More hot stats from OKCupid
August 10, 2010 10:39 AM Subscribe

OK Cupid statistics fun: We collected 552,000 example user pictures. We paired them up and asked people to make snap judgments. Here's what we found.
posted by nomadicink (109 comments total) 37 users marked this as a favorite

That link's format makes it completely unreadable!
posted by The 10th Regiment of Foot at 10:44 AM on August 10, 2010

Huh. OKCupid is really starting to do interesting things with the stats generated by their userbase. When they first started doing statistical analyses, they seemed pretty mundane but they've actually gotten more and more interesting as time has passed.
posted by zug at 10:44 AM on August 10, 2010

I hate myself for loving this shit.
posted by Mister_A at 10:44 AM on August 10, 2010 [7 favorites]

In conclusion:
"It's actually not that hard. Use a decent camera. Go easy on the flash. Own the foreground. Take your picture in the afternoon. Then visit the nearest Apple store. Done."
posted by yoyoceramic at 10:45 AM on August 10, 2010 [4 favorites]

"Don't be ugly by accident."

What about people who are ugly on purpose?

(I'm looking at you, YOU KNOW WHO YOU ARE)
posted by mr_crash_davis mark II: Jazz Odyssey at 10:45 AM on August 10, 2010

Quantifying desperation.
posted by wcfields at 10:46 AM on August 10, 2010 [2 favorites]

True story, I got a Panasonic GF-1 a few months ago, and life hasn't been the same since.
posted by Threeway Handshake at 10:46 AM on August 10, 2010 [4 favorites]

That link's format makes it completely unreadable!
posted by The 10th Regiment of Foot at 12:44 PM on August 10

Can you explain this? Because I found it to be perfectly readable.
posted by kingbenny at 10:46 AM on August 10, 2010

Mister_A: I hate myself for loving this shit.

Me too.

Back when I used to use dating/hook-up sites regularly, I always meant to put together a list of "ways to do it right"/"how to bat above your league"/"how to create a decent profile" list of tips that I meant mostly as humor but also as helpful.

OK Cupid is doing that plus "science" which equals "much love" from me.
posted by MCMikeNamara at 10:47 AM on August 10, 2010

Thank God, I was afraid this was going to be another pointing-and-laughing, "look at these ugly people" link. Charts and graphs, hooray!
posted by Gator at 10:49 AM on August 10, 2010

Can you explain this? Because I found it to be perfectly readable.

About half a second after the page opens it blots itself out with a giant "Share this with your friends banner" that takes up most of the screen and won't go away.
posted by The 10th Regiment of Foot at 10:50 AM on August 10, 2010

A million bitter, dateless young men just smacked themselves on the forehead saying "Of course! It's must have been my EXIF data!"
posted by griphus at 10:50 AM on August 10, 2010 [15 favorites]

Those DSLRs can really bring out your DSLs.
posted by condour75 at 10:50 AM on August 10, 2010 [7 favorites]

Yeah, bokeh is king when it comes to portraits. My first SLR lens was a 50mm f/1.4 and it was the best photography purchase I ever made. Everything comes out looking like Serious Business headshot portraits. I disagree that it's the intimacy factor that makes them appealing, it's more the case that the photos come out looking more like what we actually see with our own eyes. You can take a picture of a GI Joe with some christmas lights strung up in the background and people will ooh and aah over it.
posted by mullingitover at 10:50 AM on August 10, 2010 [7 favorites]

If these graphs are right, I was supposed to have had sex with 18 people by now. With my SLR.
posted by swift at 10:51 AM on August 10, 2010 [1 favorite]

I dunno. The shirtless pic of me while throwing a gang sign that my buddy took with my Motorola Razr has worked wonders so far.

