That's benchmark-dependent. For some benchmarks, under some circumstances, you can generate faster native code during execution than a traditional compiler would generate beforehand.
You can have quite a bit of fun playing various forms of solitaire with a $2 pack of playing cards. Are you really going to say that a Nintendo DS is 65 times more fun than a pack of cards? Actually, does trying to define a "fun comparison factor" even work?
Any notable increase in fun that you can afford is generally worth it. The Nintendo DS isn't 65 times better than the pack of cards, but it's still worth getting a Nintendo DS because it's fun. If you want to play games that are available on the PS3 and you can afford around $700 for the console and the game, that's worth it. If you can afford a 50" 1080p HDTV to go with your PS3, that'll be worth it too because you'll get an awesome picture.
The whole problem with Windows users using Linux is that they make things incredibly difficult.
Seriously. There seem to be two kinds of "Linux is hard" forum posts:
- A user tries to do something simple and finds the absolutely most complicated way of doing it. (i.e. compiling Firefox from CVS)
- A user tries to do something complicated and unnecessary (i.e. installing XGL by hand on a distro so old that it has libc5, or connecting a Linux machine to a Windows NT 3.51 machine by AppleTalk through an IPv6 VPN.) and then they complain that "Linux is too hard".
Why wouldn't you just install Firefox on Xandros the same way you'd install it on Windows? Sure, there's an "unzipping" step, but if you went through all the effort to install an OS you can handle double clicking on a tarball and hitting the "extract to..." button.
What's ingenious about running software on a computer? I mean, we've had handheld computers with text input since the mid-90's. Perhaps there's some ingenious wireless antenna design that you could argue is patentable, but once you have a wireless handheld computer... running network applications (like email, or a web browser, or VOIP, or networked pong, or whatever) just isn't obvious.
Why is the addition of email software to a portable computer with a wireless network connection worthy of a patent?
Sending email over a network... ZOMG Obvious.
Wireless networking... Like you said, HAMs in the 80's.
Combining the two... this is still obvious. It'd be like getting a patent on carrying groceries in an SUV based on the fact that it's a "significant innovation" from carrying groceries in a station wagon.
That's actually a really good criteria for patents: If you can duplicate the item without reading the patent, the patent isn't benefiting society. The other important rule would be: If you can't duplicate the item by reading the patent, the patent is fraudulent.
A portable email device is "bloody fucking obvious" the minute you live in a world where both email and small wireless devices are common. What are you going to do next? Get a patent on using parallel sorting algorithms on desktop workstations? Is it "not obvious" now because systems with 4 or 8 CPUs in them aren't common? Will a programmer in 2011 be wrong when he says that it's utterly obvious because he has the benefit of hindsight knowing that desktop machines generally have 4+ processors?
We have this thing called engineers. They are trained to solve this exact class of design problems. That's their job. If you call every little step they take in performing their job an "invention" and give the company that they work for a patent on it, well... that's absurd.
The solution stared at them in the face, but no one ever sat down to think it through.
That's the key thing right there. If simply sitting down to think it through will (for a competent expert in the field) usually produce a given solution, that solution is obvious.
Especially with software development, if you let every little creative step be patented, no-one will ever be able to do anything. By the standards of some of these patents, I come up with four or five amazing new inventions every time I sit down and write some code. I assure you - no possible simple combination of well known data structures with standard library functions is an invention. It doesn't become an invention when you add something common like a network, a database, or even both.
It's like if a carpenter went to get a patent for using a nail to affix a piece of plastic to a piece of wood. "It's totally non-obvious, nails were designed to attach two pieces of wood!"
My parents never limited what I could read. They didn't intentionally expose me to violent / sexually aspected television / movies before I was 12 or so, but I never remember being told to leave the room because of some content on TV (like I see other parents do with their kids).
I owned a tabletop game store (sold Magic: The Gathering, D&D books, etc) for a couple of years. The number of teenage kids who came in who weren't allowed to watch R rated moves was ridiculous. One day we were holding a LAN party - playing UT2004 - and an 11-year-old kid commented that if his mother saw that we were playing a first person shooter he'd never be allowed to come back to the store.
Basically, from what I've seen, there is a much larger chance of hurting kids through overprotectiveness than underprotectiveness - at least in my area of the USA today - and therefore parents should try to err on the side of permissiveness.
Oh... and my leatherman comment, I'm serious. I got into a conversation with someone about dangerous weapons and they suggested that everyone should need a license for knives. I asked if that would apply to kitchen knives, and they said "Only when they're being carried outside of their packaging". When I asked about a leatherman tool, they were like "people who need something like that can get a license". I was like WTF?
