Inferior as a player. Inferior in terms of skill. Not inferior as a person.
They are more comparable, but I don’t see how stratifying the data changes the conclusions.
But there’s not really any evidence to this. It comes back to circular logic. Those players are lower MMR because they’re more casual. How do we know they’re more casual? Because they’re lower MMR. It’s assuming an individual attribute is true of a population, which you can never assume validly.
And there might be. But we have no way of knowing that to be the case. What I’ve been asking for was a metric to define what it means to be casual. No one has provided that. The closest thing we’ve gotten was Sentry stratifying the old data and showing that they’re much closer in MMR, once split by MMR.
It’s quite common to stratify data, but I’ve never once seen a statistician stratifying data, then blanketly draw a conclusion about entire populations based on it.
IF Sentry were to say, “I think the data shows that Terran is the most difficult, but the data seems misleading,” I could just leave it there and agree to disagree. But to first, come up with multiple variables attempting to discredit Terran players as “new” or as “casual” only to immediately disregard said variables as inconsequential is stupid. I don’t think any rational person can disagree with that.
There would also be time spent playing in that MMR bracket, time it takes to move through each bracket range for each race etc, which we don’t currently have. That however assumes the sample-size is even amongst each. It’s important to know when having an equal sample size and a total is important, and when to utilize each to correctly make a conclusion.
As you said, applying an individual attribute to the total population isn’t really something that should be done, but can you not then state that you’ve done that with “more time spent playing” - something in which itself is a flawed metric due to being unable to differentiate which portion of the MMR ranges it applies to, and the fact that you’re also not taking into account that “more time spent playing” can also be flawed based off the general number of players within that bracket.
For example, you’re taking a sample of players in bronze/silver/gold as T/Z/P and the players come out as 300 terrans 250 zerg and 200 toss. Maybe they all spend 20 minutes playing the game, but because there’s 300 terrans, it’s going to look like the Terrans spent longer playing the game simply because there’s more of them in that sample size. So you need to also correct for that.
Again, as I’ve stated in the past, I’m not a stats person, so take what I’ve said with a grain of salt, but it seems like spliting the MMR ranges is extremely important to helping differentiate, exclude, measure or understand several other stats that can then be used to make a conclusion. It’s one thing, but one thing that can be used to aid in understanding many others.
The “activity” metric isn’t measured in time played, but in games played, where it’s been equal to or greater than their representation. So you’re already corrected for that.
If anyone’s stance were: “sure, maybe the 3 metrics we have do seem to indicate that Terran is more difficult, but there needs to be far more statistical proof before I can judge it one way or other.” I would find that a highly reasonable position. Which was what ChatGPT concluded earlier.
But to deny the validity of all 3 metrics, simply because they disagree with your initially hypothesis, is just bollocks. Then to start flinging out insults because I dare to consider the metrics remotely consequential… Well, that’s what I’ve been referring to as deluded.
When you take a subset of the data and average it, if you have a normal distribution, nothing changes.
As is evidenced by it not changing for P/Z, their distribution is more normal than Terran distribution, which indicates of a different kind of strange than the average over the whole data set, where mean averaged MMR is just lower for Terran.
Specifically, it shows that at even medium-high levels, the races are rather equal; which indicates there isn’t a problem, because it means that players who improve don’t get stopped by their race, but by their own skill.
Multiple times people have come into these massive sprawling threads and explicitly said hi i am a casual terran player, which is about as confirmed as you can get?
Furthermore, you’re the one ascribing circular logic here. The actual thing stated by everyone else has been along the lines of a completely different tack that you also randomly take offense to, namely, that Terran is the race that is ‘most familiar’, something owed to the fact that it is the most tutorialised, the race whose campaign is free, the race that has the most standard-to-RTS style of making buildings and units …
And for some reason you kept taking offense to it's the race used most in the tutorials; I can only imagine because it bears similarity to the idea that it's for new players. But I can’t imagine why that would be a bother, because for new players and for advanced players can coexist in a single thing.
suffer (: you now have the vaguest beginnings of how it feels to watch you two run in circles so stupidly
I disagree with that.
Primarily, because I don’t understand what on earth this paragraph is about.
… Then why the hell are you arguing at all? You just want people to say you’re right? A flock of people who mindlessly agree with you to affirm your intelligence?
I say this with full offense meant for once, because multiple people have stated I don't agree that your data indicates that, but a lot of their further statements is the data are bad, ie. there isn’t enough of it or there isn’t enough demonstrating that the data indicate what the argument claims the data indicate.
One guy did that. That’s called anecdotal evidence. I’ve also met a casual low level Zerg player. That’s not evidence.
I have taken zero offense to this.
Nor to this.
I even agree to some extent. The only issue I take with it, as a logical explanation, is that the game is now 12 years old and any new player base is likely long since gone.
Because no one has said that.
THIS is what I take issue with. Particularly when the data was called for by the person I was arguing with…
So you’re telling me that if you take any 500 of the low rank MMR Terrans, any 500 of the low rank Zerg and any 500 of the low rank Protoss, the average number of games between the 3 would show that Terran plays more on average for the low ranked Players?
