Camestros Felapton

FELAPTONL

Even More Hugo Wisdening

I’ve never been a fan of cricket but my family growing up were and there were numerous copies of Wisden in the house, which for those who don’t know of it is best described here https://en.wikipedia.org/wiki/Wisden_Cricketers%27_Almanack I guess some in the house hoped that I might find it intriguing and I could see the appeal but resisted.

These days we’ve got something better! All the fun of tables of dry numbers PLUS science fiction books! I don’t have a round up of other takes on the numbers yet though.

Normally Brandon Kempner at Chaos Horizon has posted something by now but there’s not been a post there since February. I hope he is OK.

Greg Hullender of Rocket Stack Rank is actually in Helsinki – and having a fun time I hope – so probably won’t post anything yet.

In the comments JJ gave links to three rich sources of data:

The first one is great for seeing EPH in action.

A quick recap. For EPH there are two things in play, a raw total of votes and a weighted total or points. The raw total doesn’t change and is the most important. The points do change. Each voters ballot per category counts as 1 point in total (I think) and those points are divided across each thing voted for. Vote for one thing then that’s one vote and one point for that thing. Vote for two things then that’s one vote each and 0.5 points for each thing. As works get eliminated, the ballots 1 point gets shared between fewer things.

It is narcissistic but I’m going to look at Fanwriter because 1. ego and 2. the Rabids had two nominees. The tables show the top 15, so there will have been some rounds of elimination before this.

tableEPH1.png I’ve screenshot this table I made because WP keeps mangling it. The extra column shows, on average how many points each vote is worth at this point. Note that Jeffro’s are different from Morgan’s, the other Castalia House blogger in the nominations.

I think it is safe to assume Morgan’s 50 votes are shared directly with Jeffro’s. Note also the difference in votes between the two is 30 and the difference in points between the two is 30. I.e. Jeffro’s points formula is 50*0.5 + 30*1. So, at this stage, Morgan appears to be exclusively on ballots with Jeffro and only Jeffro, and the other ballots with Jeffro have nobody else on it.

To see this clearly, I’ve added columns that show how the points change as the ballot continues. The changes are small because it is only points we are looking at.

ephprohr

Note that in nearly all cases, when somebody gets eliminated everybody gets some more points except Morgan and Jeffro until…Morgan is eliminated and only Jeffro gets points. OK, that is not much of a revelation given what we know already, but it is nice to see how it works in reality.

I put totals along the bottom. but I’m not sure how to read them. The total of the points column doesn’t mean much but points total / total votes should give an average sense of how much each vote was worth. However, for subsequent rounds, we don’t know how many ballots are still in play and how many are exhausted.

In terms of EPH, the table is a little hard to follow because the ordering matters with the process. On the PDF it is the name in red that matters. At each stage it is the bottom two ON POINTS who are then paired off ON VOTES (unless it is a tie – in which case points).

You can see this following Alexandra Erin’s line. She is last (or second to last) on points for several rounds but in each case, she has more actual votes than the other nominee. Low points do not mean you get eliminated, it just means that your total votes get compared more often with another candidate.

In the last stages, Natalie Luhrs has fewer points than I do but more votes and then the same with Mark Oshiro. I think this is an aspect of EPH that people find confusing. The points make it seem like some votes are worth more than others but in general raw votes are what matter. It might not quite look that way if you look at me versus Mark Oshiro but that’s misleading because the system never actually pairs me and Mark. However, Jeffro’s (semi) bullet voting works by pushing his points high enough that he didn’t end up in a knockout test with another nominee around 80 votes.

camestrosfelapton

,

20 responses to “Even More Hugo Wisdening”

  1. EPH is what pushed O. Westin, whoever they may be, ahead of me, since lots of people who nominated me also nominated you and Mike and probably Chuck Tingle and Alexandra Erin, while different people nominated O. Westin.

    Like

      • It looks as though O. Westin and their supporters didn’t understand the “Fan Writer” category definition, and possibly thought it was a “Fanfic Writer” category. I suspect if they’d had enough nominations to make the ballot in the category, that they would have been disqualified, since their work is fiction rather than Fan Writing.

        Liked by 1 person

      • Oh, that’s who O. Westin is. I actually follow MicroSFF and like what they’re doing, but I didn’t recognize the name.

