Lies, damned lies, and statistics. #KissMyArs #EqualPrizesForEqualWork

Memo Akten
11 min readSep 5, 2015

UPDATE: Having received even more responses to my tweets, I realize that my tweets have been very insensitive and misleading. One could say I’ve done a Richard Dawkins on Twitter. I'm truly sorry about that. To try and extract some kind of positive outcome from this, I can at least say I've learnt a lot. Thank you all for the feedback.

The Backstory

Ars Electronica Festival is happening. There was only 1 female winner this year. Tweets with #KissMyArs #EqualPrizesForEqualWork started trending with the following image:

The TLDR Summary:

10% female win percentage is a terrible outcome, hilighting a problem I really sympathize with. I acknowledge the systemic bias that gave rise to these results, and I try to do what I can within my limited ability.

Some tweets — not all, perhaps only a handful — claimed a bias specifically in the jury process (which there probably is), using this 10% statistic as evidence. With this particular claim in mind, I tweeted asking about the percentage of submissions from women, and I tried to point out that one couldn't use this 10% statistic as evidence for bias in the jury. Being able to squeeze concise arguments into 140 characters (including user handles, links and hashtags) is not always easy, especially when the topic is complex and delicate. I ended up in a few tweet debates that got rather long and difficult to follow and I found myself facing a plethora of straw man arguments.

In retrospect my tweets failed miserably in communicating what I’d wanted. It appeared to many that I was trying to deny the problem altogether or even ‘blame the victims’. So I wrote this text not to try and defend my tweets— which I won’t — but to try and express my original thoughts and discuss the problem which really concerns me.

I was trying to point out, you can’t use the 10% winning statistic as evidence for bias in the jury. That may seem like a tedious little detail, but I think it’s an important one. Because it’s a fact, not an opinion. You cannot use the 10% win percentage on its own as evidence for systemic bias in the jury. It’s a plain error in logic. And that error in logic — along with the lack of acknowledgement of the error by other sympathizers of the cause— fuels polarization.

Others who are sympathetic to the cause either didn't see this error, or they chose to ignore it. Most of the discussion appeared to be positive, constructive, and centred on how to improve diversity in the field. I completely appreciate the value and necessity of taking this approach. And that makes it so much more difficult to be the bad guy that brings up and focuses on this error in logic.

Because I can’t help but feel that ignoring the error, pretending that it isn't happening, and just focusing on improving diversity is a really really bad strategy.

On one hand the sympathizers are getting all fuelled up — and rightfully so, 10% win percentage is a terrible outcome that needs to be addressed. So the topic is trending (in the dedicated circles). Meanwhile, those who are apathetic, or unsympathetic to the cause — the ones that do blame the victims and deny the problem — do see this error in logic. They also see other sympathizers wave these flawed arguments around. So they conclude that they are correct in blaming the victims and denying the problem. It pushes them further away. They assume that the whole movement is based on these ‘delusional errors in logic’.

And I've now discovered, that you can’t even try to talk about this. You can’t try and point it out, because you also get accused of trivializing the issue and blaming the victims.

Yes in retrospect my tweets were terrible and I failed to communicate any of this. But I tried to start this discussion, and it bring it there. I take responsibility in my failure to do so. But it also seems like no one tried to understand. It really felt like I was touching upon a taboo.

This was a brief summary of my thoughts on the matter. Below are similar thoughts fleshed out a bit more.

The Long Version

The Premise

The shared image states 10% of winners have been women, but doesn't mention what percentage of submissions were from women.

It has been argued that focusing on the submission percentage can be misleading, because it can lead to victim blaming and “problem is they don’t want to apply” type claims. It can be seen as a denial of the problem altogether, that women bring it upon themselves. A very valid concern I appreciate. But asking what the submission percentage is, is not the same as focusing on it. Of course we should try and address all biases across the board. But that doesn't mean we shouldn't try to understand the distribution of the problem.

I am surprised at how asking the question has resulted in such an immediate backlash and a fury of straw man arguments. It seems asking the question implies that one denies the problem, or doesn't take women seriously, or tries to put the responsibility on women and ‘blame the victims’. Granted, in <140 character tweets it’s difficult to communicate the subtlety of these points. But I am concerned as to why this question immediately causes so many to assume malicious intent as opposed to a genuine curiosity and concern. I presume it shows the extent of the ‘denial of the problem’.

