I think both. After I read the rest of the thread, I also became aware that the thing can even be more likened to a ZX Spectrum because of the Z80, and then my nickel fell also on the resolution and the color range, 320x256x16 means a memory of 160 k bytes of RAM. With its 21 k of working memory, that means 4 times the memory of the Speccy.
Indeed, the big thing to work upon currently (but it seems that this is not only a EU problem) is to instil in the people of the European Commission, that it is their duty to uphold the law, and if they want to achieve something where the law blocks them, the first thing is to start up a debate about why the law is that way, not try to change the law as fast as possible.
If you look at the different ways Switzerland and the USA handle these issues, it just comes down to one thing: discipline. And that is the issue with the NRA: they do not want to take the responsibility of instilling a sense of discipline and duty upon their members and gun owners in general.
But in Switzerland, all gun owners are part of the army, so they are taught professionally how to handle and store guns.
I have nothing against the availability of guns, but I think that every gun owner has a responsibility, which should be taught to him by professionals, e.g. the police or the army, or pensioned people of said services.
Also, a distinction can be made between hunting rifles which need to be manually reloaded, and multi-cartridge weapons.
The problem with generalised parallel programming is that the possible interactions between all threads are on the order of O(n!). That is the big problem. That is also why coordinating parts of the program around synchronising objects makes the program easier to understand, because in the same way that O(n!) grows very fast, it also shrinks very fast if n can be reduced.
Another problem I see is that the current core capacity is too large for running small problems.
Take a complex mathematical expression, which can be transformed into a syntax tree with values and mathematical operations. In the set of mathematical operations, there is a big difference between additive, multiplicative and transcendent operations. Either the transcendent operations also need to be transformed into a syntax tree which can be executed on many more, much simpler nodes, or the tree must be balanced into transcendent operations, with the other simpler parts all to be run on one node with the same cost as the transcendent node.
Sugar should be as expensive as possible. Your body does not need sugar. I think that everybody who argues for the price of sugar to go down should be forced to eat sugar until he dies.
Hm, this reminds me a little bit about non-cooperation, non-violence and peaceful resistance. These people seem too old to understand tech, and too young to understand how Ghandi obtained Indian independence.
But if one takes a look at Bayes' theorem and its implications, then it is necessary to take into account the possibility of false positives and negatives, i.e. saying about someone that he is a potential danger and it is not true, and letting someone true who is dangerous, but not flagged as such by the system.
I think that developing the software for this system is the least problem.
The big work is in obtaining a database about suspects and non-suspects, and then using this to flesh out the needed parameters and their probabilities to plug them into the rule of Bayes.
Slashdot Mirror
I suggest you reread 'Huckleberry Finn' from beginning to end.
Indeed, one painter that needs to be blocked entirely is probably Rubens. All that naked flesh!
I think both. After I read the rest of the thread, I also became aware that the thing can even be more likened to a ZX Spectrum because of the Z80, and then my nickel fell also on the resolution and the color range, 320x256x16 means a memory of 160 k bytes of RAM. With its 21 k of working memory, that means 4 times the memory of the Speccy.
Yes, you are right.
Indeed, the big thing to work upon currently (but it seems that this is not only a EU problem) is to instil in the people of the European Commission, that it is their duty to uphold the law, and if they want to achieve something where the law blocks them, the first thing is to start up a debate about why the law is that way, not try to change the law as fast as possible.
My ZX-Spectrum in 1984, with its 48 kB RAM was more powerful than this thing, and I had a resolution of 256x192 pixels.
If you look at the different ways Switzerland and the USA handle these issues, it just comes down to one thing: discipline. And that is the issue with the NRA: they do not want to take the responsibility of instilling a sense of discipline and duty upon their members and gun owners in general.
But in Switzerland, all gun owners are part of the army, so they are taught professionally how to handle and store guns.
I have nothing against the availability of guns, but I think that every gun owner has a responsibility, which should be taught to him by professionals, e.g. the police or the army, or pensioned people of said services.
Also, a distinction can be made between hunting rifles which need to be manually reloaded, and multi-cartridge weapons.
Yeah, where are the days if a motorcycle passed ones house, that the TV reception was jammed?
But you could tax the usage of robots.
I do not know about other countries, but here in Belgium there is a tax on power equipment, e.g. electro-motor based things, in function of the power.
I do not think that we should repentant for us wanting him to bleed.
You know that the Gauls in Asterix are only afraid of the sky falling on their head. And their favorite exclamation is 'By Toutatis!'.
Well, the real reason why they could not stop was because Robespierre was an idealist. After he was axed, the atrocities mostly stopped.
The rich must bleed.
Even better, heat a samovar with it.
The problem with generalised parallel programming is that the possible interactions between all threads are on the order of O(n!). That is the big problem. That is also why coordinating parts of the program around synchronising objects makes the program easier to understand, because in the same way that O(n!) grows very fast, it also shrinks very fast if n can be reduced.
Another problem I see is that the current core capacity is too large for running small problems.
Take a complex mathematical expression, which can be transformed into a syntax tree with values and mathematical operations. In the set of mathematical operations, there is a big difference between additive, multiplicative and transcendent operations. Either the transcendent operations also need to be transformed into a syntax tree which can be executed on many more, much simpler nodes, or the tree must be balanced into transcendent operations, with the other simpler parts all to be run on one node with the same cost as the transcendent node.
Serial dependencies are where the problem is at. There are things that you can never do before other things, and this depends on the algorithm.
Problem is, for transparent parallelism, you need to go 100% functional / immutable
Which comes down to re-educating programmers to learn to think about their algorithms.
One book which does this nicely, without going to deep into theoretical computer science is How to Design Programs
No editors with programming experience perhaps.
The basic definition of a loop is a GOTO to a previous address! All the rest is syntax and optimisation.
Sugar should be as expensive as possible. Your body does not need sugar. I think that everybody who argues for the price of sugar to go down should be forced to eat sugar until he dies.
Maybe a horse and some hounds.
So we are already sliding back, and we haven't even had an Empire and a Foundation.
Hm, this reminds me a little bit about non-cooperation, non-violence and peaceful resistance. These people seem too old to understand tech, and too young to understand how Ghandi obtained Indian independence.
Not to mention Mittal and and other multi-nationals.
Let the rich bleed!
Indeed.
But if one takes a look at Bayes' theorem and its implications, then it is necessary to take into account the possibility of false positives and negatives, i.e. saying about someone that he is a potential danger and it is not true, and letting someone true who is dangerous, but not flagged as such by the system.
I think that developing the software for this system is the least problem.
The big work is in obtaining a database about suspects and non-suspects, and then using this to flesh out the needed parameters and their probabilities to plug them into the rule of Bayes.
And brown, and some transvestites...