Nothing new here, just summing up known issues. I ended up writing a much longer comment at File770 than I intended and as it is post-length I'm putting it here also. https://file770.com/barkley-so-glad-you-didnt-ask-80/comment-page-2/#comment-1603312
In that thread, Brian Z was being his usual self. I know he annoys people but he can sometimes be thought-provoking. He was pushing the idea that the Hugo admins decided that Babel was ineligible and then decided to stop counting its EPH points during the elimination process. I don't think that makes much sense because they'd just be plugging the ballots into the software and it isn't likely to have a "turn counting off for this finalist but don't eliminate it" button. But who knows? Maybe, they used software that did have that button! When nothing makes sense, it is hard to call a hypothesis absurd. Maybe "Babel" is the book title equivalent of Little Bobby Tables.
What that got me thinking about is firstly that issue that currently the only viable explanations are a series of coincidences/events or conspiracies or a mixture of both. There are neat general explanations including censorship or incompetence but by themselves they don't explain the specifics. We are unlikely to ever know for sure.
The other thing that I was thinking about was that we do tend to assume a particular order of events here. This is what I liked about Brian's weird theory. It looks backwards because obviously you decide eligibility LAST but that is assuming you are doing the process properly... Anyway, here is a long comment now made even longer by four extra paragraphs explaining the context. The comment was in reply to Bruce Baugh rather than Brian.
The most sensible way to rig the stats is to rig the ballot. The EPH software should be quick to run with clean data, so it is also possible to not just rig the ballot if you control the data but to do so iteratively i.e. put fake data in, see the result, tweak the fake data, put it back in etc. You'd need a bit of time, some familiarity with past results and, of course, some strong motive, poor supervision and a lack of moral integrity…but that's how to do it.
Do the published results look like that? No! They are full of absurdities and errors. So the one thing we can rule out is a competent attempt to change the results of the ballot by somebody with full control of the ballot.
There are clearly layers of issues here and Brian's question highlights something. Multiple different kinds of something must be going on but we don't have a clear idea of the ORDER of those somethings.
Babel is of particular interesting because most of the issues apply here:
- Babel's raw vote number looks inflated as do the other top 7 nominees. This is the Cliff issue that Heather Rose Jones analysed.
- The EPH total points in the first listed round EXCEED the total number of voters given at the top of the page. Marshall has documented similar issues in other categories.
- The ratio of raw votes to EPH points in that first round for Babel and the other top 7 nominees is very high (4+). This shows a kind of average number for how many other nominees still in play the nominee shares a ballot with. By this point, this ratio should be somewhere between 1ish to 2ish. The top 7 are very coordinated ballots like hundreds of voters picked between the same 7 nominees very closely.
- Babel's EPH points never change. This is unique to Babel in this category but occurs for some nominees in other categories. This is impossible as we know some people voted for works shown in the longlist that were then eliminated. The clearest example is The Mountain in the Sea. I've gathered multiple examples of people who had both Mountain and Babel on their ballots which demonstrates the EPH points for Babel cannot be correct.
- Babel is disqualified with no explanation.
As eligibility is supposed to be decided last, we all assume that was the last thing to happen but that assumption rests on the process following procedure. However, we know the process is flawed from point 4. So Brian's speculation that Babel was already disqualified when they started counting isn't totally nuts. We can't trust the assumed order of events BUT if we don't then we have to just give up at this point.
If we do trust the order of events then we could explain 1 and 3 as evidence of ballot stuffing. This would be external to the Hugo committee and not their fault. They aren't technically empowered to say "hey, these ballots look dodgy" and the likely bad PR that would follow would be a disincentive to doing so. There is also nothing anybody could do in such circumstances without clear evidence beyond the improbable numbers.
Point 2 must be an issue from within the Hugo committee. Slates or ballot stuffing or just people voting in weird ways for legitimate reasons shouldn't make the total not add up. That must be an internal error.
Point 4. This is interesting. EPH could have been run properly on a set of ballots where nobody who nominated Mountain also nominated Babel. That is what those numbers imply but that would mean that not only did we have ballot stuffing but also ballot deletion or just fabricated data altogether or in part. Alternatively, it could be just an error in making the table for the report. Maybe they just copied the number from the last round to the earlier round in the table for the report.
Alternatively, there are errors in the names listed in the longlist. Maybe Mountain in the Sea was eliminated much earlier and they just put the wrong name on the published table. If it is some other work then it is less mysterious that no points transferred to Babel.
Either way, this is an issue that MUST be something that the Hugo Committee should have been aware of regardless of its cause.
Point 5. Then, after all that, Babel is disqualified. I think it makes most sense that this happened last and at least somewhat independently of everything else.
[end of comment]
I'm inclined to lump 1,3 & 4 together i.e. EPH was run correctly but on ballots that don't represent how people voted, mainly because Nona9's numbers don't look right either. That doesn't explain 2 or 5 though. Point 2 might be sloppiness or it might indicate some other issue, i.e. an attempt to deal with or correct the dodgy ballots. None of this explains point 5.
I will, will, will change topics tomorrow even if it is just gibberish.
No comments:
Post a Comment