{/}
posted by special-k at 10:52 AM on August 10, 2010 [2 favorites]

If those graphs are right, I am supposed to have an iPhone! Statistics suck!
posted by Mister_A at 10:53 AM on August 10, 2010 [1 favorite]

Agree with 265 here... it isn't often that OKCupid misses the big picture (no pun intended), but they do here.
posted by roomthreeseventeen at 10:53 AM on August 10, 2010

According to one chart, one could make the statement that a person reaches peak attractiveness at 2:00AM. As expected that's an entirely different picture at 8:00AM.
posted by Nanukthedog at 10:53 AM on August 10, 2010 [4 favorites]

Stats!
posted by ColdChef at 10:54 AM on August 10, 2010

Right off the bat they miss an alternative explanation for the "better cameras take better photos" correlation. Namely, better cameras are also more expensive, and thus more likely to be owned by wealthier people, who for various reasons may be judged more attractive. Similarly, I suspect the iPhone probably stands out amongst camera phones not because its camera is so great (it ranges from pretty terrible to decent depending on the model) but because it's more likely to be owned by younger, wealthier people.

To an extent this is undercut by the subset that uploaded photos from phones, point and shoots, and SLRs (the SLR-taken photos of those subjects were judged best), but I wouldn't mind seeing income controlled for.
posted by jedicus at 10:55 AM on August 10, 2010 [4 favorites]

Also: 50mm, the Forgotten Lens: Why You Should Ditch That Zoom for a Classic 50mm "Normal" Lens.
posted by ColdChef at 10:56 AM on August 10, 2010 [3 favorites]

There's at least a soft correlation between expensive cameras and skilled photographers. I know it's not something you can control for, but I wonder how much it's "Have your picture taken by someone who knows how to use a camera and understands image composition" rather than "Have your picture taken with an expensive camera."

I was going to say the same thing. If you hand someone who doesn't know anything about photography a DSLR, the photo is going to look better than a blurry low-res cameraphone shot, but the average person who already uses a DSLR knows a lot more about how to use the manual features of the camera and photography fundamentals than the average person with a point-and-shoot model.
posted by burnmp3s at 10:58 AM on August 10, 2010

The best pics I've ever taken were on a Pentax ME super with a 50 mm lens, ColdChef.
posted by Mister_A at 10:59 AM on August 10, 2010

I wonder how OKCupid would rate Rita Hayworth circa 1945? Actually, I don't wonder that at all.
posted by blucevalo at 10:59 AM on August 10, 2010 [1 favorite]

Yeah, this is great. That totally wasn't the direction I expected a post on user photos to go.
posted by chunking express at 11:01 AM on August 10, 2010

Is it really the 50mm that makes a difference? Can't most P&S cameras easily do 50mm equivalent? Or is it the f1.8?
posted by smackfu at 11:01 AM on August 10, 2010

Interchangable lens cameras (like digital SLRs) make you look more attractive...

CAUSATION FAIL
posted by DU at 11:05 AM on August 10, 2010 [1 favorite]

Yeah, a whole lot of that is explained as "people with better cameras will take better photos." It's not even that the camera owner is richer (because the owner isn't necessarily the subject of the photo), just that if you're using a DSLR, you likely know how to take better photos than someone using a cameraphone.

This, however, does nothing to explain why iphone users get more action. If that's not a marketing opportunity, I don't know what is.
posted by whatnotever at 11:06 AM on August 10, 2010 [2 favorites]

I'm interested in the apparent fact that female Iphoners are more promiscuous than their age-matched male counterparts, and to a greater degree than other brand owners. Explanations? It would be nice to see a some matching weight statistics here to do some category matching.
posted by meehawl at 11:06 AM on August 10, 2010

It should be noted that on 95% of modern consumer dSLRs a 35mm lens is the equivalent to a 50mm lens on film bodies. Crop factor kids!
posted by lattiboy at 11:06 AM on August 10, 2010 [4 favorites]

My age puts me off the charts!
posted by mareli at 11:06 AM on August 10, 2010 [1 favorite]

Is it really the 50mm that makes a difference? Can't most P&S cameras easily do 50mm equivalent? Or is it the f1.8?

Prime lenses are always, always, sharper than a zoom. They are like magic. Primes also can open up really wide, and people really like the shallow DOF.

P&Ss cannot possibly come close, unless they're one of those fixed lens uber ones like the Leica, Ricoh and Sigma models (that cost as much or more than a DSLR).
posted by Threeway Handshake at 11:07 AM on August 10, 2010

Also worth mentioning that very few people (even among DSLR owners) have a f/1.4 lens, while virtually nobody can justify the expense or practicality of an f/1.2.
posted by schmod at 11:07 AM on August 10, 2010

This link confirms once again that I am a sucker for absolutely anything that attempts to quantify something.
posted by pemberkins at 11:09 AM on August 10, 2010 [1 favorite]

Yeah, fine, Mr. Statistics, but vampires don't show up in photos and they're the sexiest of all!
posted by lore at 11:09 AM on August 10, 2010 [19 favorites]

There are some zooms that approach or surpass the sharpness of some primes though. 50mm f/1.8 versus the 70-200 f/2.8L IS? No contest.