Women making up just 10% of all winners is a terrible outcome, no matter what the submissions percentage. There should be no doubt about that. It exposes a systemic bias across the culture of the festival and related fields. Women are very under-represented in many areas of STEAM — let alone media art — so this doesn't come as a huge surprise unfortunately. But one still wonders why the results are so bad. Are lots of women applying but not winning? Is there a specific bias in the jury process that needs to be addressed? Or maybe do very few women apply in the first place? What’s the ratio of women in the invitations sent out? How is the attendance at Ars? In the field in general? Of course the problem is across the board. You will find inequality at every step of the way. But surely knowing the distribution of the problem isn't a bad thing?

I tried to get an idea of this figure. It’s not reliable and actually not relevant to the rest of this article so I won’t cite it here, but you can see in the side notes if you’re curious. Even though the figures are very inaccurate, I calculated and tweeted them because no one was talking about this, and I think people should be. I wanted to raise this question and encourage others to investigate further.

The Jury Bias Claim

While some of the #KissMyArs tweets were expressing general dissatisfaction, frustration, criticism, anger, hope-for-change with the situation (a position I fully sympathize with), others had more specific allegations of a systemic bias within the Ars jury and winner selection process, that devalued the work of women, making it more unlikely for them to win (from now on I shall refer to this as the JuryBias claim).

We can rarely be certain of anything, especially when dealing with only statistics, but we can have levels of confidence in what we claim. If we knew the percentage of submissions from women, and it was significantly higher than 10%, then we could at least use it to support the JuryBias claim. But without knowing the submissions percentage, you absolutely cannot use the 10% on its own to support the claim.

(NB. This used to be a sidenote, but medium.com have recently made sidenotes inaccessible to the public: If one were to count all submissions by women, and find that it was <= 10%, then a 10% win ratio is potentially understandable. In that case the 10% win figure doesn’t actually support the JuryBias claim. On the other hand if the submission percentage turns out to be significantly > 10%, then one could use the data to support the claim. But these are just statistics, on their own they aren’t evidence. I’m not suggesting that there should be a one-to-one ratio between submissions and wins. These statistics can only appear to *support* the claim, or not. That’s it.)

And that’s my principle point for this section:

Using the 10% win figure as evidence for the JuryBias claim, without having the submissions percentage, is completely unjustifiable and a fallacy.

This might seem like a tedious little detail to some, but it is an error and I think it’s very important. Because it weakens the credibility of the overall argument, and thus fuels polarization. I mention the significance of this shortly.

Whether there is other evidence for bias in the jury, doesn't actually make a difference to the above statement. In fact it makes it even more tragic. Because the other evidence might be true, and there very well might be bias in the jury, but using the 10% statistic incorrectly like this is still logically incorrect and thus is damaging to the credibility of the argument. In the eyes of the apathetic or sympathetic, it nullifies the other evidence and further fuels polarization. More on this below.

Explanation vs Justification

It seems my tweets came across as if I was excusing Ars — and possibly some of my explanations above might too. It’s a common fallacy to ‘confuse an explanation with an excuse’, and I should make it clear that the above is an explanation not a justification.

I'm not saying any of this to defend Ars, or to deny the existence of a problem. I'm also not denying the JuryBias claim. There might very well be other evidence of this which I'm not disputing. My principle point for this section is simply the error in logic which is hilighted above, which I believe damages the credibility of the overall argument.

(NB. This used to be a sidenote, but medium.com have recently made sidenotes inaccessible to the public: I should also point out — in case some are wondering — that I have no affiliations or personal connections with Ars. I have no obligations to defend them. Yes I have won a golden nica in 2013. I went over, shook their hands, they were very nice and polite, and that’s it. I have no further relations :)

The Slight Digression

The Evidence & The Ethics

Imagine person X commits murder. Then others start publicly accusing person X of rape as well. If you think the evidence for rape is insufficient or interpreted incorrectly, it’s understandable for you to expose that, and suggest that the particular evidence doesn't actually support the allegations of rape. It doesn't mean that you forgive them for the murder. It doesn't even mean that you believe they are innocent of rape. It just means that you believe we need to be honest and careful how we present evidence.

Update: In retrospect, I appreciate how insensitive it might be, questioning the accusations of rape, while still mourning the murder. And I think that might be at the heart of this particular problem.

