File 770 has kindly published the compendium of graphs, tables and speculation that you’ve all seen glimpses of. https://file770.com/charting-the-cliff-an-investigation-into-the-2023-hugo-nomination-statistics-by-camestros-felapton-and-heather-rose-jones/
It is a long report but not exhaustive report on the inaccuracies, errors and just plain WTFery of the 2023 nomination statistics. Much of it is built on Heather Rose Jones’s original “cliff” analysis, as well as other issues identified by other people over the past weeks including Marshall Ryan Maresca.
Like I said, despite its length and sections covering every category it is not exhaustive. There is still stuff I could have covered further. Best Fancast? There is something very weird going on there. I only really looked in depth at one of the recommendation lists and not the two others (less overlap with the finalists). I only learned this morning that one of the nominees I thought was a Chinese-language magazine was actually an English language magazine (I thing I corrected all the numbers but if some are out by one, that’s why).
This is a link to the public Google Docs version https://docs.google.com/document/d/e/2PACX-1vSsNSBeLmp6MIuJX3ZEVTlw-Xj2AYygnz71j63aqnMFNvNSPxcFv-vSDgWFM17HFBN4r-Ap-nKGKaso/pub
More importantly, if you want to play along at home, here is the enormous spreadsheet of everything https://docs.google.com/spreadsheets/d/12bqjcuGMnSUqNR62SK2GEMBnfe8WqTUq7MvbpQeDkc0/edit#gid=0 This sheet includes EPH data for all years and categories from 2017 to 2023. The extra good news is that I’ll be able to repeat a lot of these calculations and graphs with any future Hugo data.
28 responses to “Hugo 2023: Charting the Cliff”
You and Heather have done a fantastic job.
LikeLiked by 4 people
This is what integrity looks like. Good work.
LikeLiked by 2 people
Guess what I’ve spent my day doing so far? Thanks, I think. 😉
Looks like you were correct in leaving Relics by K.J. Parker as a Chinese work. The collection Under My Skin from Subterranean Press which has the English version wasn’t released until the end of March 2023. I assume SF World published it in 2022.
Also I nominated Even Though I Knew the End and High Times in the Low Parliament for novella. So there’s some more WTFery since Even Though’s EPH points don’t change when High Times goes.
LikeLiked by 2 people
I’m certain if I repeated what I did with Best Novel with other categories we would find many such cases.
LikeLike
Yup, found another in Related Work… no change for Terry Pratchett when Cosplay is eliminated even though I nominated both.
LikeLiked by 1 person
! I bet there must be hundreds of examples.
LikeLiked by 1 person
And more…
Editor-long, Pomerico doesn’t increase either Sarah Peed or Lindsey Hall.
Pro Artist, no points from Micah Epstein to Tommy Arnold.
Grrr!
LikeLiked by 2 people
In a stunning reversal, I’m pretty sure my Fan Artist ballot was counted completely correctly. All 5 of my nominees are here. 3 on the longlist and 2 finalists. When the first goes, the other four gain .05. When the 2nd goes, 2 of the remaining apparently had more in common than just me and the other gets .08. When the 3rd goes, the remaining finalists both gain more than just my points. Wow!
So it seems Dave’s program can do something correctly, and it looks like my ballot wasn’t completely tossed.
LikeLiked by 3 people
Whew! I’m done! Amazing work! So nice to have all this together. Much appreciation to you, Heather Rose Jones, and all others whose insights contributed.
LikeLiked by 1 person
Fan Artist is the second least weird after Semiprozine (can’t recall what the extra weird thing was) so if anything was going to match your votes its one of those two (or maybe Lodestar)
LikeLike
Yup, that was super satisfying after all the wrong. I can’t tell anything directly from my own semiprozine choices. In Lodestar, very likely correct if not as clearcut as Fan Artist for me.
LikeLiked by 1 person
I read it last night, Cam. Great job of picking out a lot of baffling things.
Math is fun, innit? 🙂
LikeLiked by 3 people
Thank you for all of your hard work putting this together! It’s so great to have all of this data laid out clearly and thoughtfully in the same place. I especially appreciate you taking the time to point out that the impacts of these anomalies were vastly different for English-language and Chinese-language works – I think that’s a critically important piece that tended to get lost in the initial “ineligibility” kerfluffle.
LikeLiked by 1 person
One area to explore is that “Chinese” v “English” is too crude a category. There are definitely nominees that don’t fit neatly into those groups but also, it may be some subset of Chinese nominees or specific categories impacted.
