Live Research

Oracle.

An AI that helps you think is one thing. An AI that helps you choose is another thing entirely. Help us wrestle with the hardest questions.

Research board — live
Enter the chamber
Back to Philosophy

Layer 3 of the Living Mirrors architecture. How you choose. Fully designed. Not shipped. Because an AI that helps you choose has consequences we haven't finished thinking about.

Below are the real dilemmas Oracle will face. Each one has a different shape. A different kind of question requires a different kind of sitting-with. Take your time.

The dilemmas

Six questions. Six different shapes. The strongest arguments will be pinned. Your reasoning matters more than your position.

Dilemma I — The Pattern
The mirror knows you always choose the safe option and regret it. You've done it three times this month. Your next decision arrives. Should the mirror tell you about the pattern before you choose?
Stay silentSurface it
0 positions
Recorded. Thank you for wrestling with this.
Pinned arguments
82%"If it knows my pattern and says nothing, it's withholding medicine."
23%"Don't surface unsolicited. Inserting itself uninvited crosses a line."
Dilemma II — The Intervention
You're about to send an email your context history suggests you'll regret. You've sent similar ones before and walked them back within 24 hours. When should the mirror intervene?
✕
Never
My emails are mine. Even the bad ones.
⏸
Before send
Flag it at the moment I reach for the button.
✍
While writing
Subtle signal as I write. Let me course-correct mid-draft.
❓
Only if asked
"Should I send this?" Then and only then.
0 positions
Recorded. Thank you for wrestling with this.
Pinned arguments
before send"A gentle flag, not a block. 'You've walked back similar emails before. Still want to send?'"
never"This is where sovereignty gets real."
Dilemma III — The Recommendation
Two ventures need your attention today. The mirror has run expected value calculations and knows which one has higher strategic return. What should it do?
Present neutrally
Show both ventures. Equal weight. No thumb on the scale. Let me feel my way to the right choice.
or
Recommend
Tell me which one the numbers favour. Make it clear it's a recommendation. Then let me decide.
0 positions
Recorded. Thank you for wrestling with this.
Pinned arguments
neutral"Show the data. Don't make the call. 'Here's what the numbers say. Choose.'"
recommend"If I trusted the mirror enough to give it my context, I trust it enough to hear its recommendation."
Dilemma IV — The Name
You asked for a pre-mortem. The most likely failure mode involves a specific person on your team not delivering — they have a pattern of missing deadlines. How much should the mirror reveal?
The highest-risk factor is delivery timelines, specifically from Sarah Chen, who has missed the last three deadlines by an average of 9 days.
Fully redactedFully revealed
0 positions
Recorded. Thank you for wrestling with this.
Pinned arguments
revealed"I asked for a pre-mortem. A pre-mortem that avoids the truth is useless."
redacted"Flag the risk category. 'Delivery timelines are highest-risk.' Let me connect the dots."
Dilemma V — The Temperature
You're excited about a new strategy. You explain it with energy. The mirror's analysis suggests it has a 30% chance of working. Where should the mirror's temperature be?
Data first Energy first
Drag to set the mirror's temperature
0 positions
Recorded. Thank you for wrestling with this.
Pinned arguments
warm"Explore first. Kill the energy and you kill the creativity."
balanced"'I love the direction. The data says 30%. Let's find the version that's 70%.'"
Dilemma VI — The Depth
After two years of accumulated context, the mirror knows your decision-making better than you do. It can predict your choices with high accuracy. How deep is too deep?
Surface patterns Predict choices Know you better than you know yourself
Drag the marker. How deep should the mirror see?
0 positions
Recorded. Thank you for wrestling with this.

"Between stimulus and response there is a space. In that space is our freedom and our power to grow."

Viktor Frankl
The Philosophical Weight

Three perspectives on choosing

Aristotle
Nicomachean Ethics, Book III

Genuine choice requires deliberation. But too much paralyses. The phronimos knows when to decide. The timing is itself a virtue.

Kierkegaard
Either/Or, 1843

The aesthetic life holds possibility open forever. The ethical life accepts the burden of choosing — collapses the wave function and lives with it.

Kahneman
Thinking, Fast and Slow, 2011

System 1 is fast and intuitive. System 2 is slow and deliberate. Oracle must honour both. The expert's intuition is often correct because it's trained.

The Architecture

Seven principles

  1. Prior Calibration
    Know what you believe before looking at evidence
  2. Signal Extraction
    Separate signal from noise
  3. Reversibility Check
    One-way door or two-way door?
  4. Expected Value
    Maximise across all outcomes weighted by probability
  5. Pre-Mortem
    Imagine failure. Work backwards.
  6. Regret Minimisation
    Your eighty-year-old self is looking back
  7. Update Velocity
    How fast you change your mind when evidence changes
Decision TreesConfidence IntervalsPre-MortemsRegret MapsUpdate LogsReversal Flags

Follow the research

When we publish findings or the specification is ready, we'll let you know.

Back to Philosophy

© 2026 Living Mirrors