Some statisticians claim that the surest way to increase the overall correctness of the total set of one’s beliefs is: never change that set, except by rejecting a belief when given adequate evidence against it.
Conclusion
The statisticians' claim must be mistaken. i.e. The surest way to increase the overall correctness of the total set of your beliefs is not to stop adding beliefs and discard any belief you have adequate evidence against.
Evidence
If this were the only rule you followed, you'd have fewer and fewer beliefs over time. And we need many beliefs in order to survive.
Evaluate
This seems like one of those Flaw questions where the author's rebuttal is just irrelevant to the original position.
The original position is about . The author rejects this way, but it doesn't seem like he proposes an alternative method that would be a surer way to increase the overall correctness of your beliefs.
Instead, he just talks about how many beliefs one needs to survive, which doesn't seem like it's even on topic. Who was talking survival? The statisticians were only talking about the correctness of one's belief set (it was an abstract mathematical point, not advice for living).
In case their math isn't making sense, suppose you had 10 beliefs, 5 true and 5 false. Right now the overall correctness of your belief set is 50%. You're never adding beliefs, but you can get rid of beliefs. If you find out 2 of your 10 beliefs are false, you get rid of them, and now you have 5 true and 3 false beliefs. The overall correctness of your belief set went from 5/10 to 5/8, or from 50% correct to 62.5% correct.
The more false beliefs you get eliminate, the closer you'll get to 100% correctness for your belief set. And since adding beliefs could potentially lower the correctness of your belief set (if you added a false belief), the statisticians are making a boring abstract argument that by only getting rid of bad apples, the percentage of good apples in your barrel would always be rising.
Goal
Look for an answer pointing out the author's irrelevant rebuttal. He hasn't said anything to dispute the statisticians' claim that this method would be sure to increase the overall correctness of your belief set.
He seem to be concerned that following this method would imperil your chances of survival, but that doesn't go against the truth of the statisticians' claim. They didn't say we should adopt this way of living. If someone died from having too-few beliefs, but those beliefs had a super high rate of overall correctness, then the statisticians' claim would still be true.