In May 2008 the Obama campaign made a stop in Oregon late in the day. The campaign had, at that point, visited 46 states. Of the 48 contiguous states, they had visited all but two. Oregon brought that up to all but one. He made an off-the-cuff remark, in which he delivered the above quote. But the pause indicated by that ellipsis was quite long. He says "fifty" (clearly, thinking that there are fifty states), then subtracts three for Alaska, Hawaii, and whichever other of the lower 48 the campaign hadn't made any stops in yet, and then says the "... seven states, I think, one left to go," part.
Now, it is funny, because on the face of it it looks like a Harvard-educated guy, runnning for President ferchrissakes, doesn't even know how many states there are. If George Bush flubbed something that badly (as he often did), I would certainly make fun of him. And you're free to make fun of Obama for the flub.
But I wouldn't pretend it meant he really didn't know how many states there are.
Bad idea if you're doing heap allocation in a time-critical code section. Heap allocation ala 'malloc' isn't deterministic, it can take significant time if there's heap fragmentation.
With a generational garbage collector there is a pool of memory in the young generation (typically called "eden") that new objects are allocated from, and they are all contiguous, and there is a pointer to the end of the allocated space. Allocating a block of memory is literally as cheap as incrementing that pointer.
On a somewhat related note, Joel wrote about the importance of languages which manage memory for you automatically in his article How Microsoft Lost the API War. It's from a while back but it's still a great read, so if you missed it, correct that right now.
I tend to agree but as I get older and see more and more bullshit babying and coddling of the convicted and worrying more about them than their victims I tend to start thinking that these people deserve some of their own medicine.
Do you work in corrections? Where do you see the babying and coddling of the convicted? Please give three recent examples of where you, personally, have seen this. Or are you, in fact, simply full of shit?
While a murderer is certainly not a rapist is certainly not an arms trafficker, prison rape is brutal enough to sway people's consciences and not normally deadly.
So you believe that it is right that a shoplifter with a couple of moldy burglary convictions on his record should be rewarded by being regularly raped, and that a convicted rapist should be rewarded by having a nice, juicy shoplifter to be his bitch?
Apparently prison rape isn't brutal enough to sway your conscience. If indeed you have one. I see no evidence that you do.
Unfortunately few U.S. citizens today can recognize a fallacy in argumentation. In American politics debates are usually won by whoever can yell the loudest, beat the most strawmen, scatter the most red herrings, poison the well the most effectively, do a better job of making ad hominem attacks stick, or lie the most convincingly. We desperately need more of the classical liberal arts.
I agree that understanding computation has become an indispensible skill and should be considered one of the modern liberal arts. But I must take issue with your argument that a grounding in mathematics and logic is all that is needed.
Programming languages and command lines may be only one form that the underlying math behind computers takes, but learning them teaches important lessons about how information is represented and structured in computers. Of course, there are other approaches that will bring about the same understanding. But focusing on the pure abstract math behind it all without ever getting down to the nitty-gritty implementation details will never give anyone the skills needed to survive in the modern world.
One example should suffice to demonstrate the point. This evening I helped a friend with some computer problems he was having. One problem was an unwanted program that was running every time he booted up. A few moments with regedit quickly cleared up the problem. I'd like to think I'm fairly well versed in math and logic (yes, even Boolean algebra), but no amount of general knowledge would have been enough for me to solve the problem at hand, and, more importantly, it wouldn't have been enough for him, either!
I knew my way around regedit and the Windows registry; he didn't. Let's have more (and better) education in logic and mathematics (statistics is also vital but too often overlooked), but let's also teach the tools and the practical application of the concepts. Otherwise we'll just produce computer idiot-savants.
$ cat Test.java public class Test {
public static void main(String[] args) {
System.exit(42);
} } $ jikes -O Test.java $ java Test $ echo $? 42 $ ls -l Test.class -rwxrwxrwx 1 Dave None 312 Nov 1 08:56 Test.class
2048 bytes vs. 312, but you can't run the.class file directly. I assume you can just type "test" and it'll find the.NET runtime itself, or would you have to type "mint test" there, too?
(The java.class file is a binary *of sorts*, for sufficiently lax values of "of sorts.")
Anyone who uses a piece of Open Source software is a developer of that software....
If you... simply say, "Feature x doesn't work like feature y does in program z," you will be ignored; as you deserve to be.
