Joy

That movie actually is the single largest influence as to why I made the jump to data science!

And as it turns out, the voting itself is also remarkably accurate. Shockingly so!

There are exactly 9 players who failed their vote despite being stronger than the median player in the target rank. So if they were aiming for +3, they had a ranking score greater than half of the players in +3. This is only true for *9* players out of 203 failed votes.

I'm sorry to report that most of the plus vote failures are legitimate. It is actually surprising how well the plus server votes are doing to maintain their skill distribution.

Losses bugged out and I have no idea why. Fascinating. I'll have to owe you the losses for now

Tai's model - Wikipedia

Never forget this beauty

Plus servers exist because there isn’t really a better system available to measure skill, but this ends up serving a secondary purpose of giving players a goal, and a tertiary purpose of membership providing social status.

It’s essentially just the best solution in a bad system

It’s all downstream from the choice to model Splatoon’s competitive community around tournaments rather than leagues. There is no way for players to know each other’s skill level if they don’t have structured promotion/relegation mechanics, so they come up with a community approach

Here's wins and losses!

Seems like yours broke a bit because it's mostly tournaments labeled unranked. I owe you the analysis until I can get some more tournaments into this that are labeled "unranked"

The losses!

I’ll get to it when I get back home!

Here you go!

Here's both!

Wins and Losses

Here you go! I'm surprised those old sink or swims are so influential, must've been gigantic wins :D

consider it ground o7

Marvel Rivals had a fad where content creators would just spectate “the world’s top [hero] player” and break down what they did, glazing the shit out of them.

It was extremely popular. Absurdly, even

Shoutout to xenith for having such a lopsided team in LUTI S16 that it significantly improved the Div assignment algorithm

Hopefully I can kill result culture in a few weeks. Everything is looking quite promising so far, I’m testing out Div assignment via my ranking system plus a new seeding system (compared against LUTI S16) and it’s looking really nice so far. Definitely needs more work done but promising, it seems.

Wins and losses! Note that it only uses tournaments labeled as ranked on Sendou, so I might have to go through the tournaments and start manually including some

Seems like only two ranked tournaments are LUTI and Squidfin Fireboard #12. Might have to start including tournaments manually hmm 🤔

Here's your top 5 wins according to the system! (Note that this only includes ranked tournaments!)

I’m pretty sure my ranking system is pretty good now! Comment here if you’d like to get your most influential wins (and optionally your most damaging losses) to see how it’s shaking out. I’ll get to it once I’m out of work :)

After much delay, we are releasing the stat leaderboards from #SplatWorldSeries Finals! Take a look at the Splats per Game Leaderboards #compsplatoon

(All stat leaderboards require a minimum of 8 games played)

Oh I missed the last century bit. Still stands, could probably do this with philosophers like Sartre lol

Almost every academic of note. James Clerk Maxwell is one of the most important figures in all of history and yet I guarantee he is less recognizable than Miku

More cynical is that shorts is considered low signal content while long form YouTube videos are high signal, so diluting the quality of the average video means training new models on this lower quality data would significantly affect performance in a way that allowing AI on shorts wouldn’t.

I think it’s more that shorts is attempting to take eyes from TikTok rather than making money. Allowing monetization of AI content on long form videos is devaluing their primary (highly profitable) platform so they have an incentive to prevent one while allowing the other

My goal here is so that every rank you achieve feels like a *massive* achievement, hence only 8 ranks total.