Ursa's Major Issue - Confident self-promotion vs humble passionate skill, and a voting system's favoritism
Ursa, we may have a problem. Or at least so it may seem. Over the past few years the number of people voting in the furry fandom’s popular choice awards appears to be dwindling once again, despite continual growth and booming attendance at our conventions - COVID aside - revealing the growing audience and community beneath this stagnation.
But if less people in proportion are voting, is there a reason for this? One option may be that the system may be lead to some strange victors based on popularity of a franchise or personality rather than other considerations. But is this just a coincidence or could it be how the system was inadvertently crafted?
This article’s goal is to highlight why the current system is so sensitive to favoring artists who self-promote or whose fans rally on their behalf, at the expense of voters that weigh more toward judging the quality of the pieces nominated without authorship considerations. It will then propose a small change to make it more fair to both types of voters and creators, without stifling out those who show up with a passion for their artist.
Be wary, this gets long and technical. But if you’re reading this up to this point, I’m sure you knew that’s what you were in for.
A tale of two creators
This year, a single content creator had a goal to win as many awards as they possibly could, ended up getting nominated in three categories, and after a rigorous campaign of self-promotion won them all. Before we go too deep into this article, let me note that I hold no animosity for them doing so. The rules are simple and self-promotion is always part of life. If you want to win a contest based on popularity, sometimes you need to put yourself out there. If there was a fourth award for most avid fanbase, then Garrison should be proud of them because they clearly have dedicated fans. They worked consistently for a long period on their art, and it's good they were recognized for it.
However, this is not the first year that the impact of an artist’s popularity has over the measure of the artistic content has caused issues. On the other side of the spectrum, Kyell Gold won at least one Ursa Major for writing for six straight years between 2005 and 2011. He didn’t go around promoting himself as avidly as Garrison [though publishers took note]; but he has an avid fanbase, to the point where the author felt that him winning with such frequency was taking oxygen out of the room for other authors. So before voting for the Ursa Major’s began in 2012, he publicly stated that he no longer wished to have his works considered. (One of this year's winners has done something similar for 2022.)
But is this a coincidence that avid fans bring in victories, or is there something further that causes this system to be sensitive to it? To understand this we must break down the voting systems and methods of tally further.
A primer on voting systems
Before we can go over how to fix a system that has a bias, we must prove the system has one. We must also understand that there are many different ways you can collect and tally votes, and that you want the outcome to be the one that fits what the people want the best, rather than whatever gerrymandering is. The best voting systems are tailored for their purpose, but also would require less gamification on the voters’ and nominees’ part.
To go over this there are two systems we need to focus on when it comes to tallying the votes for the Ursa Majors: the first past the post method (FPP) and the alternate vote (also known as Instant Runoff voting).
Fortunately for me, instead of writing out these systems, YouTuber CPGrey goes over both of them in two videos which I have linked above. In short, the FPP method has a nasty little side-effect called the spoiler effect. Once people know that, they tend to create competing blocks that in essence eventually radicalize the nominees and voters into two binary fiefdoms in order to provide an external solution to the flawed voting system.
This is what I was alluding to in the Tweet I made referencing the situation. Though in hindsight saying “all democracies” was a bit short sighted, mostly those with these flawed voting FPP voting systems. To be fair, most tweets tend to be shortsighted, and it’s why I wear glasses.
Now, when we think of the ranked system of the Ursa Majors we may think that because we are voting for three candidates that this is actually akin to the Instant Runoff, and that would prevent spoiler effects. But as we will discuss in the next section, it is actually a First Past the Post method that is modified in such a way that it actually makes the spoiler effect worse.
Current system and its bias
Currently the rank system used in the Ursa gives each voter effectively six points to distribute among three of the five nominees: three to one, two to another, and one to the last. However they do not have to give all their six points. They can give just three to one pick and burn the remaining points. However, this and of itself is a vote: three points to one nominee and three points to nobody. So in truth it’s similar to a winner-take-all system where everyone gets six votes to distribute how they wish within those noted restrictions. The ranking actually doesn’t serve the function of ‘ranking’ but just giving votes more weight of a person’s six votes.
