OKC is a free dating site that I joined years ago to take their amusing personality tests. Last year, they started blogging about their data (
Does the cougar thing really happen?
Do pictures of self with cute animals help your cause?
Should you display abs/cleavage? etc.). While the blog is woefully heterocentric, I was nonetheless intrigued and actually added some data and pictures to my profile. (I wasn't looking very hard at the time, but I was single and open to possibilities, so it's not like I was leading people on.) It transpired that if you upload the pictures first, before filling in any text about yourself, you'll get so hammered with chat requests that it's almost impossible to finish your profile. I'll admit that's a little flattering - and way depressing. Like,
Nice pics, wanna fuck? What, you don't care enough about who I am to give me 30 minutes to finish my profile? Just as long as you can beat the crowd to the chat request? Color me underwhelmed.
Once you get all that sorted, the site's not bad, though - there's a ton of people you wouldn't give the time of day, obviously, but the utility is in how many people you do find interesting, not how many you don't, and I must say that I've met some interesting and hawt people on there.
The next amusing thing is the
scales recently tipped in your favor
part - I haven't really changed my pictures in a while, so what does that mean? Everybody else suddenly turned fug? A lot of new people who were all fug joined? They changed some algorithm? And "top half of the most attractive users" could be anything from bad phrasing for "top half of the userbase when sorted by attractiveness" (putting me at n/2+1 at worst; if the algorithm isn't new, "recently" would also indicate I made it there from the realm of the average) to "we have a group of most attractive users the size of which we won't tell you" at best (making it n=2 at best, in which case making it to the top half would technically make me the most attractive woman on the site, lol). So since it could mean pretty much anything, it's really only saying something akin to "you're not totally fug," and "you're, at least recently, for whatever reasons, above average." Could mean anything between "kinda average" and "kinda gorgeous."
If I hadn't been laughing so hard, I'd actually have found some of the answers in the next paragraph - it's click-throughs on pictures in their
features. Which I suppose means in real English, when they suggest me as a potential match, with just the picture in one and both pic and some personal data in the other case, people are interested enough to want to know more. Fair enough.
Fair enough? I've set up my likes and requirements so I corner a small segment of the market, or at least so that a small segment of the market corners me. Or so I thought. (In other words I selected,
woman interested in women
.) In which case of course, I'd get some attention from that group at first, being new. Then, attention would drop sharply as everybody in that niche has already checked me out, and new people trickle in slowly at best. In fact, even that initial interest must seem quaint at best to the outright deluge of profile views I got while I hadn't set up the
bit yet, so it really boggles the mind. By all accounts, I should have experienced a vast decline
in attractiveness, not an increase!
So, I thought, maybe they account for this? Maybe it's some new feature? Let's check out their blog!
Amusingly, I found no clarification on that issue, but a related-sounding entry
on a new feature called
My best face
. You upload some pictures to MBF, or tell it to use the ones already in your profile, and then it'll tell you which ones were the most popular, and how popular that was exactly, and with what kind of people. (Example report
. Not mine.)
Sounds great, right?
There's a few problems here. First off, the site will automatically generate icons from the pictures you submitted, presumably so you remember which ones you already put up. Thanks to the funky dimensions of some of my pictures and the dumbness of the algorithms, one of the icons shows only hair and no face. This may be fine if the person viewing it is some pervert with a hair fetish (like yours truly), but it'd still be nice if one had some way of knowing whether the actual picture that people will see got cropped, or just the icon.
Then, the way this works is by showing people sets of two pictures (of the actual sex they identified themselves as being interested in, yay!), and have them click on whichever they find more attractive. A bit like amihotornot
, only, not. To get your own profile, you must rate 60(!) sets first, which honestly seems a bit steep to me. It doesn't help that there isn't a
option (but you'll get a new set to choose from when you reload the page (by confirming the URL with return; if you simply use the reload button/hotkey, you may end up sending the same message (
liked the left picture
liked the right picture
) again, so beware!), so that works. Sort of. Sometimes both pictures are attractive, in which case it pains me to skip as per above so not to be unfair, but what can you do?