Keep it up if you never want open source to be taken seriously or break out of a narrow niche. This is an absurdly arrogant attitude, even for/. Not only do you expect every user to be a programmer (guess open source can never be used by grannies, huh?), but you also presume that typical bug reports are requests for enhancement to make open source software work like an equivalent commercial product.
Does the case of a user who isn't a coder reporting an actual bug fit anywhere in your worldview???
Have you considered the possibility of writing a script that would query RedHat's bugzilla buglist.cgi nightly, retrieve all bugs for your product, and insert any new ones into your SourceForge bug database? (And similarly for other distributions.)
Sectors were also 512 bytes back in the days of 360k floppies. Operating systems, though, generally allocate in blocks (or clusters) that may be more than one sector.
Embedded: J2ME is compiled directly to bytecode, there is no difference between C++ and Java.
All Java code is "compiled directly to bytecode." And if you meant that it's compiled to machine language, that doesn't seem to be true based on what I've read about J2ME, the CVM, and the KVM.
Yes, in theory. But in practice things don't tend to work that way. The vast majority of known polynomial-time solvable problems can be solved in cubic, quadratic, or linear times (or less). Rarely does the exponent get larger than 3. While it is theoretically possible that you would need a very large exponent for a solution to a given problem, it is considered very unlikely that such problems exist.
The assumption here is that the putative polynomial-time algorithm for some current NP problem would tend to look like current algorithms for current P-time problems. What if (part of) the reason we haven't found P-time algorithms for these NP problems is that the solutions to them differ qualitatively from the algorithms we now know? Maybe when (if) we find these algorithms, they will characteristically have polynomials of higher orders; 5th, 8th, 23rd, 42nd, or, as the writer you were responding to suggested, 237,203rd order.
Even if the situation turned out to be as you suggest, you acknowledged that the degree of the polynomial rarely gets above 3. Well, that means that occasionally it does get above 3. An algorithm that had a worst case running time of O(n^4) could make a decent encryption algorithm IF I can choose a key that gets me near that worst-case performance (the typical running time may be significantly less, but I can take all the care I want in picking a good key), and IF I pick a nice, long key (say, 3072 bits rather than 128 bits). That also means I have to have a model that lets me predict which keys get me close to that worst-case time, because we now have an algorithm with a large (maybe enormous) class of weak keys.
This would be bad, but it would not automatically relegate us to one time pads.
As is common, you are overestimating the significance of worst-case complexity, and you are assuming that "n" can get arbitrarily large. In the real world, "n" is often bounded by constraints in the real-world, and algorithms have much better average case complexities than their worst case complexities. I routinely solve problems whose worst-case complexity is O(n^8).
But in such a world where an NP-hard encryption algorithm could be broken in polynomial time, and the worst case is n^5, and the average case is just n^2, or n log n, all I have to do is carefully choose my key to be among those that gets or at least approaches the worst case complexity of n^5.
Of course, this would mean that current secrets could get broken, and it would require a re-evaluation of cipher systems and key choices, but P=NP would *not* necessarily imply that no symmetric ciphers, or even assymmetric ones, would be of any use anymore.
To be honest, I had not thought that much about "software licensing" up till this point, mainly because it seems to me that entrepreneurial hackers will always find a way around that type of thing. But when a piece of software actually requires connection to a corporate server to continue functioning (as with a virusscan program), this seems to fall only slightly short of blackmail.
First of all, I've never seen an antivirus program that didn't offer you the option of scanning with the current scan strings rather than downloading the latest ones. However, if you want to have a known-signature virus scanner that doesn't need to be updated, there is a very simple solution: just convince all the virus writers in the world to stop writing new viruses.
Why is it that so many people seem to believe that the only criteria that applies to anything is money?...
... If a group of people want to screw you over because they can and it's profitable, they can choose to do that....
Why is it that so many people seem to believe that marketing a product with the claim that it works on Windows 95, when in fact it does work on Windows 95, amounts to screwing people over? The original discussion here was about time bombs in software, but we've digressed to saying that failing to provide free upgrades is the moral equivalent of requiring people to upgrade. Or ditto for failing to provide unlimited, eternal tech support.
If I was running WordPerfect 5.2 in a DOS box, I wouldn't expect to get tech support, or support for long filenames. This is a totally different argument from the one about software time bombs. Get it?