Of course that's comparing a $100 lens to an $1,800 lens, but sharp zooms *do exist*. That shot even has a 1.4x extender on it...
posted by WinnipegDragon at 11:10 AM on August 10, 2010

"oh, also—iPhone users have more sex."

I thought this was wrong until I realized the comparison was BB and Android users.
posted by SirOmega at 11:10 AM on August 10, 2010 [5 favorites]

Anyone else notice that the time-vs-attractiveness graph is approximately inversely proportional with this?
posted by nzero at 11:11 AM on August 10, 2010 [1 favorite]

I'm not sure what's going on with Kodak all the way to the right there. They might want to consider making sharing more difficult.

QFT
posted by blue_beetle at 11:11 AM on August 10, 2010

ColdChef: "Also: 50mm, the Forgotten Lens: Why You Should Ditch That Zoom for a Classic 50mm "Normal" Lens."

My only SLR is analog and my primary lens is fixed 50mm f/1.4. This is the biggest thing preventing me from upgrading to digital. I feel silly buying a $400 body and a $1500 lens, but anything less would be a downgrade.
posted by Plutor at 11:13 AM on August 10, 2010

These are interesting numbers/theories/plots. But, I think the problems with this analysis start early in the experiment design:

2. We paired them up and asked people to make snap judgments, like so:

As someone who deals with this sort of a thing for a living, my opinion is that the study design is, well, flawed. A couple of thoughts:

1. Pairwise comparison studies are notoriously difficult to do right. You need a good set of guidelines and a set of judges who are trained to be consistent. In other words, if judge A says that X is hotter than Y, and judge B says Y is hotter than Z, it does not automatically follow that X is hotter than Z, unless the judges are deliberately applying a consistent set of guidelines and standards. Getting this right is difficult and I am sure if effort was expended on this part, it would have been mentioned. If this were an actual published study in a magazine, I would expect to see a specialized measure like Cohen's Cappa for quantifying inter-rater agreement (which requires survey overlap - see below)

2. Even among very good, trained judges, using single pairwise comparisons to obtain an absolute ordering would not usually work. To get more believable numbers, it is prudent to assign overlapping sets (says, show the same pair of pictures to 5 or 10 judges), and then take majority agreement and throw out the outliers. Of course, doing 5x overlap means that the number of judgments just went up 5x - not a cheap proposition. Again, I think if OKC folks did this, they would have mentioned it

3. Hate to keep harping about boring statistical stuff, but for these numbers to be taken seriously you need to include significance tests. Look at the first graph, for example - it says that photos taken by Panasonic Micro 4/3 cameras get best ratings. Well, it's not a really popular format (unfortunately so, I must say). In the degenerate case, if there was only ONE photo taken by that camera, and that photo got best possible rating (e.g. 5 out of 5), could you really say that Panasonic cameras are better than Canon DSLRs (e.g. a million photos with average rating of 4.5 out of 5). Reporting the value of significance test in this imaginary example would tell you exactly what's going on. Without the test... well, it's just another misleading bar graph.

I fully understand and appreciate that this is folks at OKC having fun with the big, interesting dataset that they have access to. It is a cool project that doesn't seem to aim for true research paper status, and hey all the power to them. All I want to do is point out that the numbers are not believable on their own - but they sure are fun to look at and speculate.
posted by blindcarboncopy at 11:14 AM on August 10, 2010 [18 favorites]

Plutor, a 50mm f/1.4 should only run you about $400?
posted by WinnipegDragon at 11:14 AM on August 10, 2010

mr_crash_davis mark II: Jazz Odyssey: ""Don't be ugly by accident."

What about people who are ugly on purpose?

(I'm looking at you, YOU KNOW WHO YOU ARE)"

I'm sorry. I'll leave now.

...sniffle...
posted by Splunge at 11:16 AM on August 10, 2010 [3 favorites]

Wow. According to the "number of sex partners by age by brand of phone," I am a blackberry user who has had as many partners as an iphone user. That is kind of great to know.
posted by millipede at 11:16 AM on August 10, 2010

Hrmm. The GF-1 ships with a really excellent 20mm (40mm equivalent) lens.
posted by bonehead at 11:20 AM on August 10, 2010

This, however, does nothing to explain why iphone users get more action

I call BS on this, no women seem at all impressed when I'm on the train playing Final Fantasy 1 & 2 or Chaos Rings on my iPhone.
posted by Kirk Grim at 11:21 AM on August 10, 2010

50mm, the Forgotten Lens: Why You Should Ditch That Zoom for a Classic 50mm "Normal" Lens.

Two things:

For 99% of the cameras out there, 50mm isn't.
Focal length has absolutely nothing to do with it, anyway.

posted by Civil_Disobedient at 11:24 AM on August 10, 2010 [1 favorite]

Is it really the 50mm that makes a difference? Can't most P&S cameras easily do 50mm equivalent? Or is it the f1.8?

Mostly, it's the f1.8, combined with the high quality glass often found in 50mm lenses. The mathematics/optics of a 50mm lens happens to translate to a lens that's very easy to produce both inexpensively and at a very high quality.

The long-ish focal length combined with the wide f/1.8 aperture translates to a shallow depth of field, and very "bright" photo, which reduces motion blur, high-ISO noise, and the need for a flash.
posted by schmod at 11:24 AM on August 10, 2010 [2 favorites]

Hrmm. The GF-1 ships with a really excellent 20mm (40mm equivalent) lens.

That lens is DOUBLE RAINBOW GOOD. And look how cute and tiny it is!
posted by Threeway Handshake at 11:24 AM on August 10, 2010

Also worth mentioning that very few people (even among DSLR owners) have a f/1.4 lens, while virtually nobody can justify the expense or practicality of an f/1.2.

Not true at all, because of what Winnipeg said. Anyone who owns a DSLR should own a 50mm 1.4. The Canon 50mm f/1.4 is $350. Buy it. Use it. You will take better pictures and learn to move your feet and you will love the results more than anything you've ever shot before.
posted by The Bellman at 11:25 AM on August 10, 2010 [1 favorite]

The question is whether the f/1.4 is 4 times better than the f/1.8, considering it costs 4 times as much.
posted by smackfu at 11:29 AM on August 10, 2010

oh, also—iPhone users have more sex... Finally, statistical proof that iPhone users aren't just getting fucked by Apple

If iPhone users get laid so often, why are the iPhone fans here so uptight?

That's a joke.
posted by grouse at 11:33 AM on August 10, 2010 [1 favorite]

I note that in 'Sexual Activity by Smart Phone Brand', there's no column for Palm...
posted by a young man in spats at 11:33 AM on August 10, 2010 [16 favorites]

The question is whether the f/1.4 is 4 times better than the f/1.8, considering it costs 4 times as much.

In the Canon ones, hell yes. The f/1.4 is much sharper, has better bokeh, metal mount, and USM. Well worth it. The jump to the f/1.2L is a little more questionable.

At that point, I think I would buy the 85mm f/1.2 instead.
posted by WinnipegDragon at 11:34 AM on August 10, 2010

grouse: "If iPhone users get laid so often, why are the iPhone fans here so uptight?"

Note: They don't get laid more often, they have more sexual partners.
posted by Plutor at 11:34 AM on August 10, 2010 [3 favorites]

If you're really good with your Palm, you don't need anyone else.
posted by grouse at 11:34 AM on August 10, 2010 [14 favorites]

IMO, at least in Nikon world, the 50/1.4 and 50/1.8 lenses have very similar optical performance at f/2.0 or above - very sharp, mediocre bokeh, very low CA. Both lenses are soft wide-open, which is to be expected. I own a 50/1.4 because I bought it used for cheap, and the extra f-stop makes it pretty handy for shooting music acts without a flash. But I would recommend that just about everyone get the 50/1.8 and save the cash. Except for the crop DSLR owners - 50 mm is a bit too long on a crop sensor. They'd be better off with cheap, excellent Nikon 35/1.8 (link).
posted by blindcarboncopy at 11:35 AM on August 10, 2010

