ChiefPointThief wrote: ↑28 August 2023, 15:47
But none of my tables apply to what you are saying. My combined elo doesn’t reach anywhere near the cap for 2p games.
Because you are playing with players who are less skilled than a skill-cap rating would indicate. (You are less skilled than a skill-cap rating would indicate yourself, so every game you play has somebody with a lower rating than that.)
You know what happens, over a large number of games, when you play with less-skilled players? You get scores under 30 in some of those games. Those scores make it exceedingly difficult to reach that 3.33 mean increase per game. It’s just not happening to any significant degree.
(You can look at my own recent game history for a realistic example of what happens with highly-rated players. My rating is currently a touch over 900. My last 10 games, largely played with a mix of experts and masters, have 6 scores of 30, 2 scores of 29, and 2 scores of 0. I did not deliberately bomb out either of the 0s, so this is simply the kind of distribution of scores that occur when playing in the community. Over those 10 games, my rating change is +2, a mean of 0.2. How am I supposed to make up the other +3.1, if I’m going to add 25% deliberate triple-bombs into the mix? I even lost a point in one of those 29 games, with a partner with a rating of 515.)
ChiefPointThief wrote: ↑28 August 2023, 15:47Also the 25% rate was an example. I actually looked at a players last 50 games who bombed on me on purpose for the sake of this discussion. 9/50. Which is 18%. Still very high bomb rate. Their elo remains the same. Mid 900s elo player. Also looking at the game log I discovered that in some of the games other players (masters) actually bombed on them on purpose. Which proves my point that you will still have people who do this even w/ the current system and it isn’t exclusive to low elo players.
So what? Nobody said that all players who deliberately bomb out a game sometimes will drop below 700. I said that I don’t believe that anybody with a rating over 700 is deliberately scoring 0 in 25% of their games and still increasing their rating (over a decent-sized sample of games without employing other means to cheat or distort the system). You haven’t found anybody who is doing that, and remember that if they are deliberately bombing out 25% of their games, their percentage of 0 scores will be higher than 25% because there will also be games that score 0 through genuine errors, or even other players who deliberately bomb out. That 18% 0-score player hasn’t deliberately scored 0 in even 18% of games, as you point out yourself.
The ELO rating is a measure of observed performance quality (which in turn is the best proxy for skill you will find). It allows us to filter players based on their skill/performance. You come here and complain that this does not allow us to filter out all players who deliberately lose games (because they don’t necessarily have ratings below 700, and some may never have their ratings drop that low). But what you want to filter for is not skill or performance; it’s behaviour. Of course the rating filter isn’t going to solve that problem for you.
If a player actually plays at a rating of 1000, excluding 10% of games in which they choose to misfire thrice, then an observed rating somewhere above 700 when including that 10% of games thrown is still an accurate assessment of their overall performance.
Bad behaviour is harder to measure. It’s a factor in the reputation rating, in that players who are observed by the moderators to behave extremely badly can have their reputations limited, and enough non-private red thumbs relative to green thumbs will also limit reputation, but the general case is that reputation is actually a measure of how reliably players finish games. So the one reliable indicator of bad behaviour is the red thumb you assign when you observe what you consider to be bad behaviour.
So when a partner deliberately bombs out a game with you, give them a red thumb, and then you won’t have to play with them again. Use the tools for the purpose for which they are designed.
And if you want to see a better measure of behaviour put in place, feel free to petition the admins about that, but stop trying to have ELO rating turned into that measure of behaviour. It isn’t a measure of behaviour, and it shouldn’t be.