card to see the live tension behind each idea.

01Effective libertyStarting hypothesis

A person's protected, practical capacity to understand, choose, and pursue a self-directed life without being subject to arbitrary power.

Unresolved

How should non-interference, non-domination, rights, resources, and self-authorship fit together?

02Equal AI sovereigntyCandidate model

One model to test: give each person an equal, durable share of AI-enabled agency rather than dependence on one state, firm, or optimizer.

Unresolved

Can equal founding power stay independent when infrastructure and coalition power are unequal?

03Personal fiduciary agentsCandidate model

One possible design: highly capable agents loyal to one individual, not hidden provider or state principals, and not paternalistic moral guardians.

Unresolved

What proves loyalty, protects vulnerable principals, and resolves conflicts among agents?

04Sybil resistanceCandidate design constraint

If AI-enabled power is distributed per person, cheaply creating copies, uploads, or nominal identities must not mint unlimited new claims.

Unresolved

Can uniqueness and eligibility be legitimate without universal surveillance or exclusion?

05Gradual eligibilityCandidate mechanism

A possible alternative to a cliff: full founding claims in an initial window, then a transparent declining issuance schedule.

Unresolved

No decay formula, duration, or complete account of justice for later-created persons has been chosen.

06External law + symmetric enforcementCandidate institutional model

One model puts legitimate constraints in contestable law and capable public enforcement rather than inside each personal agent.

Unresolved

Detection, defense, infrastructure, authority, and due process must remain symmetric and resistant to capture.

04 / How we work

Make disagreement legible and useful.

The pilot brings a small, varied cohort into a repeatable practice of careful disagreement—without treating consensus as the measure of success.

Small, varied cohort

Keep the pilot small enough for trust and continuity, while inviting genuinely different intellectual starting points.

Steelman first

Reconstruct an opposing case in terms its advocate would accept before offering a rebuttal.

Make assumptions visible

Name empirical claims, moral premises, institutional proposals, control points, and failure modes.

Map disagreement

Record the cruxes and competing models instead of smoothing over a disagreement for artificial consensus.

Ask what would change my mind?

End with the evidence, argument, or experience that could move each view—and the questions still open.

05 / Hosts

Hosted by

Portrait of Eric ChristiansenEric Christiansenemchristiansen.com ↗Portrait of Alex KilkkaAlex KilkkaLinkedIn ↗

The intended outcome

Clarify first. Then choose what comes next.

The pilot aims to clarify the most important questions, disagreement maps, competing models, and strongest objections. After session four, the group will decide whether—and in what form—the inquiry should continue.

For the next revision: Which session feels weakest? Which reading is missing? Which favored premise has not met its strongest critic?