All they have to do is implement a computer-based knowledge system and then the tech support monkey just asks you "Alright, what version are you running?" and the knowlege base will adjust its' [sic] answers accordingly.
And they pay the "tech support monkey" $10/hour while they haven't gotten a penny from you in years? Or are you paying a contract for tech support? Which I'm sure you'd consider unethical.
When you make a product with virtually zero duplication cost, why do you then feel you need to force consumers to buy it?
Because your landlord will evict you if you don't pay your rent? Because the supermarket expects zorkmids in return for food? Of course, you aren't forcing people to buy anything.
Here's what we should do. We should sell computers with a minimal OS, just the BIOS, a trivial, non-hierarchical filesystem, and a primitive assembler. And then we'll tell them that they can type in the opcodes of any program they want FOR FREE!
The only way of doing this would be by statute. i.e. ammend [sic] copyright law such that the maximum term is somewhere between 5 and 10 years and that if a copy is not deposited into one or more designated "libraries" at that time any profits from the software are treated as originating from an illegal source.
The length of copyright is currently far too long, and the recent extensions of both the length and scope of copyright in the Sonny Bono Copyright Theft Act and the Digital Millennium Copyright Act are bad policy. Nevertheless, 5 to 10 years is too short a term for copyright, even for software. And the depository requirement you propose is a radical change to current copyright law. The phrase "profits... from an illegal source" seems designed to invoke the RICO statute. Fail to disclose in a timely manner, lose your entire business.
Currently old software can effectivly be hidden for nearly a century.
Currently there is never any requirement that source code be disclosed, even after the copyright has expired.
Or did you simply mean that a copy of the binary should be deposited? In that case, any
end user could do so after the copyright had expired. There is hardly a need to put such a draconian requirement into the law.
It would be sensible, and good policy, to shorten considerably the current, excessive copyright term. The rest of your proposal is unworkable, at best.
Slashdot Mirror
Wonderful. A truly excellent explanation. Thank you.
Stop trying to turn kids that have a hard time reading at age 15 into astrophysicists, lawyers, and surgeons. Its a wasted effort.
What should we conclude about people that have a hard time writing at age ... how old did you say you were, again?
Tesla's don't have engines.
In May 2008 the Obama campaign made a stop in Oregon late in the day. The campaign had, at that point, visited 46 states. Of the 48 contiguous states, they had visited all but two. Oregon brought that up to all but one. He made an off-the-cuff remark, in which he delivered the above quote. But the pause indicated by that ellipsis was quite long. He says "fifty" (clearly, thinking that there are fifty states), then subtracts three for Alaska, Hawaii, and whichever other of the lower 48 the campaign hadn't made any stops in yet, and then says the "... seven states, I think, one left to go," part.
Now, it is funny, because on the face of it it looks like a Harvard-educated guy, runnning for President ferchrissakes, doesn't even know how many states there are. If George Bush flubbed something that badly (as he often did), I would certainly make fun of him. And you're free to make fun of Obama for the flub.
But I wouldn't pretend it meant he really didn't know how many states there are.
Nice. I award you full marx for that one. :)
Ignorance is bliss. Have you ever actually written a web app? If so, how do you do AJAX with just HTML and CSS?
Bad idea if you're doing heap allocation in a time-critical code section. Heap allocation ala 'malloc' isn't deterministic, it can take significant time if there's heap fragmentation.
With a generational garbage collector there is a pool of memory in the young generation (typically called "eden") that new objects are allocated from, and they are all contiguous, and there is a pointer to the end of the allocated space. Allocating a block of memory is literally as cheap as incrementing that pointer.
On a somewhat related note, Joel wrote about the importance of languages which manage memory for you automatically in his article How Microsoft Lost the API War. It's from a while back but it's still a great read, so if you missed it, correct that right now.
They do, but only till 10 and then they have to take off their shoes.
I tend to agree but as I get older and see more and more bullshit babying and coddling of the convicted and worrying more about them than their victims I tend to start thinking that these people deserve some of their own medicine.
Do you work in corrections? Where do you see the babying and coddling of the convicted? Please give three recent examples of where you, personally, have seen this. Or are you, in fact, simply full of shit?
While a murderer is certainly not a rapist is certainly not an arms trafficker, prison rape is brutal enough to sway people's consciences and not normally deadly.