You are truly suggesting that we should desensitize our children to a point where a young child wouldn't even bat an eye at seeing someone's head blown off?
No, I'm suggesting that a good chunk of the sensitivity that children have to anything is learned, and that there's no really good reason to teach children to be over-sensitized to fictional violence. Every time you "protect" your child from some video content you're reinforcing that it's dangerous and can hurt them - which generally isn't true and warps their outlook for the rest of their lives.
Now, I'm not suggesting that a six-year-old be shown the movie Hannibal. They're not going to understand it, they're not going to understand any attempt to explain it, and they're going to be left with mental images - for no benifit - of a guy being disembowled and people eating human brains. When they're 12 or 14, they should definitely be allowed to see it, because at that point they'll be able to follow what's going on enough to make seeing it meaningful.
Aside from the facts and psychology of the matter, in this day and age, WHY would you push children to 'grow up' while they are still children?
There's a difference between letting them be children (i.e. making sure they don't have to worry what they're eating and where they're sleeping) and explicitly training them that video content can be dangerous and they'll be hurt if they see stuff. The interesting thing about childhood is how fast children learn, and preventing them from having relatively harmless life experiences just reduces their chance of developing a solid understanding of the world they life in.
Very simply, overprotective parents result in children with less complete mental models of the universe. At the extreme, you end up with children who are completely incapable of coping with the world around them. In moderate cases, you end up with people who think that a leatherman is a dangerous weapon that you should need a license to carry. In lesser cases, you simply get people who are culturally retarded because their parents never let them watch South Park as teenagers. None of these are good things.
There's another possibility: My children aren't over-sensitized to fictional violence because I never made an overly big deal about it. Seriously, if your school-aged would lie awake in terror because of something they saw on TV - especially after you got the opportunity to point out to them that it isn't real - they're seriously maladjusted.
Obiwan leaves his pupil to burn to death slowly. I thought that might disturb him. Does that really seem kooky to you?
Right. That's what happens in the story. In stories, sometimes stuff that isn't nice happens. That generally makes them more interesting.
Think about it this way: By skipping that scene, what are you teaching your kid? If you don't skip that scene, what are you teaching your kid? If he complains about being disturbed by the scene, what would you then have the opportunity to teach him?
Wow. What sort of parent is so over-protective that they won't let their kid watch all of Star Wars? Seriously, kids aren't that fragile unless you really isolate them from the world around them - in which case they do a ton of drugs and get 9 venereal diseases as soon as they turn 15.
You're probably right. I was just visualizing votes get counted in my home town - the ballot boxes from the three precincts are brought to Town Hall to be counted. If you've already got observers for each candidate at the polling locations doing the count right there might be a good plan.
No, it's right. The processing power of these systems is actually increasing at that rate. Every second, BlueGene/L is able to do 280.6 trillion floating point operations more than it could do the previous second.
My point is that if you're printing a ballot that the voter is going to inspect, and the voter can clearly see that the ballot is correct, then the software can do whatever it wants - print random ballots, print ballots that only mark Green Party candidates, who cares - the voter will see it and it will get fixed with no possibility of vote fraud as the result of the voting machine code.
If the voting machine code matters, then the design has already failed. An arbitrary voter can easily verify a paper ballot, but there's no way for an arbitrary voter to verify (and therefore have reason to trust) some code running on a voting machine they're using.
First, speed of counting is not something that we can sacrifice trustworthiness (or accuracy) for. If the news reports a result, and the count comes back a week later with some other result... so be it. The politicians are going to have to stop sucking at PR and make proper public statements in the few hours after the election - the correct statement is either "Yea, the exit polls said I only got 20% of the votes, that sucks" or "This one's pretty close, I guess we'll have to wait for the official result to see who won."
Second, I agree that accuracy is something that needs to be built into the system. I'd argue that little old ladies are accurate, and that if they're not we can solve the problem with *more* little old ladies. If that's not good enough, we can move to the sort & count system I described.
In the end, I see this as being really simple: Sacrificing the ability for non-technical observers to immediately spot fraud is NOT ACCEPTABLE. I don't care if we're getting free sports cars in exchange, that's not a design property we can trade away and still have a legitimate democracy.
If the Linux kernel had never been written, a GNU/BSD-kernel system would have been released at some point. There's no reason why that wouldn't have ended up in the same niche that GNU/Linux is in now. The license difference is somewhat relevant, but then we can bring up Hurd and whether or not that would have actually seen usable releases if Linux hadn't existed. Trying to get into it too much further and we'll write ourselves an alternate-history fiction novel.
