Ack! The author doesn't quite understand how this is implemented, nor how OkCupid handles matching people. If she did, I imagine she would have found the practice itself significantly less objectionable, though her points about the e-mail's phrasing stand.
(Full disclosure: I have an OKC profile, and on My 28th, I received the e-mail mentioned in the article.)
Members on OkCupid answer questions about lifestyles, dating, sex, etc. (e.g., "Do you think it is appropriate for parents to choose their children's religion or a lack of religion for them?") and then state how their ideal match would respond, as well as how important it is that a partner have similar views.
Your set of responses is compared to every other profile's set of responses, and the intersection between each pair is used to calculate a "match percentage," which is a pretty good predictor of how folks will get on, or at least have compatible views. [0,1]
Almost any displayed link to another profile is accompanied by your match percentage with that individual. It's a major component of the site.
Now, when searching for people to message, most folks set up basic filters -- gender, orientation, age, kids, distance, etc. You have to order the result set somehow, and that's often either just the match percentage, or you can elect to sort by "Match & Distance," "Match & Last Online," etc.
As far as I know, searching is the primary means through which people find each other on the site, and this change will not impact that at all.
There are a few places where OkCupid will show you somewhat random collections of profiles: On the homepage when you log in, in the sidebar on every page, and in two features: Quickmatch and Quiver.
Quickmatch is just a procession of profiles that you're asked to rate from one to five stars. If two people rate each other highly (or poorly), the site notifies them. Your "quiver" holds a rotating set of three profiles that their algorithms think are particularly good matches which you can choose to message or explicitly ignore.
In all but one of these cases, the candidate profiles are displayed with little or no additional information beyond a picture. On Quickmatch, the user's entire profile is displayed, but you must rate them before their account name is revealed and you're able to communicate with them.
The email indicates that OkCupid is using click-through data from these sources ferret out which profiles are the most appealing to its users. The only change that results from being one of those profiles is that when OkCupid shows you a random profile, it is more likely to pick from the set of more appealing profiles.
That's it.
If your profile is not within the upper half of all profiles, then you'll still get the same distribution of recommendations as before. And your profile will still show up in OkCupid's recommendations to more appealing users, albeit slightly less often.
But again, that's only for places where OkCupid itself recommends profiles -- on the front page, in the sidebar, on Quickmatch, and in Quivers. Search ranking remains absolutely unchanged.
Contrary to the author's assertion that "the dreaded bottom 50% presumably do not have access to potential matches in the top 50%," the bottom 50% still has exactly the same access to the top 50% as before. They're just less likely to be randomly suggested to the top 50%.
I'd also challenge the assertion that "the policy reinforces the notion that hot people deserve love and happiness and "ugly" people should just get used to being alone." Rather, I think this may be designed to reward users who take the time to craft engaging profiles by showing them other profiles that have been vetted by the community.
The policy assuredly does not "[make] dating decisions on the users' behalf without considering personal preferences" since match percentages and search rankings are completely unaltered.
While the policy is indeed "kept secret," and the language is offensive, I find it hard to accept that the policy is segregational or that it has a significant impact on the site's operations. I'm not even sure it's "attractiveness-norm-reinforcing," since they're tracking click-through rates, which is not necessarily attractiveness.
(Full disclosure: I have an OKC profile, and on My 28th, I received the e-mail mentioned in the article.)
Members on OkCupid answer questions about lifestyles, dating, sex, etc. (e.g., "Do you think it is appropriate for parents to choose their children's religion or a lack of religion for them?") and then state how their ideal match would respond, as well as how important it is that a partner have similar views.
Your set of responses is compared to every other profile's set of responses, and the intersection between each pair is used to calculate a "match percentage," which is a pretty good predictor of how folks will get on, or at least have compatible views. [0,1]
Almost any displayed link to another profile is accompanied by your match percentage with that individual. It's a major component of the site.
Now, when searching for people to message, most folks set up basic filters -- gender, orientation, age, kids, distance, etc. You have to order the result set somehow, and that's often either just the match percentage, or you can elect to sort by "Match & Distance," "Match & Last Online," etc.
As far as I know, searching is the primary means through which people find each other on the site, and this change will not impact that at all.
There are a few places where OkCupid will show you somewhat random collections of profiles: On the homepage when you log in, in the sidebar on every page, and in two features: Quickmatch and Quiver.
Quickmatch is just a procession of profiles that you're asked to rate from one to five stars. If two people rate each other highly (or poorly), the site notifies them. Your "quiver" holds a rotating set of three profiles that their algorithms think are particularly good matches which you can choose to message or explicitly ignore.
In all but one of these cases, the candidate profiles are displayed with little or no additional information beyond a picture. On Quickmatch, the user's entire profile is displayed, but you must rate them before their account name is revealed and you're able to communicate with them.
The email indicates that OkCupid is using click-through data from these sources ferret out which profiles are the most appealing to its users. The only change that results from being one of those profiles is that when OkCupid shows you a random profile, it is more likely to pick from the set of more appealing profiles.
That's it.
If your profile is not within the upper half of all profiles, then you'll still get the same distribution of recommendations as before. And your profile will still show up in OkCupid's recommendations to more appealing users, albeit slightly less often.
But again, that's only for places where OkCupid itself recommends profiles -- on the front page, in the sidebar, on Quickmatch, and in Quivers. Search ranking remains absolutely unchanged.
Contrary to the author's assertion that "the dreaded bottom 50% presumably do not have access to potential matches in the top 50%," the bottom 50% still has exactly the same access to the top 50% as before. They're just less likely to be randomly suggested to the top 50%.
I'd also challenge the assertion that "the policy reinforces the notion that hot people deserve love and happiness and "ugly" people should just get used to being alone." Rather, I think this may be designed to reward users who take the time to craft engaging profiles by showing them other profiles that have been vetted by the community.
The policy assuredly does not "[make] dating decisions on the users' behalf without considering personal preferences" since match percentages and search rankings are completely unaltered.
While the policy is indeed "kept secret," and the language is offensive, I find it hard to accept that the policy is segregational or that it has a significant impact on the site's operations. I'm not even sure it's "attractiveness-norm-reinforcing," since they're tracking click-through rates, which is not necessarily attractiveness.
Links about match percentages:
[0]: http://blog.okcupid.com/index.php/2009/09/29/how-races-and-r...
[1]: http://www.okcupid.com/faaaq