        Coincidentally, MicroSFF would have been a great example for the digital Hugo panel as something that doesn’t fit the current Hugo categories.

        Like

        • Yes – Chuck Tingle doesn’t fit well either. I know the pretext was his Puppy tweets but his more general *thing* is both SFnal and interesting and entertaining (not the books per-se). Not saying “Hugo worthy” necessarily but clearly a kind of writing.

          Like

        • Yes, I saw that panel on YouTube, very interesting stuff.

          Each story is eligible in short, so it’s not that it doesn’t fit in a technical sense, it’s just that if flash is highly unlikely to get through, 140 chars doesn’t have a snowballs chance.
          I’m surprised there hasn’t been any pressure from flash writers/venues for their own category. (Or maybe there has and I’ve missed it)

          Like

      • I don’t think there’s any doubt that fanfic writers are eligible for Fan Writer under the letter of the rules, since they are so strikingly vague – they include anyone whose work appears in generally available electronic media.

        (An interesting side-effect of this: under what’s probably the intent of the rule, Mark Oshiro would not be eligible, since his fannish writing is now how he makes his living: and indeed if you follow the informal Hugo guidelines, his work is ruled out because it appears in a professional publication. Following the formal rule, however, he is eligible.)

        I think Tingle fits into fan writer fine – he wasn’t being nominated for his fiction, but for his commentary. Or perhaps for his fiction in so far as it was commentary. He was being nominated for writing about fannish matters, anyway.

        Regarding the idea of a new award, I think we have to ask, as we always do nowadays, whether there would be enough nominations and whether they would be cohesive enough – otherwise it would be too vulnerable to slates.

        Like

  2. About the totals in your last table – on each round, all voters who still have at least one nominee in the race get one point divided between their surviving nominees have been eliminated get no points. So the total in your column 1 shows that, of the original 1178 voters, only 618 still have any of their nominees in the race (there have already been 74 rounds not shown, so rather a lot of voters have not seen any of their nominees show up even on the longlist). Cora Buhlert then gets eliminated, and column 2 shows that the total is now 615 – 3 of Cora’s voters have their points eliminated along with her. Cora had 9.27 points (actually, 9.2666… 0points) from her 25 voters, so your column 2-1 shows how the 6.26 points (actuall, 6.2666…) she had from her other 22 voters are redistributed among the surviving nominees.

    Where a transfer between two nominees is small, you can sometimes infer quite a bit about the votes concerned. For instance, the 0.05 that columns 2-1 shows as having transferred from Cora to O. Westin can only represent a single voter who still had five nominees in the race before Cora’s elimination (at 0.2 to each of their nominees) but now has only four (at 0.25 each).

    Sometimes, this kind of technique works equally well on larger transfers – as you effectively demonstrate above in the Morgan/Jeffro case. And looking through the various categories, this looks very much like just one part of a scenario-testing exercise by VD (or alternatively some very tightly-drawn agreements to disagree with allies). For instance, in Related Work, three-quarters of the RP bloc voted for Ralph McQuarrie and Neil Gaiman and the other quarter just for McQuarrie – with just two of McQuarrie’s votes either being non-RP or extremely indisciplined RP (for instance, calculations suggest that one of these two voted for Gaiman and the other didn’t, but that both of them voted, among others, for LeGuin).

    Like

  3. I love this stuff! I especially like the decisions report. Interesting to see what gives people trouble when nominating. Would be good to consider for tweaking Hugo categories and their descriptions.

    Like

  4. The O Westin/micro sff thing is interesting. I’ve been following and enjoying their twitter feed.
    I doubt that writing fiction – even tiny fiction – was meant to be covered by Fan Writer, but the wording doesn’t actually seem to rule it out.

    Like

  5. I know we already have a bazillion categories, but having one for micro-fiction is good. Maybe with the idea you need enough micro work to add up to X amount, where X is small, but more than one drabble. Because you’d want to reward people who put out lots of good stuff, like O. Westin, instead of someone just tossing off one drabble.

    Liked by 1 person

  6. Huge congrats to all the winners and all the worthy finalists.

    I’ve been travelling this week, so I just got a chance to watch the parts of the BM that I cared most about, where they decided on 3SV (no), EPH+ (no), and suspending EPH+(no). There was one thing that was said repeatedly which I believe is a mistake. I’ve crossposted this at File770 and Making Light but I think I’ve finally actually found the right place for it here.