No, you haven’t. If you have 3400 Terrans in low rank, 3000 Zerg in low rank and 2500 protoss in low ranks, then the average number of games played would automatically favour the terran since there are more of them as it is. Hence my statement “you need to correct for that” - which you’d do by standardizing the amount of players you’re taking a pool from.
His point is that new players - even players coming in now - are going to use a race that is more familiar to them, especially considering mechanical similarity to other RTS and campaign.
A little bit, few enough to, in my estimation, conclude that Terran players are not, in any quantifiable way, significantly “more.” They don’t play many more games. They are a little bit more senior, but not massively so. And that’s been my point from the beginning.
The percentage of games played has always been equal to or greater than the percentage of the population. I seriously don’t understand what’s the argument here. Not being hostile in the slightest, did you ever look at the data I presented way back when? It’s in percentages and they represent slightly more activity than their populations would suggest.
And that’s where we get into my first metric. New players don’t actually pick Terran. The reasons for arriving at that conclusion were logical, but Terran players have, by and large, been more likely to be old players (I can break that down if you’d like).
The number of new players* is a weirdly ever-present number. Just last month I encountered a few Playing Wings of Liberty for the first time streams on twitch.
*However, this also includes anyone making a new account for arbitrary reasons.
Multiple people have said the second half of what you said you wanted to hear. They aren’t saying the first half because they don’t agree with it.
Which, y’know, is reasonable, because the data don’t say what you say they say.
You take issue with the fact that data is not proof, and that someone else reads the same data and comes to a different conclusion than you do?
This is a genuine honest question, no snark intended; I do not understand.
And this is exactly what I was complaining about: Nobody is saying this.
You are ascribing two unrelated things together because if they are together it either demonstrates your point or makes the people you’re talking to look like they have flimsy arguments.
I’m not talking about new players either in that sentence.
Specifically, note the clausing:
“I can only imagine [you take offense to “tutorial race”] because it bears similarity to the idea that it's for new players.” The implication, and thus the thing that’s being bothersome/insulting, in this sentence being that “for new players” and “for REAL players” are mutually exclusive - and my literal next sentence is that that’s just not true.
Please note, that, through all that, I’m not talking about new players, I’m saying that as an example of the only rational thing I can come up with to understand why you complain when we point out that “is the tutorial race” is, directly, a very strong reason why there is such a disproportionate number of bad players who decide to play Terran.
For the record, this is accounted for in the greater than their representation part - supposing the same distribution (34 - 30 - 25), the data in reference shows that there are something like 36 Terran games per 30 Protoss and 25 Zerg games.
However, that data is across the global population and as far as I know we don’t have access to good games played by rank statistics.
Percentage of games that race played divided by population says which race’s players play more games.
Miro’s point is that without the divided by population, the size of the population difference will intrinsically cause the data to fail to be useful.
I believe the data in question does already show it as related to population by default, though.
That the percentage of players who play Terran increases is indicative of either new players picking terran or more protoss/zerg quitting than terrans.
If that’s not the particular data in reference then I do not know what you are invoking.
Additionally, the two clauses here - despite their surface relation - are actually tangential ones.
I’m falling asleep, but the essence is that “random Terran player compared to random nonTerran player” is different than “on average across Terran population compared to on average across nonTerran population” and the distinction is important (though, admittedly, I actually don’t see offhand how the difference would matter).
“Terran players are more likely to have played longer” means the conversion from new to old is higher for them, or means that the turnover rate of the other races is higher. I’m not sure what that indicates at all; though it’s an interesting point…
And, from what I can tell, doesn’t demonstrate anything false about what got said: There must be some reason (or number of reasons) that the population of Terran is so high. The most likely culprit is that, when a new player comes to the game, for a variety of reasons offered up in this thread, they are more likely to pick Terran.
If there is data that directly contradicts that, then the subsequent argument is that the current incoming population is so small relative to the current population that it doesn’t adequately de-salinate the ocean.
Of course, there could also be data that contraindicates that, and then I’d be lost as to a why to that reality is this way.
No. I know very much that the data isn’t proof. You can’t prove difficulty.
What does the data say, then? Because the MMR disparity without a doubt says Terran is at a disadvantage OR Terran players are somehow inferior. Those are the only two possible explanations.
No, he legitimately said that Terran players must be newer players. Then I proved him wrong and he immediately stated that that means nothing.
Kelthar certainly did.
Wasn’t directed towards u.
You can stratify them. It doesn’t matter what League you separate out. I’ve done it many times. The total games played is roughly similar. I don’t think I’ve ever once seen a case where Terran players played less, even slightly.
It would have to be them quitting more than Terran, because the percentage increased while the player base shrunk.
Specifically, the evidence to the contrary is that the difference in MMR almost vanishes if you look at a subset of the data; and that as you go up in leagues, the representation of Terran players decreases.
That implies that it isn’t that Terran is weaker or that Terran players are worse, it’s that of the players who are bad, more of them play Terran.
The distinction is that it’s not that Terran 2000 MMR is equal in skill to 1800 or 2200 MMR, it’s that there’s a huge population overlap between players whose MMR will end up at 2000 and players who play Terran.
Now, let it be said that I also think that this is a wacky point and itself puzzling, because, quite simply, the skill distribution should more closely match to the population distribution.
Yes, it was: Because you said that I was talking about new players: