Slashdot Mirror

Yeah - I sussed out the prop thing later, and actually found a neat animation, using gears, and a moving gear track as an analogy.

http://sifter.org/~simon/journal/20101107.h.html

I wonder if a mill driven version is feasible, provided you always operate in a zone with sufficient apparent wind. I know on sail/wing craft, that the direct downwind is only impossible due to stalling of the foil, but if a turbine was used, and you have already exceeded windspeed, I can't see any reason you can't go dead down-wind (given the mill can generate enough power to overcome drag, at say 1.2x apparent wind)

This is a very interesting topic - regardless of where it's floated, it generates the same set of arguments every time, amoungst people with the same varying levels of education and intuition

Cheers

In this case, it's wheels driving a propeller, but I think the other way could work too. See this animation for a good intuitive reason as to why it works

http://sifter.org/~simon/journal/20101107.h.html

It's the movement of the wind relative to the ground that's important (to avoid the perpetual motion stuff), and the relative forces required in air vs on-ground. The animation clearly shows how a simple geared machine can move in the opposite direction to the motive force (and faster), aor move in the same direction. In this case, the prop is the inneficient mechanical interface to the wind,

I still prefer this: http://sifter.org/~simon/journal/20081009.html (if I don't say so myself...)

I believe that Netflix is still using Cinematch. You could look into movielens. It's from the GroupLens group at U Minn.

[E]ven though both probably use a pretty simple algorithm compared to Nextflix.

You do know that Netflix said on the outset "You're competing with 15 years of really smart people banging away at the problem." and it was beat in less than week.

That's not to meant as a knock against Netflix's engineers, but more about that they didn't really build a state of the art recommender system. Simple SVD (aka latent semantic indexing outperfomed them as well.) They did something a bit more than straight up kNN clustering, but that was pretty much it.

Thoughts on this?

http://sifter.org/~simon/AfterLife/index.html

Wonderful online scifi book. The author, Simon Funk, copes with many such questions with incredible imagination and vivid narration.

Lubos Motl is a "real" physicist? No true scotsman fallacy, anyone? Hell, do your research: Garrett Lisi has a Ph., D. in physics! http://sifter.org/~aglisi/Physics/CV.html He clearly isn't a "real" physicist, whatever that is supposed to mean (perhaps if Garrett spent more time performing intellectual masturbation and doing things like "debunking" global warming while comparing himself to Richard Feynman, then Garret would become a "real" physicist).

If anyone is interested in Garrett's reply to Motl's ad hominems, here it is: http://www.physicsforums.com/showpost.php?p=1496330&postcount=14

This is actually a relatively interesting approach, but I'm skeptical about the predictions of 20 new particles Garrett's theory makes. Ultimately it's up to experiment, as opposed to (say) a string of insults a maundering Harvard professor makes, that determines whether or not this theory is false or not.

None of the top teams entered the contest to win the money. The money is nice, don't get me wrong, but it's not the main motivation. The main motivation is the the data set. 2.8 million queries, with relevance judgements. (e.g. "This was good. Give me more." "This was bad. Give me less") It's a great data set for information retrieval researchers, which is exactly what all the top teams are.

We know roughly what the Bellkore team is doing. And we know roughly what all the top teams are doing. Simon Funk uses Singular Value Decomposition. Yi Zhang is using Bayesian hierarchical models and Expectation-Maximization All the top teams know what everyone is doing. It's not a secret. They publish what they're doing.

To make a long rant short: As someone who is a life-long "false positive", I am annoyed by the way things are going to say the least.

Check Atmo for a quick download link that may bypass Adobe's registration page. Atmosphere is a generic 3d world builder plus standalone browswer and web browser plugin, and it's free (builder too) so you can see if it's good enough for your purposes without risking any ca$h. It's not directly intended for what you're doing, but I've seen some pretty cool apartments and homes done in it so it should be possible.

Optimistic Muffin marketting.

Heh heh.

-Brandyn