« Downtime at WorldCon | Main | Hugo voting: Don't forget the lower-profile categories! »

Hugo voting subtleties: The No Award test

| No Comments

I wrote a blog post a few months ago about how to use No Award effectively and what else to include on your ballot, but I intentionally sidestepped the complication of the No Award test. So here's an attempt to belatedly address that.

I actually wrote this entry a while ago, but after I wrote it I realized that I didn't actually think the No Award test was going to have any effect this year, so it wasn't nearly as important to tell people about it as I originally thought. So I decided not to post it. But new information about the number of voters this year (see the end of this post) means that it's just slightly more possible that the info in this post will turn out to be relevant this year, so I'm belatedly posting it now.

But it's still unlikely to be relevant, so if you don't care about Hugo voting subtleties that have only a small chance of influencing the outcome, then may want to skip this (rather long) entry.

The idea of the No Award test is that, because of the preferential voting system used by the Hugos, it's possible that a work or person might end up winning even though a lot of people are strongly opposed to that work or person winning.

And so after all the instant-runoff rounds are over and a winner has been provisionally determined, there's one more test:

The vote-counting system looks at the number of people who put the winning item above No Award, and the number of people who put No Award above the winning item. If the latter number is higher than the former number—that is, if the number of people who would rather see no award given is greater than the number of people who want the item to win—then no award is given in the category. To see how this works, see the example below.

A further subtlety here: Remember that leaving something off your ballot is the same as voting for it to be in last place. So if I put Best Book Ever on my ballot but I don't list No Award in that category, then I'm saying that I would prefer Best Book Ever over No Award. Likewise, if I put No Award on my ballot but I leave off Best Book Ever, that counts as saying that I would prefer No Award.

As far as I can tell, the No Award test has never resulted in No Award winning any category. According to the winners by year list, there have been five times that No Award has won, but the last time that happened was in 1977, before the No Award test was created.

So if it's never happened, why am I bothering to post about it?

Because it seems to me that it has a better chance of happening this year than has happened before, because of the polarization of this year's ballot.

Here's a simplified example to demonstrate what I'm talking about:

Let's look at an imaginary ballot that contains two works known as Best Liberal Novel Ever and Best Conservative Novel Ever. Now imagine that there are 100 voters: 45 of them are ardent liberals, 45 are ardent conservatives, and the other 10 don't want politics in their fiction.

Now imagine that all 45 of the liberal voters vote this way:

  1. Best Liberal Novel Ever
  2. No Award
  3. Best Conservative Novel Ever

and all 45 of the conservative voters vote this way:

  1. Best Conservative Novel Ever
  2. No Award
  3. Best Liberal Novel Ever

and all 10 of the no-politics voters vote this way:

  1. No Award
  2. Best Liberal Novel Ever
  3. Best Conservative Novel Ever

In the instant runoff, No Award is eliminated first, because it got the fewest first-place votes. Then Liberal Novel has 55 votes, and becomes the provisional winner.

But then comes the No Award test, which finds that 45 people put Liberal Novel before No Award, but 55 people put No Award before Liberal Novel. And thus, no award is given.

Now imagine a slight variation on those votes that results in a different outcome: Imagine that in the above scenario, 15 of the conservative voters didn't fully fill out their ballot. Instead, they just voted for Conservative Novel and didn't mention No Award or Liberal Novel at all. That is, those 15 conservative voters just put this on their ballot:

  1. Best Conservative Novel Ever

In this variant scenario, the instant-runoff provisional results still go the same way; Liberal Novel gets 55 votes and becomes the provisional winner. Then comes the No Award test, but this time there are only 40 people who put No Award before Liberal Novel, because some conservative voters didn't declare a preference between No Award and Liberal Novel. And thus, Liberal Novel is declared the winner.

So what does this mean to you, the Hugo voter?

It means two things:

  • If there's a work that you really don't want to win, then you should put No Award above it on your ballot. This is really my main point in this entry—as I said before, make sure your ballot reflects your preferences.
  • If enough other people put your favorite work below No Award, then your favorite work won't win, even if it ends up with the most votes.

I've been trying keep this abstract and general up to this point, but I think all this might be a little clearer with a real-world specific example. So let's look at this year's Novelette category, which includes stories by Mary Robinette Kowal and Vox Day. I'm picking those two authors in particular because I suspect that many voters see them as being at opposite ends of the political spectrum; I expect that most liberal voters are strongly opposed to Day's story, and I imagine that some conservative voters may be strongly opposed to Kowal's story. (That may not be at all true, I don't know, but for the sake of this example let's assume it is.)

So if enough people rank No Award above Day's story, then it can't win even if it gets a lot of votes. And if enough people rank No Award above Kowal's story, then that can't win even if it gets a lot of votes.

I should say that I suspect the No Award test will continue this year to have no effect on the results. For it to have an effect in either political direction, there would have to be a whole lot of voters who are strongly opposed to the work that would otherwise have won. I don't think that there will be enough voters in favor of (for example) Day's story for it to be declared the provisional winner; and I don't think that there will be enough voters strongly opposed to Kowal's story that they'll be able to prevent it from winning (if it otherwise would have won) by ranking No Award above it.

(And in the Novelette category, I suspect Ted Chiang's story will win anyway. But there are two other fiction categories where political factors might come into play.)

In 2013, the detailed voting results showed that (for example) in the Best Novel category, 1254 ballots ranked Redshirts (the winning novel) above No Award, while only 151 ballots ranked No Award above Redshirts. The outcome of the No Award test was similar in the short-fiction categories. So last year, 1000+ additional people would have had to rank No Award higher than the winners for any of the winners to fail the test.

In the original version of this post, I concluded with the following two paragraphs:

I will be very surprised if the Correia slate ends up bringing in 1000 new voters, and even if it did, I'll be even more surprised if enough of them use No Award in such a way as to prevent a provisional winner from winning, in any category.

But it is theoretically possibly, and if the No Award test is ever going to affect a vote's outcome, I think it'll probably be in a year like this one.

But now, in mid-August, new data has been revealed: this year's WorldCon received 3,587 valid Hugo ballots, more than half again as many as the previous record of 2,100.

Which means that there are somewhere on the order of 1,500 new voters this year.

I suspect that at least half of those are Wheel of Time voters, most of whom probably don't care strongly about anything else on the ballot. And I suspect the other half of the new voters are roughly evenly split between conservatives supporting the Correia slate and liberals opposing it. And if I'm right about that, or even anywhere close to right, then the No Award test won't have any effect this year.

But I could be totally wrong about all of that; I'm talking through my hat here. When I wrote this entry, the idea of a thousand new voters coming in on one side or the other seemed so extremely unlikely that I decided it wasn't worth muddying the waters over. But it now seems slightly more possible.

So I still don't expect the No Award test to come into play in any category, but it might. We'll find out in two days!

Post a comment