Trying to think of a pervasive and ever present inflationary mechanism... it's not easy.
The only thing that comes to mind is children. If you have 1000 typical children beginners, then most of them will be really really bad at chess, and not improve much. A few will stick with it, and improve quickly and become very good, but most will be terrible and quit... this is a potential inflationary effect. If you play 10 games, lose 10 games because you're a terrible kid, and then get depressed and quit, you inject rating points... and maybe we could argue that the largest subgroup of any 1000 youth beginners is this type, and as the lower ratings becomes inflated, it slowly influences everyone up to the top players.
---
However, anecdotally this is not the case. Old players often talk about how, e.g., 1300 these days is much better than 1300 was 40 years ago.
So it's interesting to consider that perhaps deflation happens for the majority of players, and then due to top players largely isolating themselves in invite-only tournaments, they somehow create localized inflation for themselves. I vaguely recall some sore loser 2700 wish-I-were-a-top-10-player making this whiney argument...
... which doesn't make sense, or at the very least brings me back to my original point. What is the proposed inflationary mechanism exactly?
https://www.chess.com/news/view/jan-krzysztof-duda-wins-2021-fide-world-cup
In this recent Chess.com article about Duda winning the World Cup, it's pointed out that this is the first time since October 2013 that no player other than Carlsen was 2800. That made me think about the debate about rating inflation, as I hadn't thought about it in a while but was always doubtful about it and have been for a long time. Over the years, the ratings played out as I would expect for a non-inflating rating system to play out, but this striking moment of no more than one player at 2800 brought it back and made me realize how little has changed since I last had any significant musings on the issue, which is probably a few years ago by now.
Although of course this situation of only one player being 2800 is expected to be anomalous, it is interesting that after years, instead of seeing an instance of dramatic inflation, you end up seeing, of all things, something that psychologically looks more like deflation than anything (though in reality, probably there is neither inflation nor deflation). It's like the ratings were so normal for so long, that the only thing that reminded me of this debate was seeing ratings that resembled inflation even less than they normally do; one might have formerly speculated that one day we would start seeing the inflation storm to end the inflation drought going on for years, yet of all things we actually get a pseudo-deflation storm instead of that.
Years have gone by, and we even see another moment where no other player than Carlsen is 2800. As much as people like to assume otherwise, rating inflation now really doesn't appear to exist, unless all the top players have gotten so much worse at chess that they are counteracting it or something
Some other players come and go with the 2800 rating, but the constant seems to be things reliably settling down before people get used to the "new" 2800 player. I guess one can argue that the elite players play each other a lot in elite closed tournaments, and so the points just get traded within a relatively small amount of players to keep their ratings relatively stagnant, but the inflationary pressure, if a significant amount of it exists, would still need to hit soon enough somewhere and somehow, since eventually, huge amounts of former 2600 players would get into the 2700s alongside the current 2700s just through inflation alone, and you don't see that either.
It's like you really have to try to find a pattern to try to convince yourself that significant rating inflation exists, such that a 2800 in 2021 is much weaker than they would be in, say, 2000 or what have you, because a 2800 of the past would supposedly be more like a 2900 today or something.
I know that quite a few people seemed to believe in rating inflation, almost making it seem like some kind of default view, but I haven't looked into the state of things in a while. Are people as confident in rating inflation as they were? And if so, I am curious what keeps that belief alive. It is hard to see an inflating trend, and it looks as much like a deflating trend half the time, but maybe there is something beneath the surface that I am sorely missing. Anyway, an update on the state of opinions on this would be interesting and appreciated
In theory I have seen it brought up that new players who join FIDE and then quickly drop out might somehow "take away" or "add" points to the system, which is interesting to think about, even though it doesn't seem to play out in practice. Like maybe if a person with a provisional FIDE rating is actually underrated compared to actual playing ability and then as such "gives points" to some other player they lose to, and then stops playing chess tournaments before they could "even things out" by playing some more games, maybe points could be thought of as being added to the system of the rest of the players without coinciding with an actual increase in strength of the field, and maybe this process repeating enough over time could result in inflation/deflation. It was something like that, but I might be forgetting the finer points of the argument, but in any case, it seems like a new player could just as easily do the opposite, and win some games as an overrated player, "take away" points from the players beaten, and then drop out of the system before playing more games to even things out. So it seems reasonable to assume, without good evidence otherwise, that the adding and taking away of points like this would be equally likely and thus would cancel each other out in practice.
How the rating system works and reacts to different things is complicated, and I certainly have an incomplete understanding of it, but even in theory, just looking at it generally and intuitively, it just seems weird to think that statistical oddities could ever stack up so much as to taint an entire system like that. It sometimes seems like wondering what would happen to the rating system if a bunch of people just got superpowers every now and then and beat everybody they played for months and then quit chess -- or what if everyone takes money out of the bank at once -- these kinds of ideas seem so removed from how things play out and function on the statistical level.
In general there are a lot of weird scenarios that can play out every now and then (but by definition, only occurring the vast minority of the time) but if any decent amount of those things had any decent chance of affecting things, we would probably have a super chaotic, unpredictable rating system that would make any random event you learn about have you wondering about the effects on the rating system.
But no, I don't have it all worked out into some kind of official proof, but those who have good insight into the rating system relevant to inflation/deflation are free to chime in and enlighten us. In practice there doesn't seem to be significant inflation or deflation in the rating system.