LikeLiked by 2 people
One thing I wondered about is: Are all the works we label “Chinese” in the same dialect, or is there a split between Cantonese and Mandarin that may affect groupings?
LikeLike
My understanding is that anything that goes through a mainstream publication route in mainland China is written and published in Modern Standard Mandarin; even those who speak other dialects at home are educated in Mandarin and read and write with the set of simplified characters that were reformed in the 20th century.
Writing in Hong Kong, Macau, and Taiwan uses a non-simplified set of characters that represent spoken Cantonese rather than Mandarin, but if any of those made it onto the longlist, they still would have needed to have been translated into simplified Chinese for mainland-educated readers first.
LikeLike
I think (but don’t know) that they were all in simplified Chinese characters, as would be typical on the mainland, because the rec lists and marketing of the Hugos was to mainland China rather than Hong Kong
LikeLiked by 1 person
Super report. As for “play along at home” — it’s no joke, there are people who are doing just that!
LikeLiked by 3 people
A few comments on typos: You fixed Cora’s last name in the actual references, but not in the “Further Reading” section.
Also, the references in that section are just [(LASTNAME, YEAR)], when the actual references have more information [LASTNAME. (YEAR, MONTHNAME DAYOFMONTH)]. In other scholarly works, I’ve seen a lowercase letter appended to the year when there are multiple publications by the same author in the same year.
The references look a bit inconsistent — some have MONTHNUM instead on MONTHNAME. I personally would prefer YYYY-MM-DD, but that’s just me.
The Validation spreadsheet is listed as “unknown”, but you might consider adding “leaked by Hugo Admin D. Lacey, shared by J. Sanford”, or something like that, to state what provenance is actually known.
I’m a little unsure about how the File 770 link ought to be dated. The tag currently has 18 pages of results, going back to 2022, when the first announcements about Chengdu were made, and will no doubt continue to grow as further posts on the topic will be made. Maybe (2022-2024, as of 2024-02-24), or something like that?
In the footnotes, I would suggest changing [10] to “Tor.com is now called ‘Reactor Magazine’ (reactormag.com)”
Footnote [11] has “Is is” for “It is”. Or it might be clearer as “The misspelled name is”. Also in [11], I note that “mispelled” is misspelled.
This isn’t a comprehensive list; I was just looking at the last pages of the document. As one does.
LikeLiked by 1 person
Thanks 🙂
LikeLike
Thank you for doing this.
LikeLiked by 1 person
Thank you for this report! Really interesting.
I keep getting sidetracked, and thus haven’t finished reading the whole thing, but I wanted to ask about one nitpicky thing that I think may be an off-by-1 typo (or may be my misunderstanding), under “3. Unusual EPH Ratios”:
“
For example, in Best Novel the book Legends & Lattes had a total vote of 831 and in Round 9, 202.35 points. 831 divided by 202.35 comes to 4.10674574 or approximately 4.
In other words, in general, people who had voted for Legends & Lattes had also voted for 4 other nominees that were still in play by round 9. This number is not typical.
”
I hadn’t thought about EPH ratios before—neat approach! But should that second paragraph say “had also voted for *3* other nominees that were still in play” rather than “also voted for *4* other”?
It looks to me like if the number is 4, that means 4 total nominees still in play, which is to say L&L plus 3 others.
But I may be misunderstanding.
LikeLike
You are right – and to be honest I thought I’d corrected that
LikeLiked by 1 person
Oh! I see now that you did correct that in the Google Docs version; I’m reading the File 770 PDF version, but I should’ve checked Google Docs before commenting here.
LikeLike
I only just corrected it in the Google Docs version 🙂
I’ll keep updating edits and when it looks like there aren’t anymore I’ll make a new PDF
LikeLike
I’ve now finished reading the report. Really great work.
I have a couple of more-substantive thoughts that I’ll try to post soon, but for now just two more minor possible-corrections:
* In “11.1 EPH point changes”, it says: “This contrasts sharply with Long Form where there is also a split but not one that is much stranger..”—is that “not” intentional?
* In “Further Reading”, “Bulhert” should be “Buhlert” (3x). (This has been corrected in the Google Doc in the “References” section, but the typos are still there in “Further Reading”.)
LikeLiked by 1 person
Thanks
LikeLike
[…] to take a break from writing about the Hugo Awards after the time spent on the enormous mess of the 2023 Hugo Award statistics. I believe that the many fans who came together to pull apart what went wrong with the 2023 awards […]
LikeLike