I find it really very frustrating that the author talks about the "95%" percentile and then does very little to actually justify why he's talking about the 95% percentile and how he's measured it. He's got some anecdata about what he views in overwatch, but you actually need some statistical techniques if you want to justify a statistical number. Of the players in the top 5% how many times was it observed that they committed obvious game losing gaffes? How did he screen the reported rankings for the arbitrary subset of players that he considered real competitors.
I don't think the person who wrote this understands statistics because they're not playing a game where randomness is a large influence in the game. If you talk to top poker players about their game they're not going to talk to you about individual mistakes, they're going to talk to you about constructing ranges. A top 5% poker player isn't just going to beat a top 10% player in a single game or single hand, they're going to have a higher expected return over a long period of time. THe point is to come up with strategies that are successful against a large range of opponents. So whilst you can look at a single event and identify whether it was effective in that particular situation, what you're optimizing for is to get the best return over time. So to take the Overwatch example - whilst you may constantly see game losing mistakes at that level, you won't see the same people making the same mistakes constantly.
So basically - for an article that constantly talks about "95 percentile" or "99 percentile" or "moving from 10%-ile to 40%-ile". I just don't even comprehend how you can title something "95%-ile isn't that good" and then not make any effort to actually evaluate how you're measuring 95%-ile. Which, given this is an article about how people generally do things very badly for stupid reasons, I think is rather ironic.
Overwatch has a competitive ladder which ranks players on a ELO-based rating system. The percentiles were published, by the game developers, at one point and are assumed to be stable.
I play(ed) a lot of Overwatch, enough to get into the top 1% (Grandmaster). The rank he derides on, 90-95%, is Diamond, which is quite famously "ELO Hell", because that is where a lot of naturally talented players end up before deciding that blaming their teammates was the issue instead of reviewing their own gameplay. What irks me is, the top 4% of that game, Masters, I would actually consider "good". Sure there is a gap between the top 4%, top 1% and top .1%, but the mistakes you see between those groups are less about "missing fundamental gameplay mechanics" and more flawless execution, awareness and inference. (For anyone that plays the game I would consider 4250 the top .1%, which is almost entirely pros and very talented streamers).
In other words, while I agree the top 10% of the community isn't all that good, the top 5% of that very game is respectable. My beef is less with the ranking and more with his "definition" of good.
Many of the top 1% of Overwatch do regular scrims in semi-pro leagues. Comparing this to basketball, does this mean the bar for "good" is the NCAA, d-league and NBA?
There seems to be something fundamentally wrong (with the game or with the measurement or with something else) when the top 10% is considered “not good” and the top 5% is “respectable”. I can’t really think of anything where this is true. Games, job performance, educational testing, income & wealth... Top 10% is kind of by definition great!
The game's distribution is a normal distribution - and I think the problem may be with perception. What is the definition of "good"? Personally, I wouldn't claim to be very good at the game despite my past ranking - there are a number of players who would wipe the floor with me where it feels like I'm giving my all and they are barely even trying. At the same time I could probably do the same thing if I dropped down on the ranking. If you watch streams, you commonly hear top players complaining about having to play with "braindead masters players" who are the top 5%. All of this shapes the perception of what good means.
I agree that the author isn't totally rigorous in his methodology, which I understand is frustrating, but I think you're missing the bigger point he's trying to make. As I interpreted it, the article is mostly about self-improvement and the fact that even at 95%, there's still a lot of room for optimization between 95 and 99. He's saying, "hey, even this competitive Overwatch streamer makes game-losing mistakes, but it's very easy to improve on these things, if you have the right mindset". In other words, instead of focusing on your rank, you should be focusing on the low-hanging fruit, no matter where you land on the spectrum of talent.
I don't think the person who wrote this understands statistics because they're not playing a game where randomness is a large influence in the game. If you talk to top poker players about their game they're not going to talk to you about individual mistakes, they're going to talk to you about constructing ranges. A top 5% poker player isn't just going to beat a top 10% player in a single game or single hand, they're going to have a higher expected return over a long period of time. THe point is to come up with strategies that are successful against a large range of opponents. So whilst you can look at a single event and identify whether it was effective in that particular situation, what you're optimizing for is to get the best return over time. So to take the Overwatch example - whilst you may constantly see game losing mistakes at that level, you won't see the same people making the same mistakes constantly.
So basically - for an article that constantly talks about "95 percentile" or "99 percentile" or "moving from 10%-ile to 40%-ile". I just don't even comprehend how you can title something "95%-ile isn't that good" and then not make any effort to actually evaluate how you're measuring 95%-ile. Which, given this is an article about how people generally do things very badly for stupid reasons, I think is rather ironic.