Data Love. Porn Data
January 30, 2014 3:20 AM   Subscribe

Sexualitics tries to contribute to human sexuality understanding through a Big Data approach. Studies (PDF), Datasets and Porngrams (maps the evolution of words frequencies in the titles of porn videos).
posted by motdiem2 (18 comments total) 12 users marked this as a favorite


 
This Data seems... fully functional.

Gotta love Porngrams though.
posted by Mezentian at 4:18 AM on January 30, 2014 [3 favorites]


Any innuendo and that's how much I like this.
posted by iamkimiam at 4:52 AM on January 30, 2014


Footjobs are making great strides recently eh?
posted by Potomac Avenue at 5:17 AM on January 30, 2014


Porngrams for the main metafilter sites : ask is still second, meta a distant third (irl excluded).
posted by motdiem2 at 5:17 AM on January 30, 2014


> Footjobs are making great strides recently eh ?

True but jobs don't show any sign of recovery.
posted by motdiem2 at 5:20 AM on January 30, 2014


Footjobs are making great strides recently eh?

That's what I hear.
posted by Mezentian at 5:24 AM on January 30, 2014


Those Porngrams are cool, and I was surprised by the results for the words I typed in -- it showed "amateur" as dropping, for example, when I would have guessed it was increasing. Interesting.
posted by Dip Flash at 6:14 AM on January 30, 2014


I would think a longer time frame than the last five years would show any real trends, though.
posted by Curious Artificer at 6:19 AM on January 30, 2014 [1 favorite]


Bigfoot, dracula
posted by jquinby at 6:43 AM on January 30, 2014


Peak puke.
posted by Combustible Edison Lighthouse at 6:48 AM on January 30, 2014


This is interesting, but without any real viewer data it doesn't seem like there's much that can be done with it.
posted by clockzero at 7:10 AM on January 30, 2014 [1 favorite]


And I still say this is more than Big Data. Surely it's BHD (Big Hard Data) or BBD (or Big Beautiful Data) or something else when we get into this kind of data wankery.
posted by nubs at 9:42 AM on January 30, 2014


'Interracial' takes off starting in 2008, peaks in 2010 and then declines a bit before leveling out.

It's still not a huge share of total word frequency, but I thought I was seeing it pop up more post election.
posted by postcommunism at 9:48 AM on January 30, 2014


After reading the paper linked, I have a slightly more nuanced view. If anyone is interested but doesn't want to read the paper, it looks like they basically want to see what they can determine about online pornography using tag data. The analysis of tag data from two websites, xnxx.com and xhamster.com, yielded some interesting statistics that describe how various categories are over- or underrepresented with respect to inferred demand, how categories correlate with nations, and some other things.

I think there are some serious limitations to the inferential value of these data, though. For one thing, there are multiple potential confounding factors that aren't addressed, e.g.: without knowing anything about the production of metadata or the strategies employed by content hosting websites to direct traffic, it might be hard to distinguish between "organic" patterns which might enable us to make unbiased inferences about what people actually want and think and feel, and "synthetic" patterns, which might reflect tactics employed by companies for various profit-oriented quasi-technical purposes.
posted by clockzero at 9:50 AM on January 30, 2014 [2 favorites]


This should be interesting, particularly as real data underlying puff pieces like the PornHub Porn trends we discussed a few weeks ago. But so far all I've found of value are the xHamster and Xnxx datasets. Where apparently 50 megabytes counts as "Big Data". The Porngram graphs look more like random numbers than real trends. I skimmed the analysis report and didn't see any particularly interesting findings; maybe I missed something?

I did a little side project once looking at words used in Craigslist man-for-man ads, in particular looking at words that show up frequently in sex ads but not ordinary English. No big surprises, you end up with a list that starts pic host cock stats suck. But it did make me appreciate the effort required to collect these datasets.
posted by Nelson at 9:54 AM on January 30, 2014 [1 favorite]


I think there are some serious limitations to the inferential value of these data,

This should be interesting

Quite obviously, we need a full-time research team getting into the data because there should be more of it (more sites, using searches and tags - which have a potential inherent bias as noted above - and likely other points as well) to see what is going on with porn viewing habits and usage. It's a project I would love to do.

I'm not being hurf-durf "let's watch porn" here either - I think there's some interesting stuff here that needs examination, and would be fascinating to pair up with some other socio-political trends around employment, income, crime (domestic violence and sexual assault rates in particular), birth rates, divorce rates, and so on.

The internet and these sites provides for a very interesting opportunity into doing some interesting research on porn use and habits and their impacts. Or what impacts them.
posted by nubs at 10:03 AM on January 30, 2014 [1 favorite]


Bigfoot vs. Yeti vs. Sasquatch
posted by jquinby at 10:06 AM on January 30, 2014


iamkimiam: "Any innuendo and that's how much I like this."

That's what I always say: love flies out the door when money comes innuendo.
posted by Chrysostom at 2:06 PM on January 30, 2014


« Older Do Not Over Inflate   |   Designer Drugs Done Dirt Cheap Newer »


This thread has been archived and is closed to new comments