It looks like it changed in OOTP23 but no one knew it because everyone in the community was answering questions regarding it based on how it had been in OOTP22. To test, select a 1921 game and look at Babe Twombly.
https://www.baseball-reference.com/p...wombba01.shtml
Here are imports of 1921 in various game editions and the Adjust setting at 300 AB on a 3-year recalc. Weaken setting untouched. 1-100 scale, No scouting on.
Here is an OOTP23 thread on A/W:
https://forums.ootpdevelopments.com/...d.php?t=343720
Matt seems to answer the question correctly (in that he cites the way it is from OOTP23 forward). In the thread though, members of the community, unaware of the change describe their experience differently ... based on the many editions prior to OOTP23. The multiple members of the community that answer the question the way the game had been doing it for so long, don't deserve fault. They too continued with their long-standing settings on A/W throughout the OOTP23 and 24 cycles before realizing it changed. I can see, from Matt's perspective, why it is not in the OOTP25 change notes ... because it's been there for a couple years now.
Looking at OOTP23 though I wonder if it may also be the first time that the game double weighted current year stats by default, adding to the available AB's to count. Consider that in a 1921 game, 3-year recalc, Babe T has 358 AB (183 in 1921 and 175 in 1922) and if the current year of 1921 is double weighted in a 3-year recalc, he has 541 AB to consider.
He looks adjusted with a 300 setting in A/W plus 3-year recalc in prior editions. But in OOTP23, 400 AB Adjust won't knock him down. In fact, testing still... a 500 Adjust setting won't knock him down. A 600 Adjust knocks him some but not nearly like OOTP22, so the meaning of Adjust and the meaning of Weaken have probably changed over time too.
So, in looking at this, it seems the outrage is about three game cycles too late.
As for release notes, this was the only thing I saw in the OOTP23 release notes: * Improved historical balancing of ratings on import/recalc. Historical ratings should better adjust to era, and be more consistent between new games and recalc.
Sometimes brevity is good. Sometimes not.
Anyway, if anyone else wants to actually look at past editions and draw a conclusion, please do. There's been way too many posts on all of this without quantitative testing and it was driving me nuts.