# Fibbers' Forecasts Are Worthless: The Role of Source Integrity in Prediction > Coverage of lessw-blog **Published:** February 28, 2026 **Author:** PSEEDR Editorial **Category:** risk **Content tier:** free **Accessible for free:** true **Word count:** 532 **Tags:** Forecasting, Decision Theory, AI Safety, Source Credibility, LessWrong **Canonical URL:** https://pseedr.com/risk/fibbers-forecasts-are-worthless-the-role-of-source-integrity-in-prediction --- A recent LessWrong post challenges the separation of message and messenger, arguing that source credibility is a non-negotiable variable in practical forecasting. In a recent post on LessWrong, the author explores a fundamental heuristic for decision-making: the intrinsic value-or lack thereof-of forecasts provided by untruthful sources. Titled "Fibbers' forecasts are worthless," the piece offers a concise but high-signal argument regarding the separation of abstract logic from practical implementation. This topic is particularly resonant in the current landscape of AI development and systems engineering. As the industry moves toward autonomous agents, synthetic data generation, and automated decision-making pipelines, the ability to evaluate not just the data, but the _source_ of that data, becomes a critical architectural requirement. In classical logic, we are often taught to evaluate an argument on its own merits, independent of the speaker. However, this post argues that in the domain of practical execution and future prediction, ignoring the "metadata" of the source's integrity is a fatal error. The core argument presented by lessw-blog is that while abstract arguments (such as a mathematical proof or a line of code) can be verified independently, practical proposals and forecasts rely heavily on the credibility of the agent making them. If an agent-whether a human founder or an AI model-has a history of untruthfulness, their forecasts are rendered statistically worthless. This is because a forecast is essentially a probability distribution over future events based on the agent's internal model of the world. If that agent is known to distort reality (fibbing), the link between their internal model and their stated forecast is severed. The post draws on the idea that "good ideas do not require lies to gain acceptance." Consequently, the presence of deception in a proposal is not merely a character flaw but a structural indicator that the proposal itself is unsound. For engineers and product leaders, this suggests that "trust" is not just a soft skill or an ethical preference, but a hard dependency for system reliability. If an input source is compromised by a lack of truthfulness, no amount of sophisticated processing can reliably extract a valid forecast from it. For the PSEEDR audience, this highlights a necessary evolution in how evaluation frameworks are designed. It suggests that systems must be capable of tracking source reputation and integrity over time to weigh predictions accurately. A forecast from a "fibber" is not just a low-confidence data point; it is a data point that should likely be discarded entirely to preserve the integrity of the decision-making model. ### Key Takeaways * **Context Matters:** While abstract truths stand alone, practical proposals cannot be decoupled from the credibility of the proposer. * **The Cost of Deception:** Forecasts from untruthful sources lack predictive power because the signal is intentionally distorted. * **Heuristic for Quality:** High-quality ideas and robust projects generally do not require deception to secure support; the presence of lies is a strong negative signal for the project's viability. * **System Design Implication:** AI evaluation frameworks must account for source integrity, treating truthfulness as a functional requirement for reliable forecasting. We recommend reading the full post to understand the nuances of this heuristic and its application to decision theory. [Read the full post on LessWrong](https://www.lesswrong.com/posts/cXDY9XBm5Wxzort29/fibbers-forecasts-are-worthless) ### Key Takeaways * Abstract arguments can be judged independently, but practical forecasts require source verification. * A known history of untruthfulness renders an agent's future predictions statistically worthless. * Good ideas rarely require deception to gain traction. * Evaluating source integrity is critical for robust AI decision-making systems. [Read the original post at lessw-blog](https://www.lesswrong.com/posts/cXDY9XBm5Wxzort29/fibbers-forecasts-are-worthless) --- ## Sources - https://www.lesswrong.com/posts/cXDY9XBm5Wxzort29/fibbers-forecasts-are-worthless