Only on zerosum game (which WT is not) and only of they are named.teams.like.football clubs were the entire team gets an ELO and not the single players. Otherwise there are mathematical problems with the system and it does not work. It is still used by video games, but that is more Entertainment than anything serious.
Just look at the math and it kinda becomes.obvious that it does not Work.
Your conversation started about rewards and your claim that the gamemode are equally hard compared to eachother since the measure of difficulty being your opponents and with random opponents and the same skill distribution the difficulty is the same.
This is a more technically correct description of what you said.
But you made an assumption:
This is only true if the opponents skill is truly random. Which in SB is not the case since there is a preselection on who can enter.
So you assumption was incorrect. This started your entire argument that spiraled to this point. And it only spiraled because you made an incorrect assumption at the beginning.
You can just treat it as if you’re playing an opponent with the average ELO of the enemy team, and adjust your personal one accordingly on win or loss. Really not that hard.
If you’re simply arguing that this is not “100% identical to classical ELO Actshually!” because it has one word of logic different, I really really don’t care, and it’s not at all important for the conversation. Call it “Shm-ELO” if you want, instead. Whatever.
The point was that there is obviously an option to have some sort of ELO or ELO-like system for SBMM that prevents seal clubbing, and that War Thunder doesn’t use one. So SB is thus basically seal clubber heaven. So there was no point in bringing up seal clubbing, as it’s standard in all modes here. If anything, SB most of all if it has the highest skill ceiling.
gamemode are equally hard compared to eachother
Yes. And they are. A seal clubber heaven isn’t “easier” on average. It’s easy for the seal clubber, but it’s harder for the baby seals. Averaged together, everyone in the mode, the typical difficulty experienced will be the same as any other mode.
So they should not get any different crew skill points rewarded to them.
This is only true if the opponents skill is truly random.
This actually has zero effect on the fact that “all modes are equal difficulty”, and no I’m not assuming it at all when I say they are all equal difficulty. A 500 skill point person up against a 500 skill point person has just as difficult of a time as a 3 skill point person up against a 3 skill point person. So any sort of gating or change in baseline is not important.
No i am saying that the statistical estimator that is the basis of elo does not work if you use it this way.
Making the values wrong.
And i repeat WT is not a zero sum game. There is math on why elo works and it requires a zerosum game.
But experience does not equal skill, there are other factors biological factors. Those are distributed in a normal distribution, and there is limit you need to achieve on those to play sim, that doesn’t exist innthe other two modes. This increases the average skill level in SB compared to AB and RB. That’s the part you ignore. Experience isn’t everything.
But here is the thing and why randomness is important here. Skill is distributed like inteligence just use the iq scale. And distribution is important here.
Yes if two player of equal skill meet eachother the outcome is 50-50 in every mode. But the distinction is the likelihood to meet someone better or worse than you, that is what actually makes a mode difficult. And this is where random chance comes into play.
So lets use the iq point scale as it is simple to understand. So the “average” iq is 100 so thrown in a random match the expected skill level of a player is 100. average in quotation marks as it isn’t a mathematical term and can mean a lot of things, but since we are talking about a normal.distribution and all possible meanings of average would actually be the same value due to the symmetric distribution it is fine to use that word.
So here is the issue, if you bar players at a skill level of around 80 from playing (which SB air indirectly does) the chances change. Now it is no longer a normal distribution but a skewed one. The “average skill level” which in this case has to be the median skill level would now be higher than 100. Which means that the expected skill level is higher than 100.
Meaning that it is more likely to get better players as opponents compared to AB and RB, therefore making SB the more difficult mode.
This is the wrong assumption you made that the expected skill level of a random player in SB is the same as in RB and AB, which is mathematical nonsense, since there is a lower bound on skill in SB that doesn’t exist in the other two modes.
Both these topics are pure math topics, you should treat them as such.
No i am saying that the statistical estimator that is the basis of elo does not work if you use it this way.
It works just fine, I’ve used it before myself (didn’t design it but worked with the data on a game that did basically that since before i worked there). I am happy to agree it’s not formally actually “ELO”, like I said call it “Shm-ELO” if you want, but it works perfectly fine for team SBMM.
I didn’t disagree (nor agree, to be clear), I said that’s irrelevant, and explained why already. Scroll up. You’re spending all your time here defending against me saying this is wrong, which I didn’t, so… skip.
And distribution is important here.
No, it isn’t. You could have a flat distribution, a normal distribution, a chopped off normal distribution, a skew distribution, a distribution that looks like a (monotonic version of) Donald Duck, it makes no difference. Obviously since half the players win and half the players lose every match, the average win rate in the mode is guaranteed 50% no matter what, and so the average difficulty is the same as AB mode, as the same thing applies there and it’s also 50% average exactly.
Nope, that would only be true for zero sum games. WT isn’t one. There are no draws in WT if teams get to 0 at the same time both lose.
Therefore average winrate is below 50% over all players and games. Less than 50% of all games are won.
No it just doesn’t. Itay give you a value but this value is not a representation of your skill in the same way elo is. Arguably it isn’t at all, since criterias need to be met, which are not.
There are no draws in WT if teams get to 0 at the same time both lose.
lmao, this whole time you’re talking about freaking draws? I’ve literally I think had one single draw that I can remember in like 5 years of playing war thunder.
Winning against a team of higher average skill requires more skill than winning against a team of lower average skill. And, accordingly you get more points for the former than the latter. So yes, it absolutely reflects your skill, and it works perfectly fine at preventing seal clubbing as a result.
Does it reflect your skill in precisely the same mathematical way, point by point, as formal ELO does? No.
Does it reflect your skill? Yes.
Does it work for SBMM and for preventing seal clubbing? Yes, very well.
ELO works wonderfully on -0.001 sum games, just like on zero sum games. Absolute clownery to suggest otherwise. This is like saying that because the ACT had one ambiguous question on it, we should throw out the entire thing and not use it for college admissions for that entire cohort of students that whole year, lmao
You’re literally just trolling at this point and wasting people’s time.
ELO is more about calculating the win chances. That’s the main purpose.
Well no, not really. Otherwise the first question would be a yes. It either can give you a proper number or it can’t. It can’t be a rough estimate since that is all that ELO is already, an estimator.
Prove it.
It’s math so there is proof. A mathematical Proof that it works. No added stipulations like new players coming in.
I mean you can easily see that it does not meet the requirements.in the most simple setup if only two players existing. You will end up with no points being left.
Mathematically it doesn’t work.
Now add all the other problems:
-it’s a 3hour match, with players joining and leaving all the time. Should the guy plaing 20mins be as much punished on a loss than the one playing the full 3 hours?
-are the “averages” time weighted for each player?
-what about uneven team sizes? 6v4 what now? Lanchesters law suggests numbers are a big advantage, taking the average of all players no longer makes sene.
-what about the situation where the first hour it’s a 6v8 then 5 players join on one side and the second hour is now 11v8. Then 4 players leave with the last hour 11 v 4. You do realize thus is constantly in flux?
You are acting like the teams are consistent of the same players from start to finish, which they are not. It is a lobby system.
That is an oxymoron. It cannot possibly work for the SB Air matchmaker… Since there isn’t a matchmaker… It’s a lobby system.
There are other factors making this dumb, like sealclubbing actually being useful for the seals. You learn faster on good opponents than bad ones… Creating a higher spread of skill levels.
Well no, not really. Otherwise the first question would be a yes.
ELO is not the One Truth passed down by the gods themselves through their angelic messengers to mortals, lol. You sound like some sort of televangelist.
It’s just an algorithm some dude came up with awhile back as ONE WAY of trying to quantify skill in PVP.
It has all kinds of imperfections, most of all due to the fact that real people do not play an infinite number of games against an infinitely fine and even gradient of opponents. They have all kinds of wonky lumpy game histories that don’t measure at all perfectly, and often have huge systematic patterns of bias in who you play.
But also as another deeper example problem: because skill is not uni-dimensional, it’s a gross but convenient over-simplification of reality. In fact, skill doesn’t even follow rules of MULTI-dimensional coordinate spaces. It can disobey things like triangle similarity. I.e. there can be and very often are “rock paper scissor” type situations between 3 given players, for example. Playing one another repeatedly would never ever cause them to settle into any “correct” ranking between them, they would only fall into an infinite oscillation.
It’s math so there is proof.
ELO is literally not pure math. It’s a hack-y, “good enough” practical method for roughly modeling PSYCHOLOGY, and its usefulness is purely empirical.
As are my statements that similar methods work just fine, also empirical: I work in the gaming industry and almost all games have some variant of ELO, it’s almost never actually literally ELO, and yet they all do very well at making players satisfied with good, challenging games with very near 50% win rates.