So you believe that it is right that a shoplifter with a couple of moldy burglary convictions on his record should be rewarded by being regularly raped, and that a convicted rapist should be rewarded by having a nice, juicy shoplifter to be his bitch?
Apparently prison rape isn't brutal enough to sway your conscience. If indeed you have one. I see no evidence that you do.
Dave Conrad
I agree that understanding computation has become an indispensible skill and should be considered one of the modern liberal arts. But I must take issue with your argument that a grounding in mathematics and logic is all that is needed.
Programming languages and command lines may be only one form that the underlying math behind computers takes, but learning them teaches important lessons about how information is represented and structured in computers. Of course, there are other approaches that will bring about the same understanding. But focusing on the pure abstract math behind it all without ever getting down to the nitty-gritty implementation details will never give anyone the skills needed to survive in the modern world.
One example should suffice to demonstrate the point. This evening I helped a friend with some computer problems he was having. One problem was an unwanted program that was running every time he booted up. A few moments with regedit quickly cleared up the problem. I'd like to think I'm fairly well versed in math and logic (yes, even Boolean algebra), but no amount of general knowledge would have been enough for me to solve the problem at hand, and, more importantly, it wouldn't have been enough for him, either!
I knew my way around regedit and the Windows registry; he didn't. Let's have more (and better) education in logic and mathematics (statistics is also vital but too often overlooked), but let's also teach the tools and the practical application of the concepts. Otherwise we'll just produce computer idiot-savants.
$ cat Test.java
.class file directly. I assume you can just type "test" and it'll find the .NET runtime itself, or would you have to type "mint test" there, too?
.class file is a binary *of sorts*, for sufficiently lax values of "of sorts.")
public class Test {
public static void main(String[] args) {
System.exit(42);
}
}
$ jikes -O Test.java
$ java Test
$ echo $?
42
$ ls -l Test.class
-rwxrwxrwx 1 Dave None 312 Nov 1 08:56 Test.class
2048 bytes vs. 312, but you can't run the
(The java
Anyone who uses a piece of Open Source software is a developer of that software. ...
If you ... simply say, "Feature x doesn't work like feature y does in program z," you will be ignored; as you deserve to be.
Keep it up if you never want open source to be taken seriously or break out of a narrow niche. This is an absurdly arrogant attitude, even for /. Not only do you expect every user to be a programmer (guess open source can never be used by grannies, huh?), but you also presume that typical bug reports are requests for enhancement to make open source software work like an equivalent commercial product.
Does the case of a user who isn't a coder reporting an actual bug fit anywhere in your worldview???
Have you considered the possibility of writing a script that would query RedHat's bugzilla buglist.cgi nightly, retrieve all bugs for your product, and insert any new ones into your SourceForge bug database? (And similarly for other distributions.)
Sectors were also 512 bytes back in the days of 360k floppies. Operating systems, though, generally allocate in blocks (or clusters) that may be more than one sector.
Well, under DOS the shortest .com program would be INT 20h, which assembles to CD 20 (hex), which is two bytes.
Wow! I never realized until now that Yoda speaks Lisp.
Embedded: J2ME is compiled directly to bytecode, there is no difference between C++ and Java.
All Java code is "compiled directly to bytecode." And if you meant that it's compiled to machine language, that doesn't seem to be true based on what I've read about J2ME, the CVM, and the KVM.
Riverdance would be the most appropriate style of dance for stomping bugs, don't you think?
O(n^t) could be very VERY large. ...
Yes, in theory. But in practice things don't tend to work that way. The vast majority of known polynomial-time solvable problems can be solved in cubic, quadratic, or linear times (or less). Rarely does the exponent get larger than 3. While it is theoretically possible that you would need a very large exponent for a solution to a given problem, it is considered very unlikely that such problems exist.
The assumption here is that the putative polynomial-time algorithm for some current NP problem would tend to look like current algorithms for current P-time problems. What if (part of) the reason we haven't found P-time algorithms for these NP problems is that the solutions to them differ qualitatively from the algorithms we now know? Maybe when (if) we find these algorithms, they will characteristically have polynomials of higher orders; 5th, 8th, 23rd, 42nd, or, as the writer you were responding to suggested, 237,203rd order.