To add insult to injury, a lot of the
categories of people who liked this picture - artists, dorks, introverts, etc.
aren't calculated by some new profile-analyzing magic; for most of the categories, you self-identify as you sign up for MyBestFace. To make matters worse, I had a hard time finding good categories for myself.
I'm finishing my 60 while writing this. It's not always easy. The feature is called,
my best face
, but even when the faces are clearly visible and in focus and all, the request is,
Pick one for a date!
Sometimes, that makes it awfully tempting to pick someone who is just doing
something interesting in their picture. I'm down with being attractive being something you do
rather than something you are
, but when you award point this way, the whole "best face
" thing stops making sense. "My best picture" would have been clearer, it not exactly catchy.
30 to go.
get old. I wish they'd made it 25, or 30. 60 is a little much. I briefly think that they have all these click-through data from
anyway, but then remember that quiver shows you the person's primary
shows you all
their pictures, so it's insufficient for working out which is their best
25 to go.
In your report, they show you a bunch of "pageants" you were in, whether you won or lost that one, and a random person of all who voted in that comparison. So, it may show someone who voted against them. You might want to decide how comfortable you are with that before you enter.
20 to go.
If you insist on being in a picture with your friends, and your friends all look like dorks, I really don't feel comfortable voting for you.
15 to go.
This shit is beginning to make me feel like a judgmental dork.
10 to go.
Also if you're in a picture with your friends, and they're of the same gender you are, make sure we bloody know which one you are, eh?
5 to go.
Did I mention you also get a fancy map showing
people who like you are located? cos you do.
OK, the results are in. First off, only people in the US, the UK, and close to the UK voted for me. I'm not sure what that means. Do I have some particular sort of anglo appeal? MBF has people from all over vote on the pictures, not just people who live nearby (in fact that would only work of both parties in the pageant lived near each other anyway). I said that the target audience is small, so by that token, my local girls may have seen my profile before, but most of the remote ones wouldn't know me and therefore can't have vote me down on accounta "not a homegirl", "snotty #$#$@ only speaks English?", etc.
As you know if you clicked on the example
link above, the results show your pictures, in order of preference according to the public vote. Then to the right of it, they show you who liked the given picture. Those rows seem a bit hocus pocus, since if it means "preferred one of your
other pictures", with two rows you'd expect that if the Christians preferred picture #1, you'd get an equal but opposite
column for Christians in the other picture. Or if it meant, "preferred somebody else's picture to this one of you", that if a picture has no negative responses at all, from none of the groups, like my number one doesn't, you later wouldn't see people who voted against
that picture in the details. I'm assuming that the problem lies in the fact that to the extent that I can tell, people don't vote on all
or even some
of your pictures, but only (a random) one
. To rephrase, different people voted on different pictures of yours, and since people can be in several categories rather than just one, the results look a tad non-obvious at first. That, or I shouldn't be writing on statistics this late at night. Which would still be way sad considering stat was on my curriculum at uni.
Since I, perhaps stupidly, sent the thing on its merry way with my existing profile pictures, rather than choose pictures particularly for this endeavor, it ended up using the following three in the pageants - somewhat amusingly, they came in exactly the order I use on my profile, so I must be doing something right.
#1 has no negative bars … I can has rawk?
It's not even close to being one of my favorite pictures, I just figured it made a decent profile picture since you can see a little context, you can at least guess at my body-type, and you can sort of see my face (I'm in a lot "artsy" pictures where the angle or the hairdo end up obscuring my face to a degree).
Of course, I feel way
guilty now, because using model-y stuff and composed images tends to set up expectations that are hard to meet if you later meet people in real life. On the other hand, some people have some phenomenally crappy snapshots of themselves on there. A better experiment would have been to use some studio shots, and some good
"real world" photos, and see which fare better with what groups.
