No, I wouldn't really regard the Pi as a desktop replacement, but it can run one. I haven't really played around enough to see what happens when it starts swapping so I can't tell you I'm afraid. I would however imagine that this is heavily dependent on the read/write speeds of your SD card, which can be highly variable.

Well, mine was $35 plus I think $6 shipping, I plugged it into a spare phone charger I had lying around and it currently has no case, but I was planning on using a cardboard box. I haven't even had to buy a cable for it. So yeah I really honestly did spend a whopping $41, nothing like $85 (seriously is that even much money?).

Considering I was in the market for a HTPC and I was looking at AMD all-in-one fanless systems (~$150) with a nice case (~$150), I think I'm doing ok here.

Well mine is running XBMC very happily using raspbmc, and will output 1080p. It really wasn't that long ago that 256mb ram was decent. It isn't that snappy in a desktop environment, but really, we aren't talking about a particularly limited set of capabilities here.

I dunno, it sparked a pretty interesting flame war here on slashdot?

I would hardly call ~10% market share a market which has 'yielded'. Windows is absolutely the dominant platform on desktops, don't kid yourself. Servers are also less clean cut than you might think, Microsoft has a significant presence in the server market. Linux wins out in high performance scientific computing (clusters and the like) by an extremely wide margin, and high frequency trading for that matter. Not only that but 90% of mac's you see are running the office suite. Microsoft is hardly doing it tough, despite what you might like to think.

Actually, watching that right now I notice that Ballmer actually mumbles "it may sell very well, or not...". Now I've seen that video many times and laughed at it before, but really this is just a CEO doing what a CEO should do: talking up his business model. He made a reasonable point (at the time) that it did not appeal to business customers. things had to change before that was the case.

I know, its a funny video, I just noticed that he wasn't really as dismissive as people seem to think when you look back at it.

Dude, that sounds awesome. As far as I'm concerned, published papers are innovation, that is where the action takes place. Products themselves are always just an implementation of ideas that have been put out in the research for years previously - not to denigrate the huge work that goes into commercialising products, not at all. People just do not understand research, and if you are not in close contact with a lot of published research, I can see how it would be easy to think technology was just the products on the shelves.

I hope you the very best in your time at Microsoft, I would love to be in your shoes. I'm doing an honours research project for a computational neuroscience lab at the moment, and all I can say is these experiences are important, and fantastic! Good luck.

I'm sorry but Microsoft research do some really amazing stuff, sure, most of it is not commercialised, but you really have to hand it to a company that runs a huge research division at a loss in this day and age. The Kinect is a great example of something good that came out of this, they also do a lot of work in tablet research.

I recently stumbled upon some free Matlab libraries which offer optimised replacements for certain built-in functions which underperform natively, courtesy of Microsoft research. Publishing papers is innovation, fuck Apple and their trinkets.

He's a career criminal, and even back then he was disdained in his original german hacker scene.

This does not exempt him from due process. If the US DOJ fucked up their case and have no legal grounds to continue what they are doing, then the fact that this guy is dodgey has no relevance to the situation.

When I read the comment I was thinking along the same lines, I mean dealing with uncertainty can be addressed quite easily actually, you can essentially seed a simple learning algorithm with random data and evaluate the best options. You could do this with simulated annealing and genetic algorithms extremely easily. The problem is these are computationally complex approaches but more importantly, how do you evaluate a `better' choice in a completely or sufficiently general way? Nobody knows the answer to this, and from what we know of neuroscience it takes many years of training, using a computer that is many orders of magnitude more powerful and efficient than anything we know of, to achieve this task in humans. And still it is questionable how well many individuals do this.

Thanks for a sane viewpoint, this is about the only comment I came across with a rational description of HFT. Everyone just seems to convince themselves it must be bad because they don't see the point to ti...

Or it would just not connect.

The Android scripting environment already gives you Python, and a bash shell. This article is completely retarded.

Gravity is not a constant, it is a function of mass/inertia. you may carry on now ;)

Damn, that's some serious coin. Seems everyone in this dispute is being incredibly belligerent.