What Is Wittgenstein's Ruler? (How to Evaluate Information Sources)

Wittgenstein's Ruler is Nassim Taleb's epistemological principle, drawn from Ludwig Wittgenstein, that any measurement simultaneously conveys information about the thing being measured and about the instrument doing the measuring. The less you trust the instrument, the more the measurement tells you about the instrument rather than the thing.

The literal version: you use a ruler to measure a table. If the ruler is well-calibrated, the measurement tells you almost entirely about the table. If the ruler is uncalibrated (it might be stretched, compressed, or poorly made), the measurement tells you less about the table and more about the ruler's specific defects.

The Information Source Application

Applied to information: when you receive an opinion, a forecast, or a recommendation, you are simultaneously receiving information about the content and information about the source's reliability.

A weather forecast from a service with a documented 80% 48-hour accuracy tells you mostly about tomorrow's weather. A weather forecast from an astrologer tells you mostly about the astrologer.

A stock recommendation from an analyst with a documented track record across diverse market conditions tells you mostly about the stock. A stock recommendation from an anonymous forum post tells you mostly about the kind of person who posts stock recommendations on forums.

The Discount Rate for Low-Credibility Sources

The practical implication: before updating your beliefs from any piece of information, estimate the prior credibility of the source. If the prior credibility is low, heavily discount the content — because most of what the "measurement" tells you is about the measuring instrument, not the subject.

This applies especially to: - Financial media commentary (short track records, incentive to be interesting, no accountability) - Anonymous internet recommendations (no skin in the game, no verifiable history) - Expert opinions in domains with poor track records (macroeconomic forecasting, political punditry) - Research published after large search spaces (the data-mining problem)

In each case, the measurement is happening — someone is making a claim about reality. But the calibration of the ruler is poor, so the measurement mostly tells you about the source.

Taleb's practical standard: before integrating information, ask "what do I know about this source's historical accuracy?" If the answer is "not much" or "poor," the content barely updates your prior.

For the full framework, read Living With Randomness.