Very simple, the GNU tools *are* more important to the present-day existence of a usable free Unix operating system than the Linux kernel is. My evidence is that, in 1992, there existed a fully functional free Unix kernel other than Linux (BSD) that the GNU system could have used. There was no alternative to GCC. Today, it's possible to run a free Unix system on a whole bunch of different kernels (Linux, BSD, Solaris, Minix, Hurd). Off the top of my head, I can't think of another free C compiler today.
The important thing isn't the voting software, it's an effective voting procedure.
There is a known effective voting procedure using paper ballots, ballot boxes, and little old ladies (err... party representatives) to count them. This procedure has one important property: fraud attempts tend to get thwarted because the little old ladies will yell when something fishy happens. ANY VOTING SYSTEM WITHOUT THIS PROPERTY SHOULD NOT EVEN BE CONSIDERED.
It may be possible to design a voting procedure using computers that is similarly effective. Here's the important thing: it needs to retain the property that little old ladies observing the process can immediately tell if something fishy is going on. NO FULLY COMPUTERIZED SYSTEM CAN HAVE THAT PROPERTY.
Someone suggested the following system here on Slashdot:
Paper ballots are marked, either with sharpies / pens or from touch-screen ballot generating machines.
They go into standard ballot boxes.
Those ballots are brought to a central tallying location using the standard ballot-box protection procedures.
At the central tallying location, for each race:
The ballots are put into a sorting machine that sorts based on the votes in that race.
Observers check the sorted piles to make sure that they are properly sorted.
The sorted piles are put into a counting machine - there's your counts. If the counts look wrong based on pile size to any observer, it's manual count time.
If any candidate, observer, or 50 signatures question the validity of the counting machine's results - a manual recount occurs for that precinct. Every time - no "but that would be effort" bullshit.
This system takes all the properties of the hand count system and preserves them while spending money to gain two properties: Ballot generating machines for the blind, and fast counting for people who think that matters. Ballot generating machines are an easy problem, and sorting / counting machines are pretty cheap. We might have to use heavy cardstock for the ballots to survive the sort/count process for every race - that's $50 I'm willing to spend.
If there were no Linux, the popular free Unix clone would be FreeBSD. If there were no Gnu, the BSD guys would have had to write a C compiler, which would have cost the free Unix would a couple years. Linux is great (I run Linux), but we've got other free kernels - some of the Gnu tools have no good free alternative.
Slashdot Mirror
That's benchmark-dependent. For some benchmarks, under some circumstances, you can generate faster native code during execution than a traditional compiler would generate beforehand.
Fun doesn't need to increase linearly.
You can have quite a bit of fun playing various forms of solitaire with a $2 pack of playing cards. Are you really going to say that a Nintendo DS is 65 times more fun than a pack of cards? Actually, does trying to define a "fun comparison factor" even work?
Any notable increase in fun that you can afford is generally worth it. The Nintendo DS isn't 65 times better than the pack of cards, but it's still worth getting a Nintendo DS because it's fun. If you want to play games that are available on the PS3 and you can afford around $700 for the console and the game, that's worth it. If you can afford a 50" 1080p HDTV to go with your PS3, that'll be worth it too because you'll get an awesome picture.
ROFL. The whole point of Captchas is to make it so that people *can't* write code to answer them...
Seriously. There seem to be two kinds of "Linux is hard" forum posts:
- A user tries to do something simple and finds the absolutely most complicated way of doing it. (i.e. compiling Firefox from CVS)
- A user tries to do something complicated and unnecessary (i.e. installing XGL by hand on a distro so old that it has libc5, or connecting a Linux machine to a Windows NT 3.51 machine by AppleTalk through an IPv6 VPN.) and then they complain that "Linux is too hard".
Why wouldn't you just install Firefox on Xandros the same way you'd install it on Windows? Sure, there's an "unzipping" step, but if you went through all the effort to install an OS you can handle double clicking on a tarball and hitting the "extract to..." button.
With security mechanisms like that, it doesn't take much to get around them if the mechanism provides automated feedback.
What's ingenious about running software on a computer? I mean, we've had handheld computers with text input since the mid-90's. Perhaps there's some ingenious wireless antenna design that you could argue is patentable, but once you have a wireless handheld computer... running network applications (like email, or a web browser, or VOIP, or networked pong, or whatever) just isn't obvious.