For DSLRs with APS-C sensors, something in the 25-30mm range gives you equivalent view to a 50mm on a full-sized sensor or 35mm film. Unfortunately, they can be more expensive than 50mm of similar aperture.
posted by tommasz at 11:41 AM on August 10, 2010

It kills me that there is not a 35mm f/1.8ish Canon lens worth a crap at under $200. It's a huge hole in the Canon lineup compared to the Nikon offerings. The 35mm f/2 is 'okay' but outside that it's the 35mm f/1.4L at about $1,100 or nothing.
posted by WinnipegDragon at 11:43 AM on August 10, 2010

At that point, I think I would buy the 85mm f/1.2 instead.

That's a hell of a lens. I use it on movies for close-ups. rental, because there's no damn way I can afford the thing.
posted by drjimmy11 at 11:43 AM on August 10, 2010

That's a hell of a lens. I use it on movies for close-ups. rental, because there's no damn way I can afford the thing.

Yup, I rented one as well. I ended up with the 70-200 f/2.8L for it's versatility but there is nothing out there quite like the 85 f/1.2L. The 135 f/2L might be in the same ballpark, but that's about it.
posted by WinnipegDragon at 11:47 AM on August 10, 2010

It kills me that there is not a 35mm f/1.8ish Canon lens worth a crap at under $200. It's a huge hole in the Canon lineup compared to the Nikon offerings. The 35mm f/2 is 'okay' but outside that it's the 35mm f/1.4L at about $1,100 or nothing.

Well, you could always get a Sigma...
posted by daniel_charms at 11:54 AM on August 10, 2010

Focal length has absolutely nothing to do with it, anyway.

It's how you use it, right?
posted by norm at 11:54 AM on August 10, 2010 [1 favorite]

I'm surprised no FPP was made of the previous post. Admittedly, the fact that people lie in their online dating profiles is not a new insight, but it's still a fascinating post nonetheless about the degrees of which people will lie.
posted by Weebot at 11:55 AM on August 10, 2010 [1 favorite]

Well, you could always get a Sigma...

Sorry to continue the camera-talk derail here, but I know about the 30mm f/1.4 and I thought that there were some pretty serious issues with focusing and quality control on it. I could be wrong...
posted by WinnipegDragon at 11:59 AM on August 10, 2010

Am I mistaken, or didn't the first Panasonic micro-4/3rds camera come with an option of a kit lens with a 1.4f?

It seems that neither of the current versions of that camera have that lens available as a kit option, which is a shame.
posted by paisley henosis at 11:59 AM on August 10, 2010

Where do you guys get your lens reviews? I could've sworn that dpreview had lens reviews at one point, but now they just have a flash "widget" showing chromatic aberration and other technical matters; there don't seem to be prose reviews or comparison shots anymore.
posted by sonic meat machine at 12:16 PM on August 10, 2010

Where do you guys get your lens reviews?

Photozone, The-Digital-Picture, SLRgear.
posted by daniel_charms at 12:19 PM on August 10, 2010 [8 favorites]

slrgear.com is pretty fantastic for reviews.
posted by WinnipegDragon at 12:19 PM on August 10, 2010

It's that 20mm I referenced above, paisley. It's a 1.7f. With that on it, the GF-1 is about the same size as a Cannon G10.
posted by bonehead at 12:20 PM on August 10, 2010

I think everyone is missing this: "And we also found similar numbers looking only at people who uploaded all three types of photos."

So, even when you pick people who have access to Camera Phone, Point'n'Shoot, and 'Spensive Lenses, that would cut down on your money angle and your quasi-professional photography angle, but we still see the trend towards SLRs and "good" photos.

But although the subject is the same in all three, the photographer is most likely different, so you're still not really controlling for that factor. I doubt you even could control for it, because although some good photographers probably do take a camera phone photo at some point, I don't think many people who don't care about photography get their hands on a high-end DSLR.
posted by burnmp3s at 12:21 PM on August 10, 2010

Weebot: I'm surprised no FPP was made of the previous post. Admittedly, the fact that people lie in their online dating profiles is not a new insight, but it's still a fascinating post nonetheless about the degrees of which people will lie.

