This is definitely an issue that’s gone on for too long, but I think the biggest issue here is a lack of people who care about the stats which go into a tier. I’m one of the few people who actually goes through the raw stats pages willingly, because I genuinely care about trying to work with good data about a tier. However, the lack of any proper data community behind Smogon means that there will always be some level of ostracism towards people who just want to display things accurately with math. I can’t tell you how many times I come up with figures only to be met with an “okay” for a response and for three or four people tops to ever give a darn about what I publish, even if it takes me dozens of hours to make. Some spreadsheets I’ve worked on have taken over a hundred hours and gone to complete waste simply because there was no one who cared about what was in them. It seems like the average person only cares about the fancy visuals and big bolded numbers in formats so compact that they inherently dilute the quality of any data portrayed.
Ah yeah definitely felt. It really does feel like the only way to make interesting technical things related to data stick around is if they are operational -- i.e. they are easy to use in some kind of tool, which is really hard when the weighted stats are difficult to interpret. Could you link some of the things you have done?
i dont think usage stats should be used to determine tiers. i think viability rankings are much more preferable. the way that the rby tiering works is a perfect example and i think a version of what they do should be used in current gen.
Imo modern gen should use usage based tiering, while previous gens should use viability based tiering. After watching gen 1 go about it, it just makes sense imo
Oh my god, *that's* why it says provisional. I was over here like how is it still provisional after 10 years.
Anyway very interesting video.
Hahaha yep, it'll usually say provisional explicitly on showdown if your deviation is over 100.
This is definitely an issue that’s gone on for too long, but I think the biggest issue here is a lack of people who care about the stats which go into a tier.
I’m one of the few people who actually goes through the raw stats pages willingly, because I genuinely care about trying to work with good data about a tier. However, the lack of any proper data community behind Smogon means that there will always be some level of ostracism towards people who just want to display things accurately with math. I can’t tell you how many times I come up with figures only to be met with an “okay” for a response and for three or four people tops to ever give a darn about what I publish, even if it takes me dozens of hours to make.
Some spreadsheets I’ve worked on have taken over a hundred hours and gone to complete waste simply because there was no one who cared about what was in them. It seems like the average person only cares about the fancy visuals and big bolded numbers in formats so compact that they inherently dilute the quality of any data portrayed.
Ack, that's really sad, that sounds really frustrating ^^' I'd love to see some of them, if you'd like to share somehow!
Ah yeah definitely felt. It really does feel like the only way to make interesting technical things related to data stick around is if they are operational -- i.e. they are easy to use in some kind of tool, which is really hard when the weighted stats are difficult to interpret. Could you link some of the things you have done?
great video!
i dont think usage stats should be used to determine tiers. i think viability rankings are much more preferable. the way that the rby tiering works is a perfect example and i think a version of what they do should be used in current gen.
It depends how mature the meta is. Viability rankings early in a games life cycle are fairly subjective.
Imo modern gen should use usage based tiering, while previous gens should use viability based tiering. After watching gen 1 go about it, it just makes sense imo