The photo on the right shows seven men, with red ribbons around their necks, who have changed the future of the mind. What these men have done is exciting but what their work will mean to understanding human mental processes in the future is even more exciting.
These men cooperated across international boundaries without knowing each other; they created a new high powered algorithm for accessing our minds.
Some credit must go to Netflix. In its effort to improve Netflix's ability to match films members had seen before, and rated, to unseen films they might like, Netflix offered $50,000 per quarter for improvements in their algorithm and $1,000,000 at the end of three years to anyone or team that could improve the Netflix algorithm by 10%.
The team in this photo (all men you will note, in a contest open to everyone on the planet [especially relevant to all you faculty at Harvard who fired Larry Summers]) succeeded in making a 10.6% improvement in the existing Netflix algorithm and winning the $ million.
For its money, Netflix ended up getting more than 51,000 brilliant people, world wide, to work for it, almost free. Entire classes in Bombay, Beijing, MIT, CIT and Todai worked months and years for Netflix. At a $100 per hour, Netflix got a quarter $ billion of work for its money.
I entered this contest when it began, because I had a clear vision of how to improve selection/matching algorithms with branching statistics (reduction of variance in small clusters). I immediately found that the file of data Netflix gave me, and everyone, to work with was more than a 100 megabits large. Much too large for my computer at the time to handle mathematically.
Why is this algorithm so important? Because it is an accurate map of the mental structures of millions of people. It is the raw data for future research on mental constructs. The fact that one person loves The Trouble with Mary, The Big Lubowski, Speed, Man on a Wire and 30 Rock TV in such a measurable way that we can make accurate predictions from that structure means we can find out the nature of that structure with empirical testing.
Let me give you an example. I once did a test on 15 strangers who came to my house. As they entered the door they were given a letter label and a questionnaire that had them rate everyone else without talking to anyone. At the end of a subsequent discussion they were given new labels and a new questionnaire. All measures on the questionnaire were spacial. 'Which two people are more alike A,B,C ...A,B,D' etc.
The test was designed to use spacial measurement to find out how people made judgments of strangers and then how they changed those judgments after hearing the people talk for ten minutes.
It turns out when people first meet strangers, the first judgment they make appears to be masculine/femine. It was sometimes easy for subjects to discern, but in a few cases not so easy. Both men and women subjects used this same criteria.
How do I know this was the actual criteria these people used in their minds? Because I asked unrelated questions on the second questionnaire that could be matched to the spatial dimensions on the first questionnaire.
Crude and animalistic as it might sound, I found that men were looking for physical threats as they met strangers and women were looking for female competitors. Very Darwinian. The whole point of the research was to have empirical data and empirical tests to corroborate whatever I found.
Vast reliable empirical research, of a similar nature, will now be possible in the future, thanks to these seven Netflix winners.
Recent Comments