Yeah, the Wine guys were muttering about using their implementation of DirectX 10 on Windows XP, so gamers wouldn't have to upgrade to Vista to play the latest games. I don't know what became of that.
A 'Wine for Windows' or 'ReactOS for Windows' distribution replacing certain Windows DLLs with their free equivalents would be a fun toy and a useful way to get more testing for these two projects. I'd install it at once... uh, on a spare machine...
Programming in C is no more "Thinking" than programming in Java. It's just thinking about lower level crap.
Exactly! And this lower level crap is something you should learn at least a bit of during your CS course. I program in higher-level languages too but I wouldn't hire someone who couldn't understand what an address is or why byte-aligned accesses might be slower than word-aligned.
Java is much, much closer to Java bytecode than C is to assembler. You can take a.jar file and decompile it to get readable source that almost exactly matches the original (modulo private variable names). That's not true of C.
The point is that with C you can write a program and have some mental model of what the hardware is doing underneath. You learn what an address is, you can visualize the values being loaded into registers (even if an optimizing compiler will do it a bit different to how you expect), and so on. With assembler that link to the hardware is even stronger. With Java it is missing.
When I studied CS we learned a little bit of assembler, just enough to have basic familiarity with what a computer does under the hood. When we learned Java we also studied Java bytecode and how it is executed by the JVM.
It's not clear-cut that IRV is better than plurality. One property we would like from voting systems is monotonicity. That is, if you vote for a candidate that makes them more likely to win, or at least it doesn't make them more likely to lose. But in IRV and STV, giving more votes to a candidate can cause them to lose!
At least plurality doesn't have that problem. You vote for someone, they get more votes, they are more likely to win. Approval voting is also monotone and simpler than IRV.
(There are some simulations of this linked from the Wikipedia page 'Instant-runoff voting controversies'.)
But if you do vote that way, then it's possible that a dark horse candidate that you really, really don't want could come back and bite you in the ass because, well, you really did like the other guy better.
Can you give an example?
When I said 'Range voting tends to punish you if you try to play fair and share your votes among candidates' I meant that if you give varying amounts of vote to different candidates (rather than all-or-nothing) then you are wasting some of your vote, and you would be better off giving all your votes to one candidate (if you have a fixed total number of votes to give) or giving some candidates 10/10 and others 0/10 (if you can independently award marks out of ten to each candidate).
In skating competitions each judge gives marks out of 6. This is a kind of range voting. It works well because the judges are honest and don't have any particular interest in promoting one skater above another. They can just express their opinions and let the system decide. However an election isn't like that. I want to maximize my say in the results, and so does every other voter.
If everybody decides to _try_ to game it, they have to game it together, and at that point, they're not really able to game it, since their power is limited to the high amount in the range.
It's not clear what you mean here. But if you mean that everybody would end up voting either 10/10 or 0/10 for each candidate and then they wouldn't be able to game the system any further, then yes, that's exactly what would happen.
Thanks for the correction - I got misled by another poster, or more likely misunderstood his or her comment. Yes, I guess range voting does degrade to approval voting, because you'd give either 0 or 10 marks to each candidate.
But just because all rational voters would cast what amounts to an approval ballot in a range-voting election does not mean its busted.
Are you sure this is true? In approval voting, you can cast as many or as few votes as you want, but only one per candidate. In range voting you have a fixed number of votes to distribute, and if one candidate gets more all others must get less. I think range voting degenerates to plurality (you end up giving all your votes to one person), not to approval.
P.S. yes approval voting is quite a good system, and very simple.
Let's say there are three people running, A, B, and C.
You don't mind B(6) and C(10), but you hate A(0).
They don't mind A(10) and B(6), but they hate C(0).
Suppose you know roughly what 'they' are going to do. You're an individual voter and 'they' are a fairly large number of voters, as in a real election. You know that most of other people's votes will be divided between A and B. You know that C doesn't really have a hope of winning.
Given that, your choice is between B and A. Now, why would you cast 10/16 of a vote for C and 6/16 for B? You know that C can't possibly win. Why waste some of your vote and have only 6/16 of a voice? Why not use your full voting power where it matters and cast all 16 votes for the candidate who you quite like and who has a chance of winning?
That election couldn't have been done with binary voting, and everybody wins.
Yes, if everybody agrees to be honest and vote according to their true preferences, then you get a good result. But people aren't like that! Why should I vote according to my real views on the candidates if that means I'm _less_ likely to get the result I want? Surely any sensible person would vote tactically, as in the example above. And if you know that other people are going to be voting tactically and not entering their real preferences, there's still less incentive to say what you want.
Any range voting system rapidly collapses into first-past-the-post, once voters realize that the best way to influence the result in their favour is to give all their votes to one candidate. You need to find a system that not only _lets_ voters express their different preferences among the candidates, but _rewards_ them for doing that. Range voting tends to punish you if you try to play fair and share your votes among candidates.
Let's say there are three people running, A, B, and C.
You don't mind B(6) and C(10), but you hate A(0).
They don't mind A(10) and B(6), but they hate C(0).
Suppose you know roughly what 'they' are going to do. You're an individual voter and 'they' are a fairly large number of voters, as in a real election. You know that most of other people's votes will be divided between A and B. You know that C doesn't really have a hope of winning.
Given that, your choice is between B and A. Now, why would you cast 10/16 of a vote for C and 6/16 for B? You know that C can't possibly win. Why waste some of your vote and have only 6/16 of a voice? Why not use your full voting power where it matters and cast all 16 votes for the candidate who you quite like and who has a chance of winning?
That election couldn't have been done with binary voting, and everybody wins.
Yes, if everybody agrees to be honest and vote according to their true preferences, then you get a good result. But people aren't like that! Why should I vote according to my real views on the candidates if that means I'm _less_ likely to get the result I want? Surely any sensible person would vote tactically, as in the example above. And if you know that other people are going to be voting tactically and not entering their real preferences, there's still less incentive to say what you want.
Any range voting system rapidly collapses into first-past-the-post, once voters realize that the best way to influence the result in their favour is to give all their votes to one candidate. You need to find a system that not only _lets_ voters express their different preferences among the candidates, but _rewards_ them for doing that. Range voting tends to punish you if you try to play fair and share your votes among candidates.
There will be artifacts that will allow "dishonest" voters to game the system.
Are you sure about that? Of course first you need to define what you mean by gaming the system. One definition is that the system can give voters an incentive to vote in some way that doesn't reflect their true preferences - tactical voting, rather than just expressing exactly what they want. Lots of voting systems suffer from this weakness. But I believe there are some that provably do not, such as Condorcet - in Condorcet voting there is never any advantage to an individual voter or group of voters being 'tactical' and putting down some strange vote rather than just listing their real preferences in order. (Well, it might depend on what chain-resolution is applied in the result of a tie.) Approval voting also has this property IIRC. Yes, Range Voting does have the problem.
maintaining a final election between two people is probably a good thing (for this and for the more important reason that we get to focus on the candidates more during the final cycle.
Ah, and this gets into a different discussion from pure voting theory, where you assume that everyone knows which candidate they prefer at the start and can vote all at once. Yes, for real elections it might help voters make a more informed choice if there is a final showdown between the top two candidates.
Yes - but the choice tends to be either a plurality vote by constituency, which has many disadvantages but at least makes it possible to vote for an individual candidate, or a party-list system, which means voters have very little choice over which candidates from a party get elected. Party lists arguably increase the power of party bosses at the expense of the electorate. Instead of trying to appeal to voters, it's a better bet to ingratiate yourself with the party hierarchy so they put you higher up the list.
Often when people discuss voting systems there's an implicit assumption that they are talking about constituency voting, where a constituency has different individual candidates of which one or more (usually exactly one) must be elected.
Note that proportional systems are often fudged with 5% rules (as in Germany) to keep out small parties.
What is a 'legitimate' computer program? There are many people who make a living as consultants paid to test how hard it is to break into a company's systems. They might well need to use even the most dastardly and underhanded 'hacking tool' to do their work. Indeed the police and security services also use programs that help them get unauthorized access to computers. What grounds are there for criminalizing any computer program?
The times I've had the misfortune to use the Microsoft website, it hasn't been a pleasant experience. Broken links are the norm (they seem to change all their URLs every few months as a matter of principle), you have to click through lots of crap to get what you want, and the search function doesn't work. If they get the same people to redo it in Silverlight, it will do more than anything else could to put people off the platform.
Silverlight as an alternative to Flash? Yes please. It may not be a completely open standard in the same way as HTML or PNG but it is well documented and has a free implementation (Moonlight). If it came from anyone other than Microsoft then we'd all be rooting for it.
Those people who judge what program to use by measuring the 'evil factor' of the companies supporting it, rather than the properties of the program itself, will be put off Silverlight by the fact it comes out of Redmond. Yet Adobe is just as evil in many ways, pushing Flash as a closed, binary-only standard that has to be reverse-engineered; imposing DRM on users and jailing those who try to circumvent it (remember Sklyarov?). If Microsoft, for once, attempt to do the right thing (even if for the wrong reasons), I say we give them a chance.
