I've been happy with AdKnowledge's Cubics platform. I tried one other provider though, who came recommended and saw pretty dismal performance. So I think results may vary...
All good points, though your last one kind of negates the motivation I feel to implement your suggestions :)
I do mix up the order a bit to make sure people at the end don't get neglected. But people in the middle probably have fate set against them. Poor middle panelists :(
I'm very happy wasting the amount of time it took. ;)
I know, I know. I feel guilty already. I'm often torn between things I want to do because they seem exciting and controversial, and the effect it will have on people involved. I really hope I don't get any tear stained emails from panelists upset at this little project.
But to answer your question, I'm actually talking about the panels that were recommended to me by the system based on my choices. The best one, which I never would have found otherwise is this one:
http://schedule.sxsw.com/events/event_IAP5380
I'm super-excited!
And not to be too hand-wavey, but what if (I know this sounds crazy) there's actually a reciprocal effect here? What if there's a natural conceptual affinity between people who are attractive to each other? What if I only like the topics I like, because those are the ones the people I find attractive like? Or they like them because I do?
And not to be too hand-wavey, but what if (I know this sounds crazy) there's actually a reciprocal effect here? What if there's a natural conceptual affinity between people who are attractive to each other? What if I only like the topics I like, because those are the ones the people I find attractive like? Or they like them because I do?
That's an interesting idea. You'd need to do a similar experiment, swapping the topics for faces, and do a scatter plot to see if there's any covariance between topic-interest and attractiveness. I'm inclined to think that there isn't, but it'd be interesting to see the data.
You know, I was actually thinking of this. What SxSW really needs is just a tiny bit of NLP to do a tag cloud for the panels. I think that could be used to perform the kind of experiment you describe.
And I wouldn't bias such a thing by predicting the outcome, but can we agree that the panelists on topics we're interested in will be appealing to us in some way? Maybe not physically all the time, but they will be attractive to us.
I guess all I'm saying is that the inverse is true. Someone's nature, and interests do come through in their appearance (book cover judging rules!), especially in a set of people already pre-selected for common interests like the sxsw crowd.
Honestly, part of putting the word 'girls' in the title was to counter my last attempt at promoting my SxSW data. That one was very legitimate and professional. And in some ways, I think the data is even more valuable. But that one failed because the headline wasn't 'sexy' enough (IMO).
Whereas this one is getting some good buzz. Unfortunately, even using a demeaning term can sometimes get attention. I try not to be objectifying, but I fall short some times. Hopefully no one will be too offended.
If there's enough data, I will definitely post a recap. Stay tuned.
Heheh. I like to think of it as me passing along all the suckering pretty girls do to all of us. I'm just the messenger. (that sounds dirtier than I intended)
I forgot to mention one thing - some of the pictures were a little small, and so a potentially pretty girl was ignored by me because I wasn't sure. Not sure if there's a way to fix that, you might have to get creative with fetching alternate pictures for people.