Even if the situation turned out to be as you suggest, you acknowledged that the degree of the polynomial rarely gets above 3. Well, that means that occasionally it does get above 3. An algorithm that had a worst case running time of O(n^4) could make a decent encryption algorithm IF I can choose a key that gets me near that worst-case performance (the typical running time may be significantly less, but I can take all the care I want in picking a good key), and IF I pick a nice, long key (say, 3072 bits rather than 128 bits). That also means I have to have a model that lets me predict which keys get me close to that worst-case time, because we now have an algorithm with a large (maybe enormous) class of weak keys.
This would be bad, but it would not automatically relegate us to one time pads.
As is common, you are overestimating the significance of worst-case complexity, and you are assuming that "n" can get arbitrarily large. In the real world, "n" is often bounded by constraints in the real-world, and algorithms have much better average case complexities than their worst case complexities. I routinely solve problems whose worst-case complexity is O(n^8).
But in such a world where an NP-hard encryption algorithm could be broken in polynomial time, and the worst case is n^5, and the average case is just n^2, or n log n, all I have to do is carefully choose my key to be among those that gets or at least approaches the worst case complexity of n^5.
Of course, this would mean that current secrets could get broken, and it would require a re-evaluation of cipher systems and key choices, but P=NP would *not* necessarily imply that no symmetric ciphers, or even assymmetric ones, would be of any use anymore.
To be honest, I had not thought that much about "software licensing" up till this point, mainly because it seems to me that entrepreneurial hackers will always find a way around that type of thing. But when a piece of software actually requires connection to a corporate server to continue functioning (as with a virusscan program), this seems to fall only slightly short of blackmail.
First of all, I've never seen an antivirus program that didn't offer you the option of scanning with the current scan strings rather than downloading the latest ones. However, if you want to have a known-signature virus scanner that doesn't need to be updated, there is a very simple solution: just convince all the virus writers in the world to stop writing new viruses.
Why is it that so many people seem to believe that the only criteria that applies to anything is money? ...
Why is it that so many people seem to believe that marketing a product with the claim that it works on Windows 95, when in fact it does work on Windows 95, amounts to screwing people over? The original discussion here was about time bombs in software, but we've digressed to saying that failing to provide free upgrades is the moral equivalent of requiring people to upgrade. Or ditto for failing to provide unlimited, eternal tech support.
If I was running WordPerfect 5.2 in a DOS box, I wouldn't expect to get tech support, or support for long filenames. This is a totally different argument from the one about software time bombs. Get it?
All they have to do is implement a computer-based knowledge system and then the tech support monkey just asks you "Alright, what version are you running?" and the knowlege base will adjust its' [sic] answers accordingly.
And they pay the "tech support monkey" $10/hour while they haven't gotten a penny from you in years? Or are you paying a contract for tech support? Which I'm sure you'd consider unethical.
When you make a product with virtually zero duplication cost, why do you then feel you need to force consumers to buy it?
Because your landlord will evict you if you don't pay your rent? Because the supermarket expects zorkmids in return for food? Of course, you aren't forcing people to buy anything.
Here's what we should do. We should sell computers with a minimal OS, just the BIOS, a trivial, non-hierarchical filesystem, and a primitive assembler. And then we'll tell them that they can type in the opcodes of any program they want FOR FREE!
The only way of doing this would be by statute. i.e. ammend [sic] copyright law such that the maximum term is somewhere between 5 and 10 years and that if a copy is not deposited into one or more designated "libraries" at that time any profits from the software are treated as originating from an illegal source.
The length of copyright is currently far too long, and the recent extensions of both the length and scope of copyright in the Sonny Bono Copyright Theft Act and the Digital Millennium Copyright Act are bad policy. Nevertheless, 5 to 10 years is too short a term for copyright, even for software. And the depository requirement you propose is a radical change to current copyright law. The phrase "profits ... from an illegal source" seems designed to invoke the RICO statute. Fail to disclose in a timely manner, lose your entire business.
Currently old software can effectivly be hidden for nearly a century.
Currently there is never any requirement that source code be disclosed, even after the copyright has expired.
Or did you simply mean that a copy of the binary should be deposited? In that case, any end user could do so after the copyright had expired. There is hardly a need to put such a draconian requirement into the law.
It would be sensible, and good policy, to shorten considerably the current, excessive copyright term. The rest of your proposal is unworkable, at best.
I'm pretty sure that by arcade machines he meant those hulking monsters you see in video arcades, NOT home game systems that you plug into your TV.