I don’t have issues with people using hotslogs provided they point out the context of what information they’re using with an understanding that hotslog’s information doesn’t have any bearing on what the game proper actually does. Or realize the data skews when looking for more detailed information.
The amount of replays it gets are fine for sample sizes, but the lay person tends to think stats are hardline conclusions, and not facets that reflex change that need questions and analysis for trying to determine why figures go as they do. For pick and success rates, those figures are trends generally backed by people pumping information they saw from hotslogs back into hotslogs. MMR estimates, however, are estimations.
If someone backed a hotslog performance with a similar screencap of the in-game profile, then it gives a reference point of the extent of integrity to the information hotslogs has for appraising a player: if it shows 40 games logged, but the player has 200+ for that mode in-game, then hotslogs is going to skew low because of the lack of information – part of the issue of player trends for hotslogs is it does skew information based on how much it has or on whom a player may be anchored for their relative evaluation — as some mmr shifts are because other players hinged on that account information has gone down. (also, cuz K factor starting points)
Similarly, it does its own seasons or decay which may not reflect Hots mmr evaluations at all. Stormleage has a season shift on hotslogs as is so its already tacking/taking extra points due to its own “reset” of a mode that just started.
I personally don’t upload, so I don’t put much credence on its evaluation of me. However, if anyone looking at my profile sees the extent of information available, they’ll default to assuming the integrity is sound (or sound enough).
However, for the amount of games I have played, the sort of ‘anchor’ I have for hotslogs is impressively low: out of 4050 games shown, i’ve only been shown to have encountered 22 players more than 5 times – phrased differently, the amount of unique accounts I’ve played against is very high.
The effect of that reality isn’t generally going to occur to anyone else just skimming hotslogs, and in my case, because it professes I’m “diamond” I don’t get the “zomg you rank low on hotslogs therefore your opinion doesn’t count” that I used to get a few years back when my own upload exposure was low, so the mmr estimations were also low.
But since my ‘anchor’ is wide and shallow, I get stretches of uncertainty on hotslogs where hundreds of points swing one way or another as the game tries to retro-actively appraise me based on slim references.
Because I’m an example of unapparent uncertainty, I know to check for swings where mmr averages are much wider at one point in time, and then forcibly closer at another cuz that’s how hotslogs works.
Sure, I’ve had occassions where I think someone is ‘bronze’ in game, and find low counts on hotslogs, but I’ve also had the same turn of surprise where I see what I expect to be a one-trick newbie account have twice as many games to their name than I have. So because I have also been surprised to have found player/matches that surprised my expectation, I try to put a bit more stock into trying to figure out why some things come off as big as an outlier as they do.
For some players, they’re content to say 'gee look, game averages mmr" and then try to rationalize personal attacks against their own progress. However, when they do that, they only look at their own win/lose streaks, and they don’t look to see if there are streaks on the other players matched to directly offset their theory. They’re not interesting in seeing anything else but that their impulse is correct.
In that regard, I’m more inclined toward the hammer/screw analogy of the use of tools because doing such leaves unwanted holes
and to reiterate, none of that suggests that blizzard’s mmr is fine/great/perfect/working/whatev.
But that does beg the question if the matching deviates much from what matching can actually do, and not fail to meet the superstitions of an audience that doesn’t care what the system could actually do, and then fault it for their strawmen.
The biggest issue of an mmr system is that it doesn’t know the difference between a hard-fought win, or that a bum climbing to the top of the pile of other bums is still just a bum.
There’s a lot of fantasy to the whole match-making expectations for looking at “skill” and people need to broaden their perspective with other details to realize their glass has the rest of its capacity filled with air, and not ‘empty space’
So sure, I’m not bemoaning the direct use of hotslogs, but I caution people to supplement their claims with other information they can use so that they’re actually looking at more than just confirmation bias and calling it something else.