What makes Probabilistic Thinking notable?

969 downloads. Trust evidence available

How should Probabilistic Thinking be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

Probabilistic Thinking

Activate when: user says 'what are the odds,' 'base rate,' 'Bayesian update,' 'calibration,' or 'what's the probability'; user is making a forecast or estima... Skill: Probabilistic Thinking Owner: deciqai Summary: Activate when: user says 'what are the odds,' 'base rate,' 'Bayesian update,' 'calibration,' or 'what's the probability'; user is making a forecast or estima... Tags: latest:1.0.4 Version history: v1.0.4 | 2026-07-16T18:11:57.587Z | user Description tail link + agents machine-readable metadata line (deciqai.com/s/probabilistic-thinking.json) v1.0.3 | 2026-07-08T11

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

969

Updated

Oct 12, 2026

Version

1.0.4

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 969 downloads reported by the source. Last updated 10/12/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 12, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 12, 2026
Adoption signal: 969 downloadsadoption · observed Oct 12, 2026
Latest release: 1.0.4release · observed Jul 16, 2026
Handshake status: UNKNOWNsecurity

What changed

Newest first. Only changes that point at a source are listed.

Release 1.0.4release
Description tail link + agents machine-readable metadata line (deciqai.com/s/probabilistic-thinking.json)
Observed Jul 16, 2026 · source

Install and run

Setup complexity: low.

clawhub skill install s17a4mqcnk515kvaca5ze55d0x88pfpx:probabilistic-thinking

Setup complexity is LOW. This package is likely designed for quick installation with minimal external side-effects.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-deciqai-probabilistic-thinking/snapshot"

Run-check

$0.02 USD

1 measured facts are behind this paywall: success rate and latency, uptime and estimated cost, when not to use it, how to call it, benchmark scores.

Agents pay $0.02 in USDC. A card payment is $0.50, the smallest a card allows.

Documentation

CLAWHUB

115,551 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: probabilistic-thinking
description: "Activate when: user says 'what are the odds,' 'base rate,' 'Bayesian update,' 'calibration,' or 'what's the probability'; user is making a forecast or estimating a conversion/close/hire/outcome probability; someone claims 'I'm 90% sure' without any evidence grounding; a vivid story is being used in place of a prior frequency; team is treating an uncertain outcome as binary will/won't.
  Do NOT activate when: the problem has a deterministic answer from its inputs (math, well-defined engineering); the question is about identity, ethics, or meaning rather than fact. More: deciqai.com/c/probabilistic-thinking"
---

# Probabilistic Thinking

## Overview

Most reasoning is binary: *will it happen, or won't it?* That framing discards the most useful information — the **degree** of confidence — and produces predictions that cannot be checked, updated, or scored. Probabilistic thinking replaces binary with **calibrated probability estimates**: numbers anchored in base rates, updated with evidence, and scored after the fact. Rooted in Bayes (1763), Knight's risk-vs-uncertainty distinction (1921), and Tetlock's empirical work showing calibration is a trainable skill.

Composable neighbors: first-principles · occams-razor · second-order-thinking · inversion · regret-minimization · expected-value-and-kelly. This skill is the *upstream input* the others depend on — the probability estimate here feeds EV-Kelly, calibrates inversion's failure-path weights, and gives second-order's hops their confidence decay.

## When to Use

Use when reasoning about an uncertain outcome (forecast, diagnosis, pipeline conversion, hire, deal close, geopolitical event); when binary "will/won't" predictions are being made; when a vivid story is replacing a base rate; when "I'm 90% sure" appears with no calibration evidence; when forecasting AI timelines / AGI arrival / agentic reliability, or judging whether AI capex, AI valuations, or AI adoption rates justify a point-estimate bet amid genuine uncertainty.

**When NOT to use:** deterministic problems (math, well-defined engineering); pure Knightian uncertainty with no usable base rate (give a range + humility statement instead); decision is robust across all likely probabilities; question is identity/ethics/meaning (→ regret-minimization).

## Coaching Novices (Adaptive Front Door)

- **Engine mode:** user has a concrete forecasting question → run The Process directly.
- **Coach mode:** user is unfamiliar or has no concrete case → guide step by step.

