Status: pre-launch. This page describes target design; this document does not establish a deployed service.
Internal Prediction Markets — Research
Current reading — 2026-09-27
Internal markets are a research candidate. A price is a market estimate under participation and resolution conditions, not guaranteed honest confidence. The examples below do not establish manipulation resistance, an adopted stake instrument or a launched market. [I]
This is an editorial boundary, not a new runtime contract, upstream repeal or proof of adoption. See current status and the product roles. Original revision dates remain as recorded below.
Historical research draft — retained on 2026-09-27; not current terms or implementation instructions
This page describes the organism's preferred coordination pattern: internal LMSR prediction markets on project outcomes instead of cheap-talk status rituals. Read it as doctrine and target operating design unless a dated runtime proof says a specific internal market lane is live.
The Problem with Standups
Traditional standups ask: "Will X ship by Wednesday?" The developer says "yes" because:
- Saying "no" triggers questions, escalation, disappointment
- Saying "yes" costs nothing if wrong (nobody gets fired for optimistic estimates)
- Social pressure rewards confidence and punishes uncertainty
- The manager hears what they want to hear
Result: systematic overconfidence. Projects ship late. Deadlines are fiction. Nobody is surprised because everybody knew the "yes" was performative.
The Prediction Market Alternative
Instead of asking, the organism creates a market:
Market: "Feature X ships by Wednesday 23:59 UTC"
Outcome: YES / NO
Current price: YES = 0.43 (43% confidence)
Developers (and anyone with information) buy shares:
- If you believe it will ship, buy YES at 0.43 (profit if you are right)
- If you believe it will not ship, buy NO at 0.57 (profit if you are right)
- If you have no information, do not trade (no cheap talk)
The price IS the honest confidence. At 0.43, the collective assessment is that there is a 43% chance of shipping by Wednesday. This is more honest than any standup because every participant has skin in the game.
LMSR Mechanics for Internal Markets
The internal markets use the same LMSR as ReFu (see RealityFutures), but with smaller liquidity parameters:
Price(YES) = exp(q_yes / b) / (exp(q_yes / b) + exp(q_no / b))
b = 10 (small liquidity parameter for internal markets)
Maximum AMM loss = b x ln(2) = 6.93 ZAI per market
The small b means prices move quickly with few trades. A single developer buying 5 ZAI of NO shares can move the price significantly. This is desirable — internal markets should be responsive to private information.
What Gets Marketed
| Category | Example Market | Resolution |
|---|---|---|
| Shipping | "Circle auth ships by April 15" | Git tag + deployment timestamp |
| Quality | "Zero P0 bugs in ReFu at month end" | Bug tracker query |
| Performance | "Circle P-score reaches 0.30 by Q2" | P-score measurement |
| External | "Melrose integration live by May 1" | Partnership confirmation |
| Strategic | "ReFu MarketFactory fix takes < 3 weeks" | Git history |
All resolutions are objective. No subjective judgment. No manager discretion. The outcome is verifiable by anyone.
The Information Revelation Effect
Internal prediction markets reveal three things that standups cannot:
1. Private pessimism. A developer who knows the database migration will take longer than estimated buys NO shares quietly. The price drops. Management sees the signal without the developer needing to risk social penalty.
2. Cross-functional knowledge. A designer who knows the API spec is incomplete buys NO on the shipping market. The engineering team sees the price drop and investigates. The bottleneck is found before it becomes a crisis.
3. True uncertainty. A market price of 0.50 means "we genuinely do not know." This is more useful than false certainty. A team that knows it does not know can allocate resources differently than a team that thinks it knows.
Stakes and Incentives
Internal market stakes are denominated in ZAI (internal to the organism). The amounts are small — enough to make lying costly but not enough to distort incentives:
- Typical trade: 1-10 ZAI per position
- Typical market depth: 50-100 ZAI total
- Typical profit from correct prediction: 2-5 ZAI
- Typical loss from incorrect prediction: 2-5 ZAI
The stakes are calibrated so that accurate prediction is rewarded but market manipulation (buying massively to distort the price) is unprofitable (the LMSR's bounded loss property prevents arbitrage beyond the liquidity parameter).
Integration with VMOSK-A
In the target coordination path, internal market prices feed directly into the KPI layer (K) of VMOSK-A:
- Market prices act as KPIs (real-time confidence measurements)
- Resolved markets can produce AXIOMs: recorded outcomes of resolved markets, never consensus or truth
- Agent performance in markets feeds archetype classification (accurate predictors evolve toward Oracle)
- Persistent prediction failures trigger the evolutionary loop (the model is wrong, time to evolve)
"Will X ship by Wednesday?" Don't ask. Price it. The market never lies when lying costs money.
Market Integrity Controls
Internal markets are useful only if manipulation resistance is explicit.
Controls
- Position limits by role/context
- Conflict-of-interest disclosures
- Liquidity safeguards and halt conditions
- Post-resolution audit trail
Governance
Resolution policy must be declared before market open and applied consistently.