    So, the idea that I think is mistaken is that this year’s outcome shows that under EPH, bullet voting works.

    The category in question is best ELF, best editor long form. In the final elimination round, Vox Day had 83 ballots and 83.00 points; Miriam Weinberg had 131 ballots and 54.25 points; and Patrick Nielsen Hayden had 118 ballots and 65.42 points. Weinberg and Hayden went into the cage match because they had the lowest points among the final 7; Weinberg won because she had more ballots than PNH; and so Vox Day survived. People have drawn the conclusion that this shows that bullet voting works, because VD would not have been among the top 6 under the old rules, and PNH would have.

    But consider what would have happened if 82 of the 83 VD voters had voted a slate — say, Vox Day, Loco Prentiss, Vinnie Von Ritas, Cav E. Temptor, and Poe Stock. Initially, each of the slate would have had 16.4 points except VD who would have had one more, 17.4. As soon as all the ELFs with fewer points were eliminated, two of the slated ELFs (say, Cav and Poe) would have cage-matched and one of them would have been eliminated, leaving the remaining ones with 20.5 points. That would have happened again, putting the remainder to 27.33; then again, putting the remainder to 41; and finally, VD would have been the only puppy standing with his full 83 points. At no point would any of them have faced Weinberg or PNH

    What’s more, even if one of the puppies had faced down Weinberg or PNH, it would have not been VD up for elimination, as he would have stayed one step ahead of his fellow puppies. So the other puppy would have been easily dispatched by MW/PNH, and VD’s points would have surged just as if he had been the one highlandering his fellow canine.

    The upshot is that, under EPH, it is /not/ the fact that the puppies bullet voted that gave them an advantage; that was merely incidental. Their advantage came from the fact that their choices did not overlap with those of other voters. Arguably, in this case, EPH was doing exactly what it was designed to do: increasing the diversity of nominators who had some finalist who they supported. The fact that PNH had 118 ballots but only 65 points means that a majority of PNH nominators supported some other finalist or finalists. Eliminating PNH left only perhaps a score of voters without a finalist they supported; eliminating VD would have left over four score without.

    tl;dr: in EPH, there is no incentive to bullet vote by leaving weak candidates off your ballot; but there is an incentive to “free ride” by leaving people who can win without your vote off your ballot.

    (Note: some degree of “free rider” incentive is an inevitable feature of all proportional representation voting methods. EPH’s “free rider” incentive is actually relatively weak. And in the end, the “free-rider” problem is inherently self-limiting, because the more likely you think it is that other people will try to free-ride, the less of an incentive you have to do so yourself.)

    Like

    • Thanks Jameson.
      That shows up with the Jeffro/Morgan vote in Fanwriter. Jeffro wasn’t a bullet vote but one Morgan was eliminated the effect was the same. The advantage, such as it was, was having a set of voters with minimal intersection with other nominees voters rather than the number of candidates.

      Now the anti-slate property of EPH was known but this points to another feature: it is a bad idea for Vox to have hostage nominees in a category with a ‘payload’ nominee e.g. if Vox nominated File770 and Castalia House blog for Fanzine then that would only help File770 by giving it more votes while lowering the points of Castalia House blog.

      Like

      • Actually, the McQuarrie (“Star Wars Art”)/Gaiman (“View from the Cheap Seats”) vote in Related Work that I mentioned in my previous contribution here illustrates that last point beautifully. The figures strongly suggest that there were 57 RP bloc nominations for McQuarrie and Gaiman, and 19 just for McQuarrie – I don’t know how close (or indeed voluntary) an RP ally McQuarrie was, but he does looks like the RP ‘payload’ nominee in the category and Gaiman was certainly a hostage. If the 57 McQuarrie/Gaiman nominations had instead been made just for McQuarrie, Gaiman would have been eliminated in one of the two last rounds and McQuarrie would have been in second place on points (though only a point or so ahead of Carrie Fisher’s “Princess Diarist” rather than Gaiman’s 26 points).

        If VD was just trying to see what would work under EPH, without any particular concern about which candidate the RPs propelled onto the final ballot, it certainly shows what doesn’t work. If he actually wanted McQuarrie on this year’s ballot, the RPs scored a rather spectacular own goal.

        Liked by 1 person

Blog at WordPress.com.