Smaller Number of Members

[countyguy]

I just got back to Project Euler for the first time in awhile.

Among all of the changes in the statistics, there's one that I haven't found any explanation of.

Some years ago, there were more than a million members who'd solved at least one problem, and I'm pretty sure over 90% of that number had solved problem 1.

Now, there are 935K members who've solved at least one problem (a drop), and only ~600K who've solved Problem 1 (a much larger drop).

So were hundreds of thousands of members removed?

[euler]

There were a number of changes today to the back end, which although took a considerable amount of work, almost none of it will be noticed. However, one visible change is that we no longer count solves for members who choose to hide their profile (previously this was selecting the private option). We did also remove some accounts that hadn't been accessed in a very long time and hadn't engaged much with the problem set.

[countyguy]

Thank you so much for all of your tireless work!

EDIT (Monday evening 7/27):
I've been doing a lot of Euler data analysis the last few days.

Yesterday, the total number of solves from the solve count was around 10.8 million. Today, it has increased to 13,474,024 (as of earlier today). This is presumably due to the addition of solves from hidden profiles. Percentage-wise, problems with low solve counts had particularly large increases in the reported number of solves. From my understanding, there are at least three types of hidden accounts:

1. People who are voluntarily hidden/unlisted, of which the largest numerical share consists of people who joined between Summer 2020 and Summer 2023 with low solve counts.
2. Rookie accounts
3. Untrusted accounts

I'm guessing that the large numerical increases in the solve counts of the first ten problems were mostly from group 1, and the large percentage increases in the hardest problems were mostly from group 3.

Here, the total number of solves is shown to be 13,356,073, which is far higher than yesterday's combined solve count but still over 100,000 lower than today's solve count. Does this mean that the "Problems Solved" page includes group 1 (and probably group 2), but not group 3?

If so, my personal preference would be for both the solve counts and "Problems Solved" page to include both groups 1 and 2, but not 3... if people are cheating, they don't need to be included in the counts, but otherwise people shouldn't be excluded just for being new.

Lastly, I'm confused about this page. If I understand correctly, all accounts that have not been active since 2013 are automatically considered rookie accounts and not eligible for being ranked, yet the public list of dodecahedra includes accounts where the last activity is listed as being as many as 15 years ago, which would put it before 2013.

[countyguy]

I know I'm probably being really annoying with all of my questions. I'm really sorry.

[euler]

You're not being annoying at all. It's nice to see that someone is taking an interest in the data I provide, and you've made some good observations. Most of those count mismatches will be resolved as we fine tune the core issues. Regarding the point about the long term absent members, that is a good point and we're currently looking into the best way forward with that.