The Statistic Abuse

Waving around the statistic of “52% of all homicides in the US between 1980 to 2008 were committed by African Americans” by itself is irresponsible without presenting information regarding what percentage was born into poverty; born into violent environments; faced racial discrimination, prejudice and socioeconomic inequality from birth. In fact if one were to mention the 52% statistic without presenting the associated information, to infer a racial basis for violence, I will lose respect for them and dismiss them as racist (that’s perhaps a prejudice on my part).

Conclusion

So why am I so hung up on this tedious little detail on how the 10% figure is used to support the JuryBias claim?

Because sexism — as is many other biases — is woven into the very fabric of every aspect of our society. It’s so embedded into our culture and our lives that it’s like the air molecules we breathe. Such a fundamental part of our norms that we can’t even see it unless we stop to explicitly look for it, and it suddenly slaps us in the face. And when we do see it, we can’t unsee it, we realize we’re completely immersed in it, living and breathing it every day.

And that’s a huge part of the problem. That the problem itself is invisible, and hidden in plain sight. The issue isn't just about addressing biases, and increasing diversity. The issue is also about spreading awareness of the fact that there actually is an issue. The fact that so many people assumed I (and others who asked about the submissions %) were ‘deniers’ is indicative of that. These two aspects are obviously tightly coupled, but one does not automatically induce the other.

This is why I am very sympathetic to #KissMyArs and similar initiatives. Particularly in the style of Addie Wagenknecht’s tweet “In the history of #arselectronica15 10% of Nica’s go to women. We can do better #kissmyars #equalprizesforequalwork.” (I read ‘we’ as ‘everyone — men, women, writers, bloggers, speakers, artists, Ars, other festival organizers, juries etc’).

So why won’t I shut up about how the 10% figure is used to support the JuryBias claim?

Because it’s actually not a tedious little detail, it is a major error in logic, and it weakens the credibility of the overall argument — even if the claim is true.

When we start over-stretching our claims and we make logical fallacies in this quest to spread awareness, we will face critique and ridicule.

I know I'm not the only one to spot this error in logic regarding the claim. Lots of people are seeing it. Those who are already sympathizers and acknowledge the issue with systemic bias, are choosing to ignore the error, deeming it insignificant in comparison to the greater cause. Those who are not sympathizers… well that’s the issue.

The audience that is currently not aware of or acknowledging the problem of bias, is alienated if we declare logically incorrect arguments. Especially if those logically incorrect arguments are taken to heart and spread by other supporters, waving them around as if that is what the argument is based on.

While trying to raise awareness, the situation is being compromised even more by encouraging polarization. We may be trying to reach those who are currently not sympathetic to the issue, but not only are we not making them aware of the issue, we are pushing them away. All we’re doing is reminding ourselves of how bad the situation is, and fuelling our existing followers, while encouraging the apathetic to take unsympathetic views, or reinforcing their unsympathetic views even further.

The fallacy in question might not even be a terrible one, or ill-intentioned. The claims themselves might even be correct, just not presented in a sound way. But when trying to expose problems and spread awareness in delicate situations, one needs to be more vigilant than usual to not make these mistakes. Otherwise they act as ammunition for the unsympathetic. The discussion is prevented from entering larger audiences and can get stuck in a small closed circle of flag bearers, preaching to the choir. A pattern which unfortunately is quite common in almost every ‘us vs them’ debate these days.

This does not mean we should refrain from making bold, even controversial statements for fear of alienating others. On the contrary we should, but these bold statements should be accurate, not unsubstantiated. That just weakens the credibility of the argument and fuels polarisation.

“The most perfidious way of harming a cause consists of defending it deliberately with faulty arguments.”

― Friedrich Nietzsche

(NB. This used to be a sidenote, but medium.com have recently made sidenotes inaccessible to the public: There is the point that controversial statements lead to spreading awareness and productive discussions. While I have no doubt that it can trigger productive debate, I do wonder how productive that debate really is in the bigger picture if it’s also fuelling polarization. Especially if unsubstantiated or exaggerated claims are being made specifically for this very reason. I would love to see research about this).

The fact that we can’t even seem to discuss this, without immediate accusations of ‘victim blamer’ or ‘problem denier’ makes matters worse. But that’s for another time.

--

--

Memo Akten

computational ar̹͒ti͙̕s̼͒t engineer curious philomath; nature ∩ science ∩ tech ∩ ritual; spirituality ∩ arithmetic; PhD AI×expressive human-machine interaction;