My previous post about AMTSense made quite the splash. When Tim O’Reilly linked to me I noticed his next entry The System is Smarter than the Raters. I’ve quoted what caught my eye…
Google hires temp workers to evaluate search results. Often when the raters and the system disagree, there’s a bug, which Google fixes by tweaking the algorithms. But sometimes, the system is smarter than the raters.
I’d be interested to know how they go about hiring these people and collecting the results. I wonder if the sheer logistics of finding, hiring and paying these people limits the amount of data they can gather.
Let me get to the point - If Google is using human intelligence to “train” the search engine. Wouldn’t MORE human intelligence result in a better search engine? Couldn’t they tackle more of these searching glitches? What about regional differences? They could use AMT to sample data from all over the world. If they built a tool that issues mass amounts of HIT’s to AMT and then compiles and anaylizes the results.. couldn’t you call this a automated training system?
It’s funny how people only focus on scalability of their databases/web applications yet not their workforce. Are you seriously thinking about integrating AMT into your business? Give me a call and we can brainstorm about it.