This option of burning three of their votes is how self-promoters and stanning fans can effectively gain an advantage in this system. An avid fan of the artist or property will usually opt out of picking three nominees and just give three points to the one they want to see win and leave all the other four nominees to starve. You can see where this happens thanks to the statistics provided by the volunteers at the Ursa Majors.
To note before we get into these examples, this phenomenon happens in all categories in all years to some degree. The examples we are using have to do with the three Garrison nomination in 2021, but if you go back through the data you’ll see it happens all the time. Garrison being so open with their campaign gives us the opportunity to break down how the system gives advantages to those who do self-promote over others. I'm glad they highlighted this flaw in the system then someone with more nefarious intentions.
So let us take the three categories Garrison won for this year as examples. I have added additional information to help illustrate the story the numbers are telling:
Burned Points and their bias
Notice the total number of votes on Published Illustration selected for rank #1 is 275, and for #2 is 187. This means 88 voters in this group decided to burn three of their points instead of giving them to the other artists. In effect, since they are sacrificing those points, it makes their three points more powerful to their chosen person than ones who give them three and then throw points at other artists.
In essence, this gives a leg-up to a voter wanting one item to win, rather than one that takes the time to rank them and use all six of their points.
We can understand this better by simplifying it to a scenario with two voters. Voter A must use all six points distributed by the Ursas’ rules, and voter B must use three points to one candidate and burn the rest. In this scenario whoever voter B picks will win definitively if any of voter A’s points touch their candidate of choice, this has a 60% chance of outcome if done at random. Voter A will only win for their top pick if it agrees with B, a 20% chance that is shared within the 60% above. In the remaining 40% there will be a tie between Voter A’s and B’s top choice. Therefore B’s voting tactic is more powerful in this exchange, they can have their top choice win independent from A’s top choice, but not the other way around.
We can compare this to a one vote per person system. There will only be a winner if A and B agree on the nominee which will happen 20% if at random. The rest of the time there will be a tie. A and B can win and tie with equal probability, and always at the same time. There is no systemic bias.
Highlighting the Burner’s bias with a Median
The total of votes alone is not enough to dictate which of the nominees benefited from the populist pressures upon the voting system. That’s where the median of the votes can help to further highlight the phenomenon. The lower the number, the higher the probability that the person in question had their fans come in and vote for them as #1 and burning the rest of their votes. Because if a work is competitive in quality, they should have a good proportion of people considering it as their #2 or #3 choices along with those making it their top pick.
This is a bit less reliable when the masses seem to agree upon the order of their votes organically, you can see this in the comic strip votes for this year. Excluding the Garrison entry, voters seemed to definitively rank Foxes in Love for their number one pick, and Freefall as number two. This causes the medians to be much more distinct and not as correlated to the winner by points.
This could also be occurring since comic creators distribute content on a regular basis. Due to this they are more likely to have a fresh presence in the minds of people, so in essence each entry works similar to promoting oneself, which a more static style of content such as an illustration or short dramatic work does not have.
What we see in all three of the Garrison victories is a low median on their work coupled with a high top ranking distribution. As noted before, this happens all the time in the Ursas to an extent. It’s fine that people come to vote to support their senpai. But the person passionate about one person should not have an advantage over those that are analyzing the works on a more personal preference of content level.
Spirit and intent
I do not believe that when this simple system was developed that it was intended to function in this way. More likely the nerd who took it upon themselves to create this voting plan, like most of us nerds, assumed that people voting in this would be as passionate as they are for the arts and would always distribute their three options, thus leading to the best art winning given everyone would use all three of their ranks. They did not think that a person would be passionate for only the artist behind one of the works and only hand out three points to them. However, you can see that this appears to always happen to some extent. From my scan of the data, the number of #1 picks tends to be greater than those who pick a #2 option, is greater than those who pick a #3 option.