What's weird is I could swear there was such a post, as I even recall jokey comments from that about how guys who were 6' tall would now need to list themselves as being 6'2", etc.
posted by hincandenza at 12:24 PM on August 10, 2010

Also, Ken Rockwell.
posted by togdon at 12:25 PM on August 10, 2010

Yes, 35mm on a crop DSLR is equivalent to 50mm on a full frame or film camera, but for the purposes of dating site profile pics, the 50mm is way better for portraits, IMO. A small amount of foreshortening can be attractive.
posted by rocket88 at 12:25 PM on August 10, 2010

Ok, bought the iPhone. Now which app will let me finally do sex?
posted by dr_dank at 12:25 PM on August 10, 2010 [1 favorite]

Oh, great. I searched for "iPhone pickup app", trying to make a snarky reply to dr_dank. Now I regret having eyes.
posted by griphus at 12:31 PM on August 10, 2010

Also, Ken Rockwell.

You are going to open up a can of worms with that one, Ken is not universally loved amongst other reviewers and enthusiasts.
posted by WinnipegDragon at 12:34 PM on August 10, 2010

So the take-away from this is to strip your EXIF data.
posted by geekyguy at 12:38 PM on August 10, 2010

So the take-away from this is to strip your EXIF data.

No, I think the take-way is to falsify it!

(This comment shot at ISO 25, f/0.8, 1/20000th shutter)
posted by WinnipegDragon at 12:40 PM on August 10, 2010 [1 favorite]

You'd want to go the other way. This comment was done at ISO 800 on a Kodak camera I won in a drawing at Wal-Mart. It is the sexiest comment you will read today. How did I do that? I am just that good.
posted by mikeh at 12:47 PM on August 10, 2010 [1 favorite]

It may not be scientific, but I love the fact that the Panasonic cameras take the "most attractive" photos. It helps me keep my resolve buy a GF1 instead go into crippling debt for an M9 (although an M9 would still be awesome).
posted by jnrussell at 12:56 PM on August 10, 2010

I'm surprised no FPP was made of the previous post.

I thought pretty seriously about it, but realized I've already posted two OKCupid stats-blog things, and it might start to look pretty astroturfy. That said, I'm glad this one got posted since it seems to have spawned a really interesting discussion. The conclusions out of the "Lies" piece were not all that unexpected, I thought.

Some of the analysis and resultant articles they've done have been better than others (this is one of my favorites so far), but in general they seem to be doing some neat things with their dataset. It's too bad that most of the things that make it so interesting also mean that they can't easily roll it into a big tarball and post a torrent link to it, so that everyone could have a go.

I'm sure an OKCupid MetaFilter-style infodump would turn up some really, really interesting stuff.
posted by Kadin2048 at 12:56 PM on August 10, 2010

Wow, I turn my back on this thread and all the camera nerds come out! Yay!

The question is whether the f/1.4 is 4 times better than the f/1.8, considering it costs 4 times as much.

The real question is, when the 50 1.4 is so good, and so cheap, why is Canon overcharging people on the 1.8 when it's nowhere near as good as the 1.4?

I love the 50 1.4--it is, I think, the best value in lenses around. These days though, I shoot almost exclusively with the 35 1.4, which is just dreamy. I have the 85 1.2, as well, but the focus by wire is annoying, and the fact that the objective element is flush with the mount makes me loath to swap it in when I'm off shooting. But when you throw enough light at it, the clarity is truly remarkable.
posted by Admiral Haddock at 12:58 PM on August 10, 2010

This comment was hand-written on the back of a napkin in black pencil, shot with a pinhole camera using a low-sensitivity film, exposed for like 24 hours, then processed, printed, scanned, OCR-ed, then piped straight into the comment box. All I had to do was click 'Post Comment'. If it isn't the sexiest comment you'll read today, I will shoot myself. With a camera.
posted by daniel_charms at 1:02 PM on August 10, 2010 [3 favorites]

What's that spike at 4am on the photos-by-time-taken graph? The noon peak I can understand -- camera's default time 12:00, I guess -- but what brand of camera has their factories set the time at first power-on to 04:00?
posted by sleepcrime at 1:08 PM on August 10, 2010

This comment was originally written by Ansel Adams, but I bought it at a garage sale and now I'm posting it.
posted by WinnipegDragon at 1:08 PM on August 10, 2010 [2 favorites]

FredMiranda.com has a pretty good section for lens reviews.