Why is the addition of email software to a portable computer with a wireless network connection worthy of a patent?
Sending email over a network... ZOMG Obvious.
Wireless networking... Like you said, HAMs in the 80's.
Combining the two... this is still obvious. It'd be like getting a patent on carrying groceries in an SUV based on the fact that it's a "significant innovation" from carrying groceries in a station wagon.
That's actually a really good criteria for patents: If you can duplicate the item without reading the patent, the patent isn't benefiting society. The other important rule would be: If you can't duplicate the item by reading the patent, the patent is fraudulent.
A portable email device is "bloody fucking obvious" the minute you live in a world where both email and small wireless devices are common. What are you going to do next? Get a patent on using parallel sorting algorithms on desktop workstations? Is it "not obvious" now because systems with 4 or 8 CPUs in them aren't common? Will a programmer in 2011 be wrong when he says that it's utterly obvious because he has the benefit of hindsight knowing that desktop machines generally have 4+ processors?
We have this thing called engineers. They are trained to solve this exact class of design problems. That's their job. If you call every little step they take in performing their job an "invention" and give the company that they work for a patent on it, well... that's absurd.
That's the key thing right there. If simply sitting down to think it through will (for a competent expert in the field) usually produce a given solution, that solution is obvious.
Especially with software development, if you let every little creative step be patented, no-one will ever be able to do anything. By the standards of some of these patents, I come up with four or five amazing new inventions every time I sit down and write some code. I assure you - no possible simple combination of well known data structures with standard library functions is an invention. It doesn't become an invention when you add something common like a network, a database, or even both.
It's like if a carpenter went to get a patent for using a nail to affix a piece of plastic to a piece of wood. "It's totally non-obvious, nails were designed to attach two pieces of wood!"
My parents never limited what I could read. They didn't intentionally expose me to violent / sexually aspected television / movies before I was 12 or so, but I never remember being told to leave the room because of some content on TV (like I see other parents do with their kids).
I owned a tabletop game store (sold Magic: The Gathering, D&D books, etc) for a couple of years. The number of teenage kids who came in who weren't allowed to watch R rated moves was ridiculous. One day we were holding a LAN party - playing UT2004 - and an 11-year-old kid commented that if his mother saw that we were playing a first person shooter he'd never be allowed to come back to the store.
Basically, from what I've seen, there is a much larger chance of hurting kids through overprotectiveness than underprotectiveness - at least in my area of the USA today - and therefore parents should try to err on the side of permissiveness.
Oh... and my leatherman comment, I'm serious. I got into a conversation with someone about dangerous weapons and they suggested that everyone should need a license for knives. I asked if that would apply to kitchen knives, and they said "Only when they're being carried outside of their packaging". When I asked about a leatherman tool, they were like "people who need something like that can get a license". I was like WTF?
No, I'm suggesting that a good chunk of the sensitivity that children have to anything is learned, and that there's no really good reason to teach children to be over-sensitized to fictional violence. Every time you "protect" your child from some video content you're reinforcing that it's dangerous and can hurt them - which generally isn't true and warps their outlook for the rest of their lives.
Now, I'm not suggesting that a six-year-old be shown the movie Hannibal. They're not going to understand it, they're not going to understand any attempt to explain it, and they're going to be left with mental images - for no benifit - of a guy being disembowled and people eating human brains. When they're 12 or 14, they should definitely be allowed to see it, because at that point they'll be able to follow what's going on enough to make seeing it meaningful.
There's a difference between letting them be children (i.e. making sure they don't have to worry what they're eating and where they're sleeping) and explicitly training them that video content can be dangerous and they'll be hurt if they see stuff. The interesting thing about childhood is how fast children learn, and preventing them from having relatively harmless life experiences just reduces their chance of developing a solid understanding of the world they life in.
Very simply, overprotective parents result in children with less complete mental models of the universe. At the extreme, you end up with children who are completely incapable of coping with the world around them. In moderate cases, you end up with people who think that a leatherman is a dangerous weapon that you should need a license to carry. In lesser cases, you simply get people who are culturally retarded because their parents never let them watch South Park as teenagers. None of these are good things.
There's another possibility: My children aren't over-sensitized to fictional violence because I never made an overly big deal about it. Seriously, if your school-aged would lie awake in terror because of something they saw on TV - especially after you got the opportunity to point out to them that it isn't real - they're seriously maladjusted.
Right. That's what happens in the story. In stories, sometimes stuff that isn't nice happens. That generally makes them more interesting.
