Skill compression describes the relative difference in skill between two players randomly chosen in a given bracket. For example, GM covers about 1% of the playerbase, while Gold covers around 32% IIRC. The players in GM are playing within a much narrow band of skill variation than players in Gold, and they’re relatively more equally skilled then players in Gold. Hence, it leads them to rely more heavily on differences in power discrepancies in heroes–the problem here being that if you push skill compression too far, those power discrepancies can become extremely small yet have large effects on the meta, resulting in apparently blown out viability/power/usage (to apply arbitrary units, a rank with a skill compression factor of 80% might derive viability from a delta down to 15 “viability units,” but a skill compression factor of 99% might derive viability from a delta down to 0.65 “viability units.”)
This leads ranks under heavy skill compression to develop phenomenon not seen in other ranks with lower skill compression, even those which encompass highly skilled players–for example, even in GM, which itself demonstrates skill compression, Sombra was never truly a viable choice, yet she was a strong pick in OWL. Was it because Sombra is actually this super powerful pick? Well… no. Even GM couldn’t make her work effectively. It’s because every team in OWL is so relatively close in skill and exhibits such adequately high teamwork (largely because they train together multiple hours a day) that they were able to exploit the tiny difference in viability to make the strategy and hero work.
But, realistically, nobody on ladder plays with that level of teamwork or under such closely competitive circumstances–no, not even GM. So, if we then balance Sombra based on OWL, she’s going to be underpowered in everywhere but OWL because nobody has the necessary level of teamwork/skill/training to make her feasibly work. This is ultimately the failure of top down balancing when you use a highly compressed reference point–it makes the usability of many heroes not rewarded but contingent on having that level of skill. This is fine for players in top ranks, but for players not in the elite, it creates an extremely unfun and unbalanced experience like what Hajile shows, where “balance” (very loosely using this term because, again, Blizzard doesn’t actually “balance” around top ranks) for the top ranks actually made balance across the board considerably worse–even in “skilled” ranks.
This… is not true. Something OP at low ranks is OP… at low ranks. Just because it’s not also OP at high ranks doesn’t make it not OP in the low ranks… There are mechanics, for example, that can excessively punish errors. At high ranks, this isn’t a big deal, because people are less prone to making errors, so it’s relatively balanced. At low ranks, it’s atrociously oppressive and makes improvement more difficult, because people tend to make a lot of errors, making it very OP in those ranks.
FYI, Diamond covers the 85th through 95th percentile of players. If that “isn’t skilled at all,” then you’re engaging in massive gatekeeping on who is good at the game, and it becomes very difficult to take your opinion seriously. I agree that we shouldn’t balance the game around Bronze or Silver or Gold, but calling objectively skilled players “not skilled at all” in an attempt to justify a bad balance model is dishonest.