Mostly it's just user reviews, so you need to be willing to sift through the comments to piece together a more complete picture, but it's a decent resource that has helped tip my decision on several lens purchases.

FWIW they also have a buy/sell forum where I picked up my last two lenses (the buy/sell forum requires a registration fee if you want to make a transaction, but you can poke around without it; the registration fee paid for itself by a wide margin with the price on the two lenses I picked up there).
posted by Davenhill at 1:09 PM on August 10, 2010 [3 favorites]

FredMiranda rocks for used lenses. Fo sho.
posted by WinnipegDragon at 1:10 PM on August 10, 2010

I love the 50 1.4--it is, I think, the best value in lenses around.
I wonder if I have a bad copy; mine is soft as hell below f/2

while virtually nobody can justify the expense or practicality of an f/1.2.
Depends on what you mean by "justify". It's only money, and if you've got enough to eat and a house then it's time to visit Mr L. Bokeh at the 85mm house. But bring your patience, because it's not going to focus until the end of the week.
posted by bonaldi at 1:26 PM on August 10, 2010 [1 favorite]

Dear Askme: My iphone is evidently not working as intended. Do I need a newer version?

(Grumble: knowing the mean is only so useful. A full distribution would give much more info..)
posted by nat at 1:38 PM on August 10, 2010 [1 favorite]

Couldn't those who try a lot of partners also be the ones to try a lot of phones? And iPhone is the one to be trying.
posted by Obscure Reference at 1:40 PM on August 10, 2010

The real question is, when the 50 1.4 is so good, and so cheap,

The real question is how much camera equipment you need to buy for $400 to count as "so cheap". I kid!
posted by smackfu at 1:46 PM on August 10, 2010

blindcarboncopy: 1. Pairwise comparison studies are notoriously difficult to do right. You need a good set of guidelines and a set of judges who are trained to be consistent.

I don't think they were trying to do this entirely correctly for scientific purposes; in particular, daters would probably be less interested in some sort of platonic ideal of attractiveness than the completely subjective guidelines of entirely untrained judges, i.e. the people they are trying to meet, i.e. other OKC users.

2... it is prudent to assign overlapping sets (says, show the same pair of pictures to 5 or 10 judges).... Again, I think if OKC folks did this, they would have mentioned it.

As OKC said: This article aggregates 11.4 million opinions on what makes a great photo... We collected 552,000 example user pictures.

In other words, assuming photos were randomly presented, each was judged an average of 41.3 times. I'm not a good enough stats nerd to do the analysis on this, but it seems like a good start for sample size.
posted by Homeboy Trouble at 1:49 PM on August 10, 2010 [1 favorite]

If you plotted a site's ability to turn this article into a discussion of focal length and aperture against hotness, my science says you get a slope of +1, much better returns than just switching cameras. Well played, MeFi.
posted by carbide at 2:48 PM on August 10, 2010

The real question is how much camera equipment you need to buy for $400 to count as "so cheap". I kid!

Actually picked mine up used from Craigslist for about $250 a couple of years ago. But yes, I have spent far too much on camera equipment, particularly considering I should have been buying Panasonic micro 4/3 gear if I wanted to get laid instead of just getting 1) brokerer and 2) a crick in my neck and a sore shoulder.
posted by Admiral Haddock at 2:57 PM on August 10, 2010

@Homeboy Trouble:

Regarding your point #1, there is is no talk of a platonic ideal of attractiveness. But there is a widely-accepted understanding that different people are attracted to very different things that are easily discernible in photos (weight, height, hair styles, race, etc). When you mix all your judges together, you get the equivalent of the average playboy centerfold composition portrait - not useful to anyone in particular. I dare say, that having judges and guidelines is actually much more useful in this case.

Regarding your point #2: is somewhat irrelevant.

If OKC did absolute comparisons (rate this photo on scale of 1 - 10), and presented pairs of photos at a time to set "context", maybe the absolute ordering would work out from this data set.

Instead, if I read the study the same way that you do, it seems that they showed each photo rogether with some other photo 41.3 times. That seems like a lot, unless you look at the flip-side - this photo was not shown together with the other ~ 552,000 photos. Suddenly it doesn't seem all that significant anymore.

