"Most frequent ending numbers" is misleading

[NilsEnevoldsen]

I was shocked that 91 was not among the most frequent "ending" numbers listed on https://isthisprime.com/game/record.php, and I now see that it's because this is a table of raw frequencies irrespective of the frequencies with which these numbers appeared. I think this table is of limited interest without also seeing that base rate.

I think a more interesting table would be the numbers under 100 (or some other threshold) most frequently missed when they appear.

Based on my analysis of the data, it appears four numbers under 100 are substantially harder than all others: 87, 51, 91, and 57. (In fact, after removing obvious data errors, these four appear to be harder than basically any other numbers, presumably because only experts encounter very high numbers.)