In a quantum random-number experiment, why do physicists compress millions of imperfect raw detector bits into a much shorter string before they publish it?

Answer: To squeeze out the bias so the shorter string is much closer to truly uniform

To prove the raw bits were already perfectly uniform — Wrong. Compressing the string proves nothing about whether the raw bits started out uniform; if they had been perfect, there would be little reason to throw so many away. The logic runs the other way: the experiment first certifies that enough genuine entropy is in the raw data, and the extractor then concentrates that entropy into the short public result.

To squeeze out the bias so the shorter string is much closer to truly uniform ✓ — Correct! Raw quantum measurement strings are random but not perfect: detector quirks and bias creep in, so some bits are more predictable than others. A randomness extractor distils a long, slightly biased string into a shorter one that is much closer to uniform. For certification, fewer bits really can mean cleaner bits.

To hide the detectors' flaws from anyone checking the result — Wrong. Hiding flaws would defeat the whole point of a certified randomness experiment, which exists so that outsiders can check how much usable randomness the setup produced. The detectors, the assumptions and the statistics are published openly; the compression step is not a cover-up, it keeps only as much randomness as the evidence can back.