When computer science meets human science

A couple months ago, a fuss arose because some researchers had published a study based on Facebook data[1] that had been collected without anything like what we in academia regard as informed consent. Will Robertson responded with the most comprehensive explanation of why this study should not have been conducted and should not have been published I have seen,[2] and if you haven’t already done so, this is a must read, because Robertson nailed it (and, in the process, effectively reminded me that I could probably use a refresher on the topic of research ethics myself). Along these lines, Tim Carmody has tackled questions not only of informed consent but of how sites like Facebook and OkCupid represent their users.[3] This, too, is a must read.

Now the founder of OkCupid, Christian Rudder, has sought to defend Facebook’s practice, saying his outfit does the same.[4] Actually there are some differences, as David Holmes at Pando Daily has noted, that probably mitigate OkCupid’s experimentation to some degree.[5]

I want to begin with where I agree with Holmes. His main point is that “the OkCupid founder’s arrogantly flippant attitude in regards to these experiments raises some red flags about how responsible the company will be in the future when it comes to toying with its users’ emotions.”[6] Yes, this is absolutely right. If nothing else, when asked about the possibility of bringing in an ethicist, Rudder asked rhetorically, “To wring his hands all day for a $100,000 a year?”[7]

Holmes zeroes in on this,[8] an apparent joke,[9] but for me this is, before anything else, a red flag that yes, OkCupid does indeed need an ethicist. Desperately.

I have some qualms about dating sites anyway and these need to be explored in order to illustrate the ethically tenuous ground these sites trample on. First, they are run by profit-making enterprises, which means that they are exploiting the very human need for contact, and especially intimate contact, for selfish gain. If we, in this society, were not so indoctrinated with neoliberal ideology, surely this would seem suspect. One might even say it would seem downright creepy.

Second, the matchmaking methodology is entirely and inherently quantitative, which is to say it is hopelessly superficial. Even where space is offered for users to explain their answers, the algorithms have no way to take account of this. They can only go by the true/false or multiple choice responses to questions which, in OkCupid’s case, are generated by users, which means the questions will often be malformed. Such questions may fail to allow for all possible responses and they may be ambiguously worded. What is presented as a single question may in fact be more than one question, making a single answer from the choices offered even more difficult.

In short, the data which OkCupid relies on for matchmaking is garbage data. Yet we are to trust this organization’s founder when he claims that these experiments are all “just part of the scientific method.”[10]

That’s not to say that OkCupid’s research questions were not legitimate. The one about whether “people just like each other because they think they’re supposed to,” worrying about the site’s own power of suggestion, a power which seems to have been significant,[11] seems to me to be tremendously worthwhile. But ironically, the importance of that question raises its own ethical issues. Here’s Carmody:

First, going on “a date you didn’t like” isn’t a frivolous thing. It definitely incurs more material costs than not seeing a Facebook status. And bad (or good) messages or a bad or good date can definitely have a bigger emotional impact as well.[12]

So the quality of OkCupid’s matchmaking is itself a question with serious ethical ramifications.

It also raises the matter of informed consent. One of the foremost ethical values in human research is a notion that humans are autonomous beings, entitled to make informed decisions about the experiments that are conducted on them. Here again, Rudder is flippant:

I was in some psych experiments when I was in college, just because they give you 20 bucks to go to the department and you, y’know, you sign a form. But that is informed consent – which users can’t see, but I’m putting in quotes.[13]

I don’t know about Rudder’s experiences with psychological experimentation in college, but just “sign[ing] a form” does not meet the requirement. Researchers are supposed to ensure that participants understand the research that is being conducted, fully understand the risks and benefits of their participation, and agree to participate without being coerced or unduly enticed into doing so. (The $20 inducement, however, is probably acceptable.)

That’s why burying “consent” in web site terms of use doesn’t cut it. That’s why all those university psychology experiments where freshman-level students are “encouraged” to participate by professors who have power over them (grades are just the beginning here) to participate don’t cut it. And that’s why OkCupid’s experiments on people hoping to find significant relationships (for even a fairly loose sense of ‘significant’) don’t cut it.

One of the reasons that the Proceedings of the National Academy of Science cited in excusing its own handling of the Facebook study was that, in an Associated Press paraphrase, “as a for-profit company governed by its own terms of service, Facebook had no obligation to adhere to those scientific principles.”[14] The same logic would apply to OkCupid, except in the sense that at least Facebook brought in university-affiliated researchers to study the data. OkCupid isn’t even doing that. And as Christian Rudder acknowledged, “OkCupid doesn’t really know what it’s doing.”[15]

Actually, I do know what OkCupid is doing. It’s messing with people in intimate details of their lives while only employing the pretense of scientific method.

High technology is one of the few areas where the United States is ascendant. But what we see is that especially the people in power within this industry tend to value technical competence and to devalue the humanities.[16] Indeed, my own experience of people in the high tech world is that they often devalue a college education (which itself is only minimally demanding in the humanities). They think they know better.

But what we are also seeing is that they know nowhere near enough about the ethics of human experimentation. And for an industry that is assuming the kind of power over our lives that high tech is, that’s a very worrisome thing.

  1. [1]Adam D. I. Kramer, Jamie E. Guillory, and Jeffrey T. Hancock, “Experimental evidence of massive-scale emotional contagion through social networks,” Proceedings of the National Academy of Science 111, no. 24 (2014), doi: 10.1073/pnas.1320040111
  2. [2]Will Robertson, “Facebook’s Unethical Research Project,” Skepchick, June 28, 2014, http://skepchick.org/2014/06/facebooks-unethical-research-project/
  3. [3]Tim Carmody, “The problem with OKCupid is the problem with the social web,” Jason Kottke, August 1, 2014, http://kottke.org/14/08/the-problem-with-okcupid-is-the-problem-with-the-social-web
  4. [4]Christian Rudder, “We Experiment On Human Beings!” OkCupid, July 28, 2014, http://blog.okcupid.com/index.php/we-experiment-on-human-beings/
  5. [5]David Holmes, “OkCupid’s human experiments weren’t that bad — but its cofounder’s flippant attitude over user ethics is,” Pando, August 4, 2014, http://pando.com/2014/08/04/okcupids-human-experiments-werent-that-bad-but-its-cofounders-flippant-attitude-over-user-ethics-is/
  6. [6]David Holmes, “OkCupid’s human experiments weren’t that bad — but its cofounder’s flippant attitude over user ethics is,” Pando, August 4, 2014, http://pando.com/2014/08/04/okcupids-human-experiments-werent-that-bad-but-its-cofounders-flippant-attitude-over-user-ethics-is/
  7. [7]Christian Rudder, quoted in Alex Hern, “OKCupid experiments are standard ‘scientific methods’, says founder,” Guardian, August 4, 2014, http://www.theguardian.com/technology/2014/aug/04/okcupid-dating-ethics-facebook-experiments
  8. [8]David Holmes, “OkCupid’s human experiments weren’t that bad — but its cofounder’s flippant attitude over user ethics is,” Pando, August 4, 2014, http://pando.com/2014/08/04/okcupids-human-experiments-werent-that-bad-but-its-cofounders-flippant-attitude-over-user-ethics-is/
  9. [9]Alex Hern, “OKCupid experiments are standard ‘scientific methods’, says founder,” Guardian, August 4, 2014, http://www.theguardian.com/technology/2014/aug/04/okcupid-dating-ethics-facebook-experiments
  10. [10]Christian Rudder, quoted in Alex Hern, “OKCupid experiments are standard ‘scientific methods’, says founder,” Guardian, August 4, 2014, http://www.theguardian.com/technology/2014/aug/04/okcupid-dating-ethics-facebook-experiments
  11. [11]Christian Rudder, “We Experiment On Human Beings!” OkCupid, July 28, 2014, http://blog.okcupid.com/index.php/we-experiment-on-human-beings/
  12. [12]Tim Carmody, “The problem with OKCupid is the problem with the social web,” Jason Kottke, August 1, 2014, http://kottke.org/14/08/the-problem-with-okcupid-is-the-problem-with-the-social-web
  13. [13]Christian Rudder, quoted in Alex Hern, “OKCupid experiments are standard ‘scientific methods’, says founder,” Guardian, August 4, 2014, http://www.theguardian.com/technology/2014/aug/04/okcupid-dating-ethics-facebook-experiments
  14. [14]Associated Press, “Journal expresses ‘concern’ over Facebook study,” July 3, 2014, http://bigstory.ap.org/article/journal-expresses-concern-over-facebook-study
  15. [15]Christian Rudder, “We Experiment On Human Beings!” OkCupid, July 28, 2014, http://blog.okcupid.com/index.php/we-experiment-on-human-beings/
  16. [16]See, for example, Guy Raffa, “What the Head of Hiring at Google Doesn’t Understand About Skills,” Chronicle of Higher Education, May 28, 2014, http://chronicle.com/blogs/conversation/2014/05/28/what-the-head-of-hiring-at-google-doesnt-understand-about-skills/

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.