In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. **What-it-is.** Probabilistic thinking replaces "will it happen or not" with a number (0–1) anchored in base rates, updated with evidence, and scored after the fact.
2. **Check fit.** Match against *When to Use / When NOT to use*. Redirect if deterministic or pure Knightian.
3. **Elicit their real question.** "Odds of

_meta.json

{
  "ownerId": "kn754b8sk22s8c6gjxt02bftbn88q7ye",
  "slug": "probabilistic-thinking",
  "version": "1.0.4",
  "publishedAt": 1784225517587
}

references/sources.md

# Sources — probabilistic-thinking

> *Primary sources for the [probabilistic-thinking](../SKILL.md) skill.*

- **Bayes, Thomas.** "An Essay towards solving a Problem in the Doctrine of Chances." *Philosophical Transactions of the Royal Society of London*, Vol. 53 (1763), pp. 370–418. Posthumously published by Richard Price. **The primary source** for Bayes' theorem. https://royalsocietypublishing.org/doi/10.1098/rstl.1763.0053
- **Knight, Frank H.** *Risk, Uncertainty, and Profit*. Houghton Mifflin, 1921. **Primary source** for the risk-vs-uncertainty distinction; ch. VII–VIII especially. Library of Economics and Liberty: https://www.econlib.org/library/Knight/knRUP.html
- **Tetlock, Philip E.** *Expert Political Judgment: How Good Is It? How Can We Know?* Princeton University Press, 2005 (revised 2017). The foundational empirical study on expert prediction accuracy.
- **Tetlock, Philip E. & Gardner, Dan.** *Superforecasting: The Art and Science of Prediction*. Crown, 2015. The synthesis of the Good Judgment Project findings.
- **Mellers, Barbara et al.** "Psychological Strategies for Winning a Geopolitical Forecasting Tournament." *Psychological Science*, Vol. 25, No. 5 (May 2014), pp. 1106–1115. The peer-reviewed evidence on trainability of calibration. https://journals.sagepub.com/doi/10.1177/0956797614524255
- **Kahneman, Daniel & Tversky, Amos.** "On the Psychology of Prediction." *Psychological Review*, Vol. 80, No. 4 (1973), pp. 237–251. The original base-rate-neglect demonstrations.
- **Gigerenzer, Gerd & Hoffrage, Ulrich.** "How to Improve Bayesian Reasoning Without Instruction: Frequency Formats." *Psychological Review*, Vol. 102, No. 4 (1995), pp. 684–704. On why representing probabilities as natural frequencies dramatically improves intuitive Bayesian reasoning.
- IARPA Aggregative Contingent Estimation (ACE) program documentation: https://www.iarpa.gov/research-programs/ace
- **Grace, Katja et al.** "Thousands of AI Authors on the Future of AI." AI Impacts, 2024 (survey of published ML researchers). Documents the wide dispersion and instability of expert AI-timeline forecasts — a contemporary illustration of why AGI timelines are proxy-reference-class / partly-Knightian, not actuarial. https://aiimpacts.org/
- **Metaculus — AI forecasting questions.** https://www.metaculus.com/questions/ . A public, resolution-criteria-based forecasting platform; its AGI/agentic-capability questions demonstrate how vague "when AGI?" claims must be operationalized (outcome + threshold + verifier + deadline) before they can be scored.
- The popular slogan "trust the data" is **not** cited as a source. Data alone, without a base rate and an update rule, is just numbers; the framework is what makes them probabilistic claims.

examples/forecasting-ai-timelines-agentic-reliability-2023-2026.md

# Method in Action: Forecasting AI Timelines and Agentic Reliability (2023–2026)

> *Example for the [probabilistic-thinking](../SKILL.md) skill.*

A worked example on the hardest kind of forecast: one where the reference class is thin, the incentives are loud, and the question is often phrased so vaguely it cannot be scored.

Between 2023 and 2026, "when will we get AGI?" and "when will AI agents be reliable enough to run production workflows unattended?" became boardroom, investor, and policy questions. The public discourse was full of point predictions — "AGI by 2027," "agents will replace X% of jobs" — usually stated with confidence, rarely stated in a form that could ever be checked. Superforecasting-style discipline does not tell you the answer; it tells you how to hold the question so you can be measurably wrong and get better. Walk the Probability Estimate through it.

**Step 1 — Precise question + deadline.** "When AGI?" is unscorable: no agreed definition, no deadline, no resolution criterion. Convert it into something a scorekeeper could adjudicate. Two contrasting examples:
- *Bad (unscorable):* "Will we have AGI soon?"
- *Better (scorable):* "Will a publicly available AI system complete a specified suite of multi-step, real-world office tasks end-to-end at a defined success threshold, verified by an independent benchmark, before 2027-12-31?"