Think about it this way: By skipping that scene, what are you teaching your kid? If you don't skip that scene, what are you teaching your kid? If he complains about being disturbed by the scene, what would you then have the opportunity to teach him?
Wow. What sort of parent is so over-protective that they won't let their kid watch all of Star Wars? Seriously, kids aren't that fragile unless you really isolate them from the world around them - in which case they do a ton of drugs and get 9 venereal diseases as soon as they turn 15.
Have you ever considered that a University degree isn't always about job training?
You're probably right. I was just visualizing votes get counted in my home town - the ballot boxes from the three precincts are brought to Town Hall to be counted. If you've already got observers for each candidate at the polling locations doing the count right there might be a good plan.
No, it's right. The processing power of these systems is actually increasing at that rate. Every second, BlueGene/L is able to do 280.6 trillion floating point operations more than it could do the previous second.
My point is that if you're printing a ballot that the voter is going to inspect, and the voter can clearly see that the ballot is correct, then the software can do whatever it wants - print random ballots, print ballots that only mark Green Party candidates, who cares - the voter will see it and it will get fixed with no possibility of vote fraud as the result of the voting machine code.
If the voting machine code matters, then the design has already failed. An arbitrary voter can easily verify a paper ballot, but there's no way for an arbitrary voter to verify (and therefore have reason to trust) some code running on a voting machine they're using.
First, speed of counting is not something that we can sacrifice trustworthiness (or accuracy) for. If the news reports a result, and the count comes back a week later with some other result... so be it. The politicians are going to have to stop sucking at PR and make proper public statements in the few hours after the election - the correct statement is either "Yea, the exit polls said I only got 20% of the votes, that sucks" or "This one's pretty close, I guess we'll have to wait for the official result to see who won."
Second, I agree that accuracy is something that needs to be built into the system. I'd argue that little old ladies are accurate, and that if they're not we can solve the problem with *more* little old ladies. If that's not good enough, we can move to the sort & count system I described.
In the end, I see this as being really simple: Sacrificing the ability for non-technical observers to immediately spot fraud is NOT ACCEPTABLE. I don't care if we're getting free sports cars in exchange, that's not a design property we can trade away and still have a legitimate democracy.
If the Linux kernel had never been written, a GNU/BSD-kernel system would have been released at some point. There's no reason why that wouldn't have ended up in the same niche that GNU/Linux is in now. The license difference is somewhat relevant, but then we can bring up Hurd and whether or not that would have actually seen usable releases if Linux hadn't existed. Trying to get into it too much further and we'll write ourselves an alternate-history fiction novel.
Very simple, the GNU tools *are* more important to the present-day existence of a usable free Unix operating system than the Linux kernel is. My evidence is that, in 1992, there existed a fully functional free Unix kernel other than Linux (BSD) that the GNU system could have used. There was no alternative to GCC. Today, it's possible to run a free Unix system on a whole bunch of different kernels (Linux, BSD, Solaris, Minix, Hurd). Off the top of my head, I can't think of another free C compiler today.
The important thing isn't the voting software, it's an effective voting procedure.
There is a known effective voting procedure using paper ballots, ballot boxes, and little old ladies (err... party representatives) to count them. This procedure has one important property: fraud attempts tend to get thwarted because the little old ladies will yell when something fishy happens. ANY VOTING SYSTEM WITHOUT THIS PROPERTY SHOULD NOT EVEN BE CONSIDERED.
It may be possible to design a voting procedure using computers that is similarly effective. Here's the important thing: it needs to retain the property that little old ladies observing the process can immediately tell if something fishy is going on. NO FULLY COMPUTERIZED SYSTEM CAN HAVE THAT PROPERTY.
Someone suggested the following system here on Slashdot:
At the central tallying location, for each race:
If any candidate, observer, or 50 signatures question the validity of the counting machine's results - a manual recount occurs for that precinct. Every time - no "but that would be effort" bullshit.
This system takes all the properties of the hand count system and preserves them while spending money to gain two properties: Ballot generating machines for the blind, and fast counting for people who think that matters. Ballot generating machines are an easy problem, and sorting / counting machines are pretty cheap. We might have to use heavy cardstock for the ballots to survive the sort/count process for every race - that's $50 I'm willing to spend.
If there were no Linux, the popular free Unix clone would be FreeBSD. If there were no Gnu, the BSD guys would have had to write a C compiler, which would have cost the free Unix would a couple years. Linux is great (I run Linux), but we've got other free kernels - some of the Gnu tools have no good free alternative.