Hello,
one of the things I like to check from time to time are the general statistics available at https://qcsalon.net/en/stats?general
they give some useful information, for me the most interesting about screen reader / operating system / browser usage.
Speaking of which, ever since you have added weekly / monthly / yearly percentages, the screen reader statistics became a little broken, possibly inaccurate as well.

Here is what I mean:
Make sure you are in the screen readers used table, and check the data presented there. The first row is None or unknown, totaling at 3,47% (500,/14.416)This is completely fine. However, if we move to the next row, the first column of the row itself says, 1,63% (130,/7.977), instead of naming the next screen reader in the table.
What is this unnamed entry and why is the first column showing some random percentage?
I believe whatever this is, it also breaks the next row, because now we have Jaws at a weekly percentage of0,02% (1,/5.917) and I highly doubt that only 1 user used Jaws in the last week.
There is such a random percentage column after the Window eyes entry as well, with the data showing 0,02% (3,/14.416)

A side question when talking about this, do you know what is NVDA 2004? As far as I know such a version never existed, NVDA being released around either 2006 or 2005, but I do believe that when using alpha versions of NVDA, the client doesn't detect the version correctly and thinks you are running NVDA 2004. This however used to be the case only with the client version 2 I believe, because the debug file would always say,
"Initializing UniversalSpeech...
NVDA Version: 4.0.0"
This doesn't happen in V3 however, there it only says that it is unable to get the version of NVDA, not sure if that influences statistics in any meaningful way.

Thanks for looking into this.

Hello,

The bug in the table has been fixed. Thank you for signaling.

For NVDA 2004, this is still an enigma for me as well. I assume that it's some test version, possibly non public, or, since NVDA is open source, maybe a custom self-compiled version. In truth, I don't know.

The windows 98 appearing in OS stats is also an enigma for me. I have never tried the playroom client on Windows 98 of course, but as far as I know, it's totally impossible that it could have worked on something before 2000, even version 1 (version 1 for which I had no stats at that time).

Then there is also the case of Firefox 255 as well. For this one I don't know either. Probably a test version, too.

Hello,
yes same, I also wondered about Windows 98 and Nokia browser too since as far as I know the web client doesn't work on that one, assuming of course that Nokia browser here means the old Simbian browser. The Windows 98 entry could perhaps happen if you run the Playroom client and set the compatibility mode for Windows 98, depending on what it does and how the client might think it's running under an older Windows. Of course, no idea why anyone would do that, but that's the only possible reason that comes to my mind.

Thanks for the table fix. So there appear to be 2 entries for Jaws. The first one with a total of 5,96% (858,/14.408) and then, after you pass all Jaws version specific entries, there is another one just labeled Jaws with a total of 34,93% (5.032/14.408) I assume this second one is the correct total number for Jaws, and not sure what this first Jaws entry is, perhaps when no version number is detected for some strange reason.

Yes, the first entry is when the exact version couldn't be detected for whatever reason, while the last is the total of all versions together.