LogFAQs > #900274079

Now, if I haven't hit you with TLDR territory yet, I'll get into a bit of what I've said in Discord about the seedings.

First, let's get this out of the way. Yes, there is host bias. That's essentially unavoidable. MC_XD's stats show that I'm one of the people at the top of the "voted for winners" list. Some of that is because I have pretty decent mainstream taste where the board is concerned, and some of that is self-fulfilling prophecy. I like to hope it's more of the former. I do, of course, have my foibles. I'll crap on Banjo-Kazooie or Earthbound, and I'm sick to death of Persona 3 and 4 music that doesn't sound like anything special to me. (P5, I will freely admit, is a lot better in my book.)

So how do I do the seeding? I listen to the songs, try to guesstimate what I think of them, with consideration for if my opinion will reflect the board's tastes, and then compare that against the song's past contest history, if any. Usually, that works out pretty well. I'd like to think that, in general, I do a decent job of appraising songs. But I do occasionally miss badly.

When I overseed, it's usually because either I over-hype a song, thinking that if I like it that much, clearly everyone else should like it too, right? I was pretty good about that this year with my pet songs. The most overseeded songs (perhaps) were Jump Up Super Star, which seemed to me like it was popular, and a Cuphead track that I thought might be the latest hotness. I was in love with neither. I also gave Blue Radiance a 1, and put it against Amusement Park. I still think AP is garbage, which says how poorly Blue Radiance performed.

OTOH, when I underseed, it's much more concerning. I can sometimes bite my tongue and accept a song's power, but we get into trouble when I can't see the value in a song at all. (See my previous remark about Amusement Park.) The most famous example of this, as discussed in chat, was 12th Chapter in the Sky from last year, which I assigned a 16 seed and still put up a decent fight. I was absolutely dumbstruck by that match, and couldn't believe it didn't get completely wrecked. If you want an example of that from this year, go back to the beginning of the contest and how I treated Besaid, where I unrepentantly fed it to Laguerra in R1.

And sometimes, when I miss, it's because I misjudge a song on the first listen. I can only put some much time into each of these songs before contest. I have other things to do, and you're all screaming "We want the damn bracket!" so I don't go absolutely nuts here. I also don't appeal for feedback. I did in my first year, where I recruited Ed to help out, and he helped as much as he hurt in terms of fixing bad seedings, with us needlessly arguing back and forth over some stuff.

It's not that I wanted to talk that laboriously about the places where I err, but those are the bumps in the road that stick out to what is otherwise a fine bracket. And when people gripe about positioning in a way that seems to have a legitimate case, I feel guilty, so maybe I had to get some of that out of my system.

So that's enough about my errors. Let's move on.

After I've given the songs a preliminary position, I plunk in the top songs as 1 seeds, and then position the next tier as 2, and do my best to make sure the 1 and 2 -- should they meet, will be an interesting match. Not two songs that sound the same, not two songs from the same company, etc.

Then I grab a cluster of other strong tracks, and drop them into the 3 and 4 positions under the same criteria -- if they should meet 1 vs 4 or 2 vs 3, we want reasonable matches that contrast style. If people have trouble voting, it should be because the tracks are both legitimately good in different ways and the listeners can appreciate that.

5 and 6 take the same tack. I'm a little less worried about what # I give to 7-10, but my R1 pairings are built around this paradigm.
---

Video Game Music Contest 12: Now in progress!