IMO the draw of multiplayer games are primarily the superiority of human intelligence in relation to the artificial intelligence we enjoy in games today. AI continues to be a difficult problem for game developers and no matter how hard they try, they probably won't ever be able to create ingame AI driven opponents with anywhere near the level of creativity/ingenuity the common gamer holds. Who wants to demolish seven computer zerg opponents when you can ambush that dude who keeps zerg rushing the hell out of your siege tanks with a swarm of battlecruisers... as you explain to him how much his life sucks.
I agree that the idea has potential to be incredibly useful. But this is highly contingent on the quality and consistency of the "community tagging" that this services hopes to exploit. Working in research labs at a reputable engineering research university I find that the most difficult thing is formulating your problem into discernable parts. Once you formulate the problem into 'academic speak' so to say, finding the solution (or finding if there already is a solution) is straight forward. That's kind of the whole point of 'academic speak,' to give consistency to the way all this research is described so that there is a way to connect ideas between different projects. Also, what kind of credibility will these community tags hold? Is it really going to be that much more informative than the latest textbooks? If I'm into programming parallel distributed robotic systems, shouldn't I just read the latest releases from conferences like the AAAI?
Tangent note, from the perspective of a computer science major, I would like to see this embracing of descriptive community tags applied to something like Google's code search tool. It'd be great to say something like "I want a script that takes as input X, and outputs Y." This is because the programming community benefits the most (at least the most quickly) from collaboration, ie Rosetta Codes mantra of "don't reinvent the wheel."
The ~8 spe's each have the capability of processing ~256 different values from memory at the same time. All 256 processes must execute the exact same instruction on each of the various pieces of data. This means that unless every computer algorithm can be divided into executable sets of 256 pieces of data without any remainders you will never be fully utilizing the processor. This reminds me of this [http://science.slashdot.org/article.pl?sid=06/06/ 23/185217] article on slashdot which points to this [http://www.newscientisttech.com/article/dn9394-at omic-simulation-most-intensive-computer-program-ev er.html] article online about the most computationally intensive program ever written. So please everyone stop running algebraic calculations on gpu and bus performance, it's an inherent drawback/rule of parallel processors. You're just confusing everyone.
Slashdot Mirror
IMO the draw of multiplayer games are primarily the superiority of human intelligence in relation to the artificial intelligence we enjoy in games today. AI continues to be a difficult problem for game developers and no matter how hard they try, they probably won't ever be able to create ingame AI driven opponents with anywhere near the level of creativity/ingenuity the common gamer holds. Who wants to demolish seven computer zerg opponents when you can ambush that dude who keeps zerg rushing the hell out of your siege tanks with a swarm of battlecruisers... as you explain to him how much his life sucks.
I agree that the idea has potential to be incredibly useful. But this is highly contingent on the quality and consistency of the "community tagging" that this services hopes to exploit. Working in research labs at a reputable engineering research university I find that the most difficult thing is formulating your problem into discernable parts. Once you formulate the problem into 'academic speak' so to say, finding the solution (or finding if there already is a solution) is straight forward. That's kind of the whole point of 'academic speak,' to give consistency to the way all this research is described so that there is a way to connect ideas between different projects. Also, what kind of credibility will these community tags hold? Is it really going to be that much more informative than the latest textbooks? If I'm into programming parallel distributed robotic systems, shouldn't I just read the latest releases from conferences like the AAAI? Tangent note, from the perspective of a computer science major, I would like to see this embracing of descriptive community tags applied to something like Google's code search tool. It'd be great to say something like "I want a script that takes as input X, and outputs Y." This is because the programming community benefits the most (at least the most quickly) from collaboration, ie Rosetta Codes mantra of "don't reinvent the wheel."
The ~8 spe's each have the capability of processing ~256 different values from memory at the same time. All 256 processes must execute the exact same instruction on each of the various pieces of data. This means that unless every computer algorithm can be divided into executable sets of 256 pieces of data without any remainders you will never be fully utilizing the processor. This reminds me of this [http://science.slashdot.org/article.pl?sid=06/06/ 23/185217] article on slashdot which points to this [http://www.newscientisttech.com/article/dn9394-at omic-simulation-most-intensive-computer-program-ev er.html] article online about the most computationally intensive program ever written. So please everyone stop running algebraic calculations on gpu and bus performance, it's an inherent drawback/rule of parallel processors. You're just confusing everyone.