I don't think research is a pissing contest to see who has the most powerful computer.
The metric for 'most powerful' also seems flawed. If you just count operations per second, then a large enough cluster of Linux PCs will appear 'more powerful' than any supercomputer, even if they are connected by UUCP over 2400 baud modems. Yet the supercomputer is much faster at most difficult computational tasks because it has faster connections between the nodes. The Linux cluster would only outperform it for drawing a large picture of the Mandelbrot set or other 'embarrassingly parallel' problems. This isn't an academic distinction; even gigabit Ethernet is much slower (higher latency) than the links used in a real supercomputer.
top500.org do categorize each supercomputer as 'cluster' or whatever but I think their ranking is just on raw operations per second.
The thought that Jimmy Wales, cofounder of Wikipedia could have an open site without abuse is laughable. You operate under the sham of an open community, yet exclude those outside a very narrow political agenda. Your a fraud, using open source principals as a smokescreen that presents your personal world-view set as fact to the world....
Save up for just a little longer and buy something for at least $450 that runs Windows Vista, or get the ASUS Eee PC 4G laptop.
The Eee PC is another Linux-based system. From reading the review, I don't think the author is biased against Linux, he just thought this particular distribution and this slow hardware wasn't much fun to use.
This could be one reason why running perl5 in the Parrot VM never really took off. Perl 6 will be a bit better specified, I hope, and so easier to write on top of a specified virtual machine.
Of course I've heard of those things - what I meant was that (AFAIK) there are absolutely zero plans from the Python community to adopt Pirate as their runtime environment or even one of the supported runtimes along with CPython and Jython. I'd be delighted if it all works out and the different scripting languages can find a way to play well together (after all, the semantics are similar enough for most useful libraries to be used across languages) but I am not optimistic.
Slashdot Mirror
Yeah, the Wine guys were muttering about using their implementation of DirectX 10 on Windows XP, so gamers wouldn't have to upgrade to Vista to play the latest games. I don't know what became of that.
A 'Wine for Windows' or 'ReactOS for Windows' distribution replacing certain Windows DLLs with their free equivalents would be a fun toy and a useful way to get more testing for these two projects. I'd install it at once... uh, on a spare machine...
Yeah, look inside the jar file and you'll see the class files which are Java bytecode.
Java is much, much closer to Java bytecode than C is to assembler. You can take a .jar file and decompile it to get readable source that almost exactly matches the original (modulo private variable names). That's not true of C.
The point is that with C you can write a program and have some mental model of what the hardware is doing underneath. You learn what an address is, you can visualize the values being loaded into registers (even if an optimizing compiler will do it a bit different to how you expect), and so on. With assembler that link to the hardware is even stronger. With Java it is missing.
When I studied CS we learned a little bit of assembler, just enough to have basic familiarity with what a computer does under the hood. When we learned Java we also studied Java bytecode and how it is executed by the JVM.
It's the thought that counts.
It's not clear-cut that IRV is better than plurality. One property we would like from voting systems is monotonicity. That is, if you vote for a candidate that makes them more likely to win, or at least it doesn't make them more likely to lose. But in IRV and STV, giving more votes to a candidate can cause them to lose!
At least plurality doesn't have that problem. You vote for someone, they get more votes, they are more likely to win. Approval voting is also monotone and simpler than IRV.
(There are some simulations of this linked from the Wikipedia page 'Instant-runoff voting controversies'.)
When I said 'Range voting tends to punish you if you try to play fair and share your votes among candidates' I meant that if you give varying amounts of vote to different candidates (rather than all-or-nothing) then you are wasting some of your vote, and you would be better off giving all your votes to one candidate (if you have a fixed total number of votes to give) or giving some candidates 10/10 and others 0/10 (if you can independently award marks out of ten to each candidate).
In skating competitions each judge gives marks out of 6. This is a kind of range voting. It works well because the judges are honest and don't have any particular interest in promoting one skater above another. They can just express their opinions and let the system decide. However an election isn't like that. I want to maximize my say in the results, and so does every other voter.It's not clear what you mean here. But if you mean that everybody would end up voting either 10/10 or 0/10 for each candidate and then they wouldn't be able to game the system any further, then yes, that's exactly what would happen.
Thanks for the correction - I got misled by another poster, or more likely misunderstood his or her comment. Yes, I guess range voting does degrade to approval voting, because you'd give either 0 or 10 marks to each candidate.
