2023 Hugos nomination stats part two

[psocoptera]

Ok, some wild speculation regarding the stats. Other people have pointed out some other interesting features, like the ways the raw number of nominating votes goes off a cliff in some categories after the finalists. Dave McCarty, the one non-Chinese official who seems most likely to know what actually happened, has indicated that questions can be emailed to the committee, so, uh, so much for his credibility, but he's apparently willing to go down with this mess.

I guess I'm fascinated by the whole thing because, insofar as it seems like Something Dubious has happened here, it's interesting to me what it is and isn't? Like, if, hypothetically, you had a committee willing to go in full-throttle on falsifying results, I feel like you could do it much *better* than this, just looking at what the numbers looked like from the past couple of years and then cooking yours to resemble them. You could have Babel come in a narrow seventh for the ballot and we all might say "damn, okay, that's super surprising but I guess that's what happened".

Whatever happened here is... not that. You have the things that were brute-force de-eligibilized. You have Best Series nomination counts in the 800s and 900s - Wayward Children got 242 nominations in 2022, and now October Daye got the *least* nominations of the finalists with *816*??

It seems like there must have been either some kind of internal struggle or disagreement, such that the "honest" committee members weren't able to prevail, but are now able to expose what happened, or for the "dishonest" powers involved (whoever they might be) just don't give a shit, or are actively pleased to be able to show their power - because, after all, what's more powerful than knowing you can do as you like and never face any kind of consequences for it.

Comments

[muccamukk]

I just don't get WHY they'd do something dubious. Like... to what end?

I've been tending towards "maybe they're just bad at this?" but they'd have to be *really* bad at it, so it seems very odd.

[psocoptera]

I don't think we're ever going to really know. But I can imagine a situation in which, rather than one overall goal, there were multiple people with the ability to impose demands on these results. Maybe you have one person who says "no, I don't want person X on the final ballot" and another person who says "hey, I do want person Y on the final ballot", for whatever personal or political reason, and if the situation is such that it's generally understood that these are normal requests when getting approval for a public event in China, now there are multiple crisscrossing manipulations we can't untangle.

[cahn]

I feel like you could do it much *better* than this

Yeeeeeah, that's a really good point.

When you said that I immediately thought: but this would be exactly the kind of thing you would expect if an external force (let's call it "the government") made you falsify the results, but you wanted everyone to know that they weren't the right results...

[nenya_kanadka]

This was 100% also my first thought. It's one thing you might do in that situation, especially if you figured that the cooked books might look plausible to anyone but Hugo awards stats nerds, but you also knew that Hugo awards stats nerds are a vibrant and talkative subset of the eventual audience for the document. It might, theoretically, be a way to flag it up for people other than whoever was putting on the squeeze.

[glassonion]

I said a little about this on SFDT just now, but i suspect it's not just one thing, and it's worth keeping in mind that different dubious decisions must have been made at different times --- like, Babel was marked "ineligible" during the nominations phase, whereas presumably "the cliff" was inserted last week when someone was like "okay, this deadline is actually happening and we actually have to release a PDF, let's do this thing."

I guess a thing to be optimistic about is: this is clearly a bad job of cooking the books. So at *some* point, someone will run the actual math, and be able to report on how it was done.

Here's my best personal conspiracy theory guess, we'll use novel as an example:
* Let's say the goal of the book-cooking was to make it appear that more Chinese works were long-listed, than actually happened. And let's say that goal was assigned to the Hugo administrators out of nowhere after the awards happened.
* Let's say that the mechanism by which that goal was achieved, was by adding fake ballots to the nominations list, and then running EPH as usual (because EPH is confusing and hard to fake --- i'll leave it to someone else to report if it's mathematically impossible that EPH was actually run, but for now let's assume it was)
* Some motivating math: In 2022, there were 1151 nominating ballots vs 1867 voting ballots (61.6%). In 2023, there were 1637 nominating ballots vs 1674 voting ballots (97.8%). Assuming the voting number is real and 2022 is representative, that means there are 643 "extra" nominating ballots.
* So each of the extra ballots, was assigned 5 works that actually appeared in the top 7 (they needed 7 because of Babel), and 1 of the six Chinese works they wanted to promote, in a random order.
* That means that each of the works that really was in the top 7, got 5 / 7 * 643 = 459 fake votes
* And each of the introduced works got 643 / 6 = 107 fake votes
* So that doesn't quite math out (107 fake votes for the introduced works is too high, and 459 fake votes for each of the real-top-7 works is too low), but it's kind of ballpark. You could imagine lots of tweaks around the edges (i think if you have somewhat more fake ballots than i said, but less than 100% chance that each ballot includes an introduced work, that would do it).

My point is: if you assume:
* the real vote had Daughter of Doctor Moreau edging out The Mountain in the Sea, so that bounds the number of introduced votes
* Siren Queen (or whatever clearly-overlooked work you prefer) had 58 nominations
* the goal of the shenanigans was to introduce works to the long list while not making the makeup of the overall long list look totally implausible at a glance

then the mechanism i described seems like it would suffice to explain what happened, and seems like something someone could actually do if they were trying to hit a goal on a deadline. (And maybe they didn't care that the Ns would make it obvious something was wrong, or maybe they were just really really in a hurry and didn't think of it.)

[psocoptera]

That a bunch of this must have happened after the finalists lists were already announced and possibly/probably after the awards themselves were awarded makes *so* much sense, thank you. Like... somehow that hadn't quite clicked for me, but, yeah, of course you get some of this because you've already locked in the results and that limits what you can do with the numbers.