Of course showing every photo together with every other photo is not tenable. Instead, to gain statistical significance for this process, you could stop doing random sampling (show this photo with a random other photo 41 times) and move to a process that ensures a desired degree of overlap. For instance, pick a number of photo pairs (say, 100,000 of them) and ensure that each of those pairs is seen by at least 10 judges. Now you have 10 judgments of the same pair of photos to work with - if 9 of the judges said person A is hotter than person B, you have arrived at a majority point, and can throw away the outlier judgment. Get enough of these majority-rule judgment points, and you can start calibrating individual judges to give their votes more or less weight depending on how they agree with majority on these "gold sets". From here on, you can eventually arrive at an absolute ordering with the desired degree of precision. (Better hire a real statistician for the job though, not myself). This is just one way to go about getting interesting insights from this data, there are others as well. But overlap through random sampling in a large set like this gets you pretty much nowhere.
posted by blindcarboncopy at 3:01 PM on August 10, 2010 [1 favorite]

It is interesting that they are trying to do research, as opposed to just data mining. Given that, maybe they should stick to the data mining.
posted by smackfu at 3:39 PM on August 10, 2010

while virtually nobody can justify the expense or practicality of an f/1.2.
I have a 55mm f1.2 that I can stick on my DSLR, it's about 40 years old. A lot of fun, not really practical, but fun. Occasionally I get amazing results with it but the field of focus is paper thin. What it can do in low light situations is amazing but trying to grab a focus in low light is, well, difficult. It is sometimes perfect for portraits though.
posted by Locobot at 6:34 PM on August 10, 2010

So I pretty much go around getting people to buy this lens: Sigma 30mm F1.4.

It's not hard. It's a machine in low light.
posted by effugas at 7:04 PM on August 10, 2010

I probably have quite enough Canon camera bodies (anyone want to send me a 1Dm4 to complete my set? I promise to try to port my code to it), but I'm considering a Micro 4/3's camera just for the Noktor 50mm f/0.95 since the Zeiss f/0.7 isn't available in a PL mount.

(Digital sensors have improved to the point that we can shoot with only three candlepower at f/1.8 at ISO 1250 and still have it look acceptable. Take that Kubrick!)
posted by autopilot at 7:40 PM on August 10, 2010

When I switched from film to digital, one thing didn't change. The majority of my photos are still candid portaits in natural light (or with bounced flash when that isn't feasible) with a 50 1.4.
posted by [expletive deleted] at 8:25 PM on August 10, 2010

Also, since no one has said it yet, the conclusions of OKCupid are essentially the conventional wisdom of shooing people: use a decent camera with a prime lens in the standard or medium telephoto range; use depth of field to isolate your subject; avoid direct forward flash; shoot during the golden hour.

That last point was especially well illustrated by their data. It clearly shows spikes in the percieved attractiveness of photos strongly correlated with the hour after sunrise and hour before sunset.
posted by [expletive deleted] at 8:39 PM on August 10, 2010

female Iphoners are more promiscuous than their age-matched male counterparts, and to a greater degree than other brand owners. Explanations?

Rich girls are sluts.
posted by Jimmy Havok at 11:51 PM on August 10, 2010

But most importantly....never have a blurry photograph taken of yourself. Destroy all blurry photographs of yourself.
posted by a non e mouse at 12:00 AM on August 11, 2010

I buy my man only the finest, Canon L-series lenses. And I get laid all the time. ;-)
posted by Goofyy at 10:08 AM on August 11, 2010

This suggests that there is an exploitable market inefficiency here.

Find people with portraits taken at the wrong time, using a cheap camera or camera phone using flash.

These people are undervalued! Invest in them now! Contact me for a subscription to my newsletter if you want other hot tips.

[That's right - I didn't bother to read the rest of the thread. I've got investments to monitor. ]
posted by srboisvert at 12:37 PM on August 11, 2010 [1 favorite]

« Older The three annoying train monsters shown in the pos... | Ted Stevens has died in a plan... Newer »

This thread has been archived and is closed to new comments

More hot stats from OKCupid August 10, 2010 10:39 AM Subscribe

More hot stats from OKCupid
August 10, 2010 10:39 AM Subscribe