All those reasons you quoted have basically nothing to do with rating. If you lost because you made a mistake, you catch it regardless of rating. Higher win rate against different players also doesn’t have anything to do with their rating. If you keep beating players, you will get a significant contribution to rating. Again, whole point of trueskill is a 50% win rate. And no, the win rate stats on the leaderboard are just terrible data and shouldn’t have been there in the first place because of problems in how game results were recorded on FAF.
I do not understand this mentality at all. Are you saying you just don’t analyze losses as much when a player way worse than you wins? It doesn’t matter that he played better? All that matters is whether he has a higher trueskill?
Whenever I’ve watched replays what rating gives me as a barometer is when to expect the worst mistakes combined with whether I can expect the player to bounce back from said mistake. Plenty of 1800+ players screw up starts, but it’s something like putting 1 or 2 too few engies on energy production. An 800 just forgot to scale power after 5 pgens. Sometimes a 2000 forgets too, but they recover from it using the tools available to them much faster which may or may not enable them to still keep pace in the game depending on the mistakes of the opponent. The fundamental error is still there to see, though.