Bit hard to know what happend here, since apparently, the picture got cropped in weird ways. That probably didn't help. Very possibly, some people also hated it otherwise, but it's hard to tell how much of which, and you gotta hate when that happens, it makes for crappy science!
Even if in this case, the "science" obviously requires scare quotes. I can only guess at what happens here.
Vegetarians tend to have a certain overlap with the granola and henna hippie fraction, so the glamor might not score with those. Perhaps similar for dorks, whatever that even means
in this context. Introverts may be put off but that whole "rockstar" (-wannabe) thing. But what about the geeks? Maybe the important thing here is that it's geek women
. (Someone once told me I look like a RealDoll in that picture …) But even so, the thing that puzzles me most is the pronounced rejection by the 23-30 olds. Now you could argue that I'm "slowly but surely drifting out of that age group" (in fact, I'm already older than that in LA, and I'm scared to convert it; in metric, I'm probably 50!), but if "presumably older than myself" counted against, you'd expect the same negative reaction or worse from the 18-22 year olds. Curious and curiouser still!
Now this was universally panned.
It probably shouldn't even be in there, it's already somewhat problematic even in the profile (A lot of the "judges" are in the US. The UK is a drinking culture; the US, not necessarily so much. Don't believe me? Check out the prevalance of "drinks often" in UK vs US profiles. Even if you just view it in how socially acceptable it is, and not in terms of how much drinking actually occurs.). But at least in the profile, I could add a caption that would sort-of indicate that it's a joke shot. Fishnets, of course, also aren't necessarily the epitome of class. Picture does showcase hair and legs though, and at least hints at body-type. Face could certainly be more visible, but that didn't keep the 23-30s from liking it. In fact, they were the only ones who liked the picture at least a little
. Again, this one should have been a more regular shot instead, out and about, or out with friends, or something. Even so, it's fascinating just how
badly this fared - if it were just me, period, all the pictures would do badly. With some of the other pictures, only people who I seriously don't find attractive either voted against them, so it's all good (whether we're people into seriously different ideas of looks, or whether it's that "gravitating towards people of our own attractiveness level" that comes out in some studies is hard to tell, and perhaps irrelevant). In this case though, while some hot people voted for it, some hot people also voted against
it, so this picture is definitely "a lost opportunity." Interestingly, in one case it even lost against a picture that is so dark that you can't see much at all of the person - face, body type, nothing -, so there must be something that really
antagonizes people. So apparently the answer to, "Which one would you date?" is, "I'd prefer no
date to her
, she looks like a lush!" Or maybe, it's the fishnets.
In fact, the more I look into the details, the less they make sense. Even with the other pictures, I sometimes won against people I consider more attractive than myself, and lost to people less so. Goes to show how varied human taste is, bending in all directions like a blade of grass in the wind, like they say in Sweden. Love is a many splintered thing.
That probably goes even more for lesbians. Holy fuck, some of those choices were so weird that it's hard to believe people weren't simply poisoning the well, or clicking all 1s or all 2s to get through the 60 questions already and see their own MyButtFace results already. (That should statistically even out in the end, but gives rise to some headscratching when you look at individual rows in good faith.)
So, just so you don't think I'm kvelling too much, there was also a good amount of rejection in the mix, but between the badly auto-cropped #2 and the questionable joke #3, I have half a heart to run another set of pictures at some point, for clearer results. If I do, I'll probably amend this article, or write another one. (On a side-note, if you blog about this, you may wish to make sure your blog doesn't try to put a trackback on the OKCupid-blog, thereby exposing your identity to potential date-wannabes!)
That said, I have far less faith in this tool than in their main matching algorithm, and even so, MBF needs better documentation. It's not quite up to OKC's usual standard yet.
By special request, here are clearer versions of #2 and #3:
I play the guitar, the piano, and the fool. Two out of those badly, as well.
This chick Marla Singer did not have testicular cancer. She was a liar.