I would like to think that the Ursa Majors’ intention was to be awarded to the best content as voted by the fandom as a whole, rather than who make the best content and can run the most effective voting campaign. As GreenReaper responded to my tweet above, not everything is about politics. I would agree, so we need to have a system that does not show favoritism toward being able to art and schmooze. Just the art, thanks.
So what can be done? Luckily for the Ursas I believe how the votes are collected can remain the same, but how they are tallied must change. This is where the Alternative Vote comes into play.
Solution: true Instant Runoff Voting
I believe this was the original intent of the Ursa Majors’ pick-three system, but the way in which it was implemented created the problem mentioned above. It took some ideas of FPP voting and instant runoff voting and bred them together. However instead of mitigating the spoiler effect, it just exacerbated it.
The foundation of this solution is that everyone should have one vote, and one vote only. None of these three votes here, two votes there, etc. This is Highlander, there can be only one. However, the ranks you put in give you an opportunity for your voice to be heard even if your first pick won’t win.
Currently we are saying “I give three votes to A, two votes to B, and one vote to C”. Instead it will change to “I give my vote to A, but if A loses I give it to B, and if they lose I give it to C”.
You could extend this kind of ranking to all five nominees if you wanted to update the way things are tallied, but I think starting off without changing how we collect votes will be the easiest way to change things before changing anything further if desired. Plus I think having a #5 rank may be unnecessarily harsh to the nominees if the results are shared publicly.
In theory what this system would change is that for the first several rounds, the self promoter would win handily. As the other nominees get voted off, though, the secondary votes of those voted off would congeal into nominees that will more than likely not be the self-promoter. We can theorize this due to the median distribution. In the final round, the self-promoter is going to be facing a more humble artist who gathered power from their fallen humble colleagues, and then come out as a much stronger opponent in the final round.
Voters will continue to vote the same way, ranking items 1, 2, and 3. Also like now, the voter can pick one vote they want as their #1 and leave the other options blank. But if said voter has their choice lost in the runoff voting process they simply forfeit their vote being passed on to another choice.
Instead of distributing points, you go through a recursive loop and eliminate the least favorite until only the favorite remains. The loop is as follows:
- Take all #1 ranked votes and discover which nominee has the least of them.
- Eliminate the nominee found in step one. If tie for bottom eliminate all worst (what if final 2, number of prior top wins in prior rounds?).
- For the eliminated nominee. See which voters picked that as their #1 ranked vote.
- For these voters, take their Rank #2 votes and transform them into Rank #1.
- Repeat the above steps until only one nominee remains (up to 4 times).
Of course one big issue with this is that publishing the results may not be as clean as the currently provided statistics. Basically you’d probably have to show each round of voting in each category, up to the maximum of four rounds, to get an idea as to their ranking and how ‘close’ things were. Also what if there is a tie at any point the process and how that would get resolved, as noted in the list above that may need some additional thought.
But the good news is we wouldn’t need to test this system on live data in a future Ursa. We can use prior Ursa votes as a template to see how utilizing this would impact the results. It would be an excellent way to test how well a system like this would help make things fairer for those not unionizing their votes.
For we should all want a system where we all have a choice and that choice isn’t made with consideration of having to mitigate the possibility of a group of people ganging up prior to the selection process. It should be you, and the vote, without giving an unnecessary edge to artists who can also lead a tribe.
In a system that removes the additional bias toward the confident and self-aggrandizing, the humble may have a chance to breathe, and show a passion for the Ursa’s that may get stifled otherwise. And that is something everyone participating should want, especially since it is about promoting the art.
But if nothing changes, I at least look forward to winning the dramatic short in 2022. Just kidding, fortunately I suck at self promotion.