P.S. yes approval voting is quite a good system, and very simple.
Given that, your choice is between B and A. Now, why would you cast 10/16 of a vote for C and 6/16 for B? You know that C can't possibly win. Why waste some of your vote and have only 6/16 of a voice? Why not use your full voting power where it matters and cast all 16 votes for the candidate who you quite like and who has a chance of winning?
Yes, if everybody agrees to be honest and vote according to their true preferences, then you get a good result. But people aren't like that! Why should I vote according to my real views on the candidates if that means I'm _less_ likely to get the result I want? Surely any sensible person would vote tactically, as in the example above. And if you know that other people are going to be voting tactically and not entering their real preferences, there's still less incentive to say what you want.Any range voting system rapidly collapses into first-past-the-post, once voters realize that the best way to influence the result in their favour is to give all their votes to one candidate. You need to find a system that not only _lets_ voters express their different preferences among the candidates, but _rewards_ them for doing that. Range voting tends to punish you if you try to play fair and share your votes among candidates.
Yes - but the choice tends to be either a plurality vote by constituency, which has many disadvantages but at least makes it possible to vote for an individual candidate, or a party-list system, which means voters have very little choice over which candidates from a party get elected. Party lists arguably increase the power of party bosses at the expense of the electorate. Instead of trying to appeal to voters, it's a better bet to ingratiate yourself with the party hierarchy so they put you higher up the list.
Often when people discuss voting systems there's an implicit assumption that they are talking about constituency voting, where a constituency has different individual candidates of which one or more (usually exactly one) must be elected.
Note that proportional systems are often fudged with 5% rules (as in Germany) to keep out small parties.
What is a 'legitimate' computer program? There are many people who make a living as consultants paid to test how hard it is to break into a company's systems. They might well need to use even the most dastardly and underhanded 'hacking tool' to do their work. Indeed the police and security services also use programs that help them get unauthorized access to computers. What grounds are there for criminalizing any computer program?
The times I've had the misfortune to use the Microsoft website, it hasn't been a pleasant experience. Broken links are the norm (they seem to change all their URLs every few months as a matter of principle), you have to click through lots of crap to get what you want, and the search function doesn't work. If they get the same people to redo it in Silverlight, it will do more than anything else could to put people off the platform.
Silverlight as an alternative to Flash? Yes please. It may not be a completely open standard in the same way as HTML or PNG but it is well documented and has a free implementation (Moonlight). If it came from anyone other than Microsoft then we'd all be rooting for it.
Those people who judge what program to use by measuring the 'evil factor' of the companies supporting it, rather than the properties of the program itself, will be put off Silverlight by the fact it comes out of Redmond. Yet Adobe is just as evil in many ways, pushing Flash as a closed, binary-only standard that has to be reverse-engineered; imposing DRM on users and jailing those who try to circumvent it (remember Sklyarov?). If Microsoft, for once, attempt to do the right thing (even if for the wrong reasons), I say we give them a chance.
Don't forget to have a good look at the Interface Hall of Shame for examples of what not to do.
I don't think research is a pissing contest to see who has the most powerful computer.
The metric for 'most powerful' also seems flawed. If you just count operations per second, then a large enough cluster of Linux PCs will appear 'more powerful' than any supercomputer, even if they are connected by UUCP over 2400 baud modems. Yet the supercomputer is much faster at most difficult computational tasks because it has faster connections between the nodes. The Linux cluster would only outperform it for drawing a large picture of the Mandelbrot set or other 'embarrassingly parallel' problems. This isn't an academic distinction; even gigabit Ethernet is much slower (higher latency) than the links used in a real supercomputer.
top500.org do categorize each supercomputer as 'cluster' or whatever but I think their ranking is just on raw operations per second.
YOU HAD LOWERCASE LETTERS? BACK IN MY DAY ASCII WAS UPPERCASE ONLY. DAMN KIDS!
aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa
aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa
This could be one reason why running perl5 in the Parrot VM never really took off. Perl 6 will be a bit better specified, I hope, and so easier to write on top of a specified virtual machine.
Was there ever a time when the count of vulnerabilities was higher for Windows than for a Linux distribution?
Of course I've heard of those things - what I meant was that (AFAIK) there are absolutely zero plans from the Python community to adopt Pirate as their runtime environment or even one of the supported runtimes along with CPython and Jython. I'd be delighted if it all works out and the different scripting languages can find a way to play well together (after all, the semantics are similar enough for most useful libraries to be used across languages) but I am not optimistic.