The Extra Cool Group! (of people Michael is experimenting on) discussion

More specifically, will book recommendation websites ever be as good as humans at recommending good books? I'm referring to http://bookish.com/ and the newly Goodreads-acquired http://www.discovereads.com/ as well as Amazon's recommendations.

Do you feel like there is a need for them?

I don't feel there's a need for them, but I do feel they often speed up the process of finding things I'd be interested in. Netflix is usually pretty accurate about what I'll think of movies. But, I must say, Amazon is usually pretty far off. That might be because my book taste is harder to pinpoint than my movie taste...anyway, that's off topic.

I don't think they're needed as long as we have people to talk to about books. I expect to always pay more attention to my friends' reviews than I do some auto-recomendation robot.

The robot might be better than friends at telling me what I'd like, but I doubt they'd be as good at getting me EXCITED about reading something.

I know you didn't say they'd be robots. But, robots are way cooler than websites, so I'd rather have a recommendation robot.

I've found this site of some use. You input an author and it displays a cloud of authors who are similar.

http://www.literature-map.com/

Another site based on books rather than authors is:

http://www.whatshouldireadnext.com/?

The literature-map site is interesting & fun. Thanks.

Computers will probably be able to get pretty good at recommending new reads, but the search will have to be personalized & based on a large set of good reviews with a lot of metadata including similar readers' reviews.

GR might have the data since it has ratings for 1700 or so books that I listed here & it knows who my friends are. Even with all that information, there are a lot problems figuring it all out though. I'd probably have to be able to tweak any algorithms myself to some extent based on trial & error.

Michael: interesting, but I've had a completely different experience. I find Netflix's recommendations to be downright AWFUL. Amazon's is pretty bad too. The only similar site that I find surprisingly good is Pandora (for songs). I tried Pandora a while ago and recently tried it again and was surprised at how much better it had gotten. They probably changed their algorithm a lot or had more metadata than before. Of course, there are still things I don't like, but at least it's not recommending me the same 4 artists I already know about.

Yeah, I've had good experiences with Pandora and with last.fm. I'd never been sure if that was because they were good at it, or because my taste in music is incredibly broad and not at all picky. Good to know it works for others as well!

Netflix is usually within a star of what I'll end up giving movies. Every now and then, it'll be totally off, but that's not very often. What impresses me is it is usually right about whether I'd put a movie in the so-bad-it's-good category and give it a high rating, or if I'd give it a -so-bad-its-really-just-bad rating like one star. I'm impressed their algorithm can differentiate with that.

I'd say Netflix's algorithm is so bad it's good.

Michael said: "Netflix is usually pretty accurate about what I'll think of movies. But, I must say, Amazon is usually pretty far off."

I don't have Netflix, but Amazon is usually a bit crap for recommendations. Half the time they're recommending books I already have, usually bought from them! I suppose it's better that way, as it prompts me to buy less books, but most of the time I just ignore their suggestions completely.

Computers already do beat humans: The IBM system which won the natural language Jeopardy game shows the shape of things to come. They have been beating humans at chess for a long time and they do number crunching as no human can approach in real time.

We will see and exponential increase in the use of AI, especially in my field, Medicine. Big changes coming.

My question—Will computers ever become sentient? Will they evolve on their own? These have been explored in depth in SF and cyberpunk. "Resistence is Futile"

I really felt the buzzer was tilted to Watson's favor on Jeopardy, but was amazed by how well it answered. I loved Jennings' final comment.

I didnt actually see it Jim, just read about it. What was the final comment?

Jennings said something about looking forward to serving our new lords & masters.

I both read about it & watched it. Both Jennings & Rutter were super on the buzzer against other humans, but in that contest they were often frustrated because they couldn't buzz in fast enough. I think the issue was that Watson got the clue as a text as soon as it was revealed & then got a go-ahead signal when the buzzers were freed up. The humans have to wait until Alex stops talking - much less definite timing. I don't know the stats, but Watson got to answer first most times. It also got almost all of the daily doubles. It had an uncanny knack for being able to find them.

Only if we keep fanatically advancing them. I read a few years back about the project they are working on to combine something wth computer cells with human brain cells to improve a computer. With this sort of mad science The terminator really wouldn't be that far off lol