Here's Umeki's MU chart for Peach. I'm particularly interested in Pika being -2 for Peach. Even ESAM put that MU as even.
That is surprising, and for once I agree with ESAM's appraisal. Though TBH I haven't figured out why; it goes noticeably better and feels easier than it did in Brawl. Anyone
who isn't sick of talking about Pikachu have ideas?
I was thinking about how, man, people really are up in arms over these DLC characters. They want them nerfed, they think they're ruining the game, that they invalidate half the cast, and the fact that top players are switching to them means the end of the world for certain character metas.
Then I remembered that this happens AFTER A BIG TOURNAMENT, EVERY TIME.
Last week everyone was looking at all of those tournaments and being impressed that the spread of top 3 across all regions was impressively diverse. This week because DLC characters did well it's back to "these characters are killing the diversity of the game!"
...at the same time that characters not placing for like 3 weeks means that their meta is FAILING FAST, GOTTA GET THEM ON LIFE SUPPORT.
This is why projects like
Das Koopa
's are so important. The failure mode of results-based thinking is imagining patterns out of noise. It's easy enough to tell a convincing story about how every notable fluctuation represents a drastic shift in the metagame, even though the likihood of observing
some drastic fluctuations in performance at any large event is very high. This is a great way to produce false positives. In fact, if any of you have been following current trends in experimental psychology (and sociology...and medicine...
and...), you'll already know what I'm talking about. This very problem has led to entire bodies of research that accrue seemingly-compelling evidence, only to later
evaporate under slightly different conditions. There
are real and profoundly important trends in the metagame, just as there are real effects out there in (bio/psycho/neuro/socio/etc.)logy. But the noise (unexplained variance in outcomes) is often much greater than the trends themselves. If we take the long view of results, which is admittedly
hard even when you're aware of these problems, then we'll be able to get our rate of false positives down to an acceptable level. Let's do better than most practicing scientists!
As an aside, this is why I wouldn't bother using game counts within individual sets at all (they should be aggregated across several sets if used at all); they're just too prone to being skewed by unusual interactions. We already have enough problems with using individual sets as "data" already.
And also who said Pit's meta is on life support? It's basically the same as it was before genesis.
Earth used Corrin at his last tournament (Sumabato 7, where he drowned in pools). I'm not sure if it's a permanent thing, but he was never exclusively Pit anyways. This could be one of those things that ultimately winds up being noise.