The reason WHY it’s not actually perfect ELO is due to quirks of the game (such as being team games, as the example above) that make it simply not fit. So it isn’t even possible to measure “whether ELO would have achieved closer to 50%” because ELO is just un-usable there.
So it’s a worthless question to talk about. And basically just trolling / you’d eventually get fired from those companies if you kept pestering people about it for no practical reason.
I’ll just highlight the critical example here more clearly. Let’s say a game has only 3 players:
Ellen ALWAYS 100% of the time beats Sam when they play
Sam ALWAYS 100% of the time beats Katy when they play
Katy ALWAYS 100% of the time beats Ellen when they play
The reason could be “One always plays a knight civilization in Age of Empires, and another plays a pikeman civilization which hard counters knights, and another plays archers which die to knights but kill pikemen”. Or whatever.
If they all play each other constantly for a couple of weeks, the ELO system will have assigned them like 990, 1,000, and 1,010 ELO or something like that, depending on who was most recent.
So ELO is basically predicting that when Katy and Ellen play next, one of them has a 50.4% chance or something like that. But ELO IS WILDLY WRONG here, because the correct answer is that Katy has a 100% chance to win and always has won vs. Ellen a hundred times already.
This very much happens in real life. It actually happens so much more in some games than it does in chess that they make custom adaptations to ELO to adjust for it and do better than ELO for their game, which may be very prone to rock/paper/scissor situations. Although it does happen in chess, too, it just isn’t worth the time and expense of worrying about it there.
So let me get this straight
Your idea of an example is a scientific impossibility?
100% does not exist in real life due to the inductive way we gather information.
And yes ELO works iterative… No shit.
Nobody said it is perfect. That wasn’t my claim. My claim was that your method does not work, because it doesn’t do what it sets out to do (what elo does).
But lets go into more detail why your method is BS.
You take an average (you don’t even explain what you mean by average, that isn’t a mathematical term, my guess is you mean the mean) of the players in a team.
WT would be similar and as you can see it is very much skewed. Since you use the arithmetic mean on a skewed distribution, there is already a red flag, since the arithmetic mean only works as an estimator for symmetric distributions.
Here the measures are explained and why the arithmetic mean doesn’t fit skewed distributions.
It’s even more skewed if you take into account that ELO does not work linear. So you would treat a 2 player per team game as follows:
Team A consists of a 1000 and a 2000 elo player.
Team B consists of both players being 1500 rating.
Your system would only work if the chance of a 2000 pts player beating a 1000pts player with a chance of 2:1. It’s actually a 99.96% chance for the 2000 point player to win.
1500 beats 1000 with a chance of 94.61%
2000 beats 1500 with a chance of 93.97%
So lets look at it as a Deathmatch dogfight where the opponents split up (which is a fair assumption since leaving on opponent alone is an almost instant fail and has no longer to do with the skill levels also the competent pilot would always engage the split, it’s fighter doctrine anyway) and look at the likelyhood of outcomes.
1000(A) beats 1500(B) and 2000(A) beats 1500(B) - A wins = 5.06%
1500(B) beats 1000(A) and 1500(A) beats 2000(A) - B wins = 5.70%
1000(A) beats 1500(B) and 1500(A) beats 2000(A) after that 1000(A) beats the remaining 1500(B) - A wins = 0.02%
1000(A) beats 1500(B) and 1500(A) beats 2000(A) after that 1500(B) beats the remaining 1000(A) - B WINS = 0.31%
1500(B) beats 1000(A) and 2000(A) beats 1500(B) after that 2000(A) beats the remaining 1500(B) - A wins = 83.54%
1500(B) beats 1000(A) and 2000(A) beats 1500(B) after that 1500(B) beats the remaining 2000(A) - B wins = 5.36%
So we get a chance of A winning to be roughly 88.62%
While your system thinks it’s 50%. It’s not as easy as just taking the arithmetic mean.
To give you an idea, the team rating for Team A in that composition should be around 1900.
So your system fails at what is supposed to do. It does not calculate the win chances properly.
As i said mathematically it doesn’t work, if it did it would be used in football, as it would be more precise than the team ratings.
Lol no, you don’t get to dismiss the entire point you find inconvenient with something as moronic as
as saying it’s “impossible” for one player to always beat another player in a game (Especially since that obviously wasn’t even the point, 99% would have been the same conclusion). If you want to keep discussing, you first address the provided counterexample of ELO failing incredibly hard, directly and in good faith.
Then I can speak to the rest, because “ELO often fails incredibly hard at describing human games” is going to be an important foundational point before that.
