7̶ɯ0ᴚ9🧵
My first real involvement with the @[email protected] community was when I discovered “Goose Twitter” in connection with the 2021 #goosebracket — when I inexplicably started posting tourney stats.
I repeated that #goosery for the 2022 and 2023 #goose JOTY. And now I’ll do the same for 2024.
In that 2021 tournament, I started an Excel tracker that grew in detail as the rounds continued. After, I went back to 2020 and then used the same template for 2022 and 2023.
So, at this point, I have four years of past longitudinal data that runs through PowerQuery into a consolidated sheet.
As for 2024, let’s start with the basics:
There are 30 unique songs, played across 43 shows in 23 states.
Typical jam platforms have the highest concentration of picks: 5 each for Red Bird, Arrow, Pancakes Drive; 4 for Tumble; and 3 each for Wys, Borne, Madh, Hungersite, and new entrant BMo!.
New York (Cap + FH + SPAC) and Showhio (shocker!) lead in states with 9 and 7 picks, respectively.
Cincy N1 and Atlanta N3 had the most picks, with 3 each.
Few surprises in the most-represented songs (jam platforms get featured in jam bracket). But interesting to note what’s missing.
For the first time, there's no Need, FF, or Flee.
No Arc for the second year in a row. (sigh)
Here's a couple charts showing historical song distribution.
And here’s a year-over-year comparison of song, show, and state distributions.
As playing has grown more consistent, more shows and states have been represented.
Surprisingly, even with the introduction of more songs, distribution has remained relatively flat.
Is there a length bias?
Yes. As always, there's a downward sloping trendline, driven in part by 1-seeds Cap Drive and Fiddler’s Madh both at mid-30'.
I don’t think long songs get ranked higher BECAUSE they’re long; they rank higher because they cover more ground and accomplish more.
There seems to be a reverse recency bias.
This is true for 1-64 and 1-128 this year, and has been consistently true across years.
Recent jams have had less time to be discussed and to percolate in the ether. They tend to get less attention and, therefore, get ranked with less enthusiasm.
Looking at set distribution, there’s a clear downward trend for S1 and a countervailing upward trend for S2.
We all know first sets have gotten more and more songy. The data demonstrate that.
I’ve recently started tracking key slots across sets (O = opener, M = middle, P = penultimate, C = closer, E = encore).
M is obviously the largest contingent because there are 3x M slots vs O, P. or C.
Set openers have become a greater source of big jams; the opposite is true for encores.
As voting begins and rounds progress, I’ll provide more analysis. Send any ideas/questions to @[email protected], Director of Unearned Fluffing, Stats, and Complaints.
In the meanwhile, huge thanks to the AAT crew for facilitating the tourney! I've met so many great folks through past ones.