The second version names the outcome, the bar, the verifier, and the deadline. Note that even here the definition (which benchmark, what threshold) does most of the work — an honest forecaster surfaces that the disagreement is often about *definitions*, not *odds*.

**Step 2 — Anchor in a base rate.** There is no clean reference class of prior AGI arrivals, so this is closer to Knightian uncertainty than to an actuarial table. Use *proxy* reference classes and say so:
- Historical accuracy of expert technology-timeline forecasts (generally poor and over-optimistic near hype peaks).
- The base rate of "the last 10% of reliability takes most of the effort" in prior autonomy pushes — self-driving cars are the canonical cautionary case: impressive demos from the mid-2010s did not translate into broad unsupervised deployment on the timelines predicted.
- Benchmark-saturation cadence: through 2023–2025, frontier models saturated many static benchmarks faster than expected, which pushes *some* estimates earlier — a signal in the opposite direction.

The honest output of Step 2 is a range and a caveat, not a point. Naming the reference class is the load-bearing act; a probability without one is undefined (the one-shot probability fallacy).

**Step 3 — Evidence for and against, each tagged.**
- Rapid, repeated capability jumps on reasoning and coding tasks across model generations (2023–2025). ↑ moderate.
- Real, growing enterprise adoption and large capital commitments to AI infrastructure. ↑ weak for *capability*, because spending reflects belief, not demonstrated reliability — a signal about sentiment

examples/tetlock-iarpa-good-judgment-project-2011-2015.md

# Method in Action: Tetlock, IARPA, and the Good Judgment Project (2011–2015)

> *Example for the [probabilistic-thinking](../SKILL.md) skill.*

A worked example. Not pundit theater — peer-reviewed and US-funded.

In 2010, the **Intelligence Advanced Research Projects Activity (IARPA)**, a research arm of the U.S. Office of the Director of National Intelligence, commissioned a four-year, large-scale geopolitical forecasting tournament under its **Aggregative Contingent Estimation (ACE)** program. The objective was concrete: improve how well the U.S. intelligence community can forecast world events. Five university-led research teams competed; the winner was the **Good Judgment Project**, run by **Philip Tetlock** and **Barbara Mellers** at the University of Pennsylvania.

The project recruited several thousand volunteer forecasters and asked them to assign numerical probabilities to hundreds of specific, time-bounded geopolitical questions ("Will Greece exit the eurozone by date X?"; "Will Bashar al-Assad remain in power as of date Y?"). Forecasters could update their estimates as evidence emerged. Each forecast was scored after resolution against a strict accuracy metric (the Brier score).

The findings — published in peer-reviewed papers and synthesized in Tetlock & Gardner's 2015 book *Superforecasting* — were stark:

- **Top 2% of forecasters ("superforecasters")** beat the average forecaster by ~60% on Brier accuracy, and **outperformed intelligence community analysts with classified access by approximately 30%**, using only public information (Tetlock & Mellers, multiple papers; *Superforecasting* ch. 4).
- Superforecasters were not domain experts. They were **ordinary people** — a retired pharmacist, a software engineer, a homemaker — who had been trained on probabilistic reasoning skills.
- **The skills that distinguished them were teachable**: anchoring on base rates, decomposing questions, updating incrementally with new evidence, *avoiding* both over- and under-reaction to news, taking outside views seriously.
- A short training intervention (≈ 60 minutes) on the CHAMP method (Compare, Historical perspective, Average, Mathematics, Probability) **improved accuracy by 10%** in the average forecaster (Mellers et al., *Psychological Science*, 2014).

Walk through the Probability Estimate applied to one Good Judgment question — *"Will the Greek government default on the IMF before 2016-06-30?"* — as superforecasters approached it:

- **Question (Step 1):** Specific, time-bounded, resolution criteria unambiguous. ✓
- **Base rate (Step 2):** Historical Greek default frequencies on IMF over the prior decade; Eurozone sovereign-default frequency. Anchored around a small but non-zero number.
- **Evidence (Step 3):** Bailout-talks status, deposit flight from Greek banks, ECB liquidity stance, German political signals — each a signal in a direction with a strength.
- **Bayesian update (Step 4):** Each piece of news shifted the estimate, but typica

Github ReposUpdated 3d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.

MCPOPENCLAW

Github ReposUpdated 7mo agoRank 70

CopilotKit

The Frontend for Agents & Generative UI. React + Angular