Everyone is whining about "security through obscurity". Well, sorry, the other forms of security are already gone -- that's what it means that there's a security bug. Security through obscurity is better than nothing.
I agree.
I think the./ elite need to remember that are many, many users that do not continuously download the latest security fixes. Only people who are full-time admins or don't have a life can keep up with this stuff. The rest of us at best manage to update our system with the latest version of our favorite distribution.
Ideally, the system should update itself, but that has its own problems. Maybe this is available already?
I think almost all of the comments are missing the main problem that Bill Joy is pointing out. The problem is the potential lack of human control over the new technologies. As bad as nuclear and chemical weapons are, at least (some) people control whether they are used or not.
A manmade gene/bacteria or a nanomachine might start replicating itself out of control. At some point, it will start running out of whatever resources are needed. Depending on the resources, we have an event anywhere from a blip to a catastrophe.
A smart enough robot will believe that it knows what decisions are best for you. It could become all too easy to simply accept whatever the robot recommends or to let the robot make "minor" decisions without getting your permission. It is not too far off from that point to loss of human control. Whether or not we can achieve perfect AI is a non-issue; you just need enough AI to automate your finances, read your mail, operate your applicances, decide what articles are interesting, etc. This future is not so far off.
Will we have the foresight to design and enforce the checks and balances that are needed? I think Bill Joy is right that we need to do this before we cause some disaster.
I agree with most of the article, especially that Linux is not ready for non-techies. E.g., what do you do if an application dies (or goes into a coma?) and leaves it window up. I've often had to resort to using ps to find the process and kill to kill it. Easy for many of us to do, but not exactly what a non-tech user wants.
Regarding open source management, if some company wants a change RIGHT NOW, the simple solution is that the company puts some money where its mouth is. If the company (or any user) is not willing to pay the price (or help coding), then they don't have any hold on the open source developer.
free will merely refers to the ability of an entity to make decisions not based on simple reaction to stimuli or due to external forces, but based on its own nature and consciousness.
This is a reasonable commonsense definition, but I don't think that many philosophers will accept this formulation. I recall that my parents and teachers would react negatively when I acted according to my nature and would apply stimuli and external forces in an attempt to change my nature, with some amount of success. So much for my free will?
The issue is not free will vs. determinism, but free will vs. materialism. Our brains/bodies are made up of the same kinds of material (matter/energy) as everything else in the universe. The fundamental axiom of Physics is that matter and energy do not behave arbitrarily, but according to physical laws. How do you implement free will with material that cannot, by our best understanding, "choose" how to behave? It is important to note here that quantum laws do not allow choice. Randomness, yes, but not choice.
Maybe you want to give up or relax the "physical law" axiom. That's fine, but there's no science to back it up, only our desire to believe that we are not slaves to Physics.
Also, there are still good reasons to punish wrongdoers if there is no free will. Among other things, we are still very sophisticated learning machines, and negative feedback for bad actions is a good learning principle.
I think the subtle point that should be made is this. It is hard to solve all instances of the traveling salesman problem, and it is impossible (undecidable) to solve all instances of the halting problem, but it is often not-so-hard to solve or approximately solve many, if not most, of these instances. If you have read Penrose's book, he seems to have no clue about this.
For example, computer programmers are faced with debugging programs every day, and debugging is very easily shown to be impossible to do all the time. Nevertheless, many programs are successfully debugged, and many, many programs are debugged enough to be useful, like the Netscape browser I am using right now.
I wouldn't get too excited about genetic algorithms, neural networks, or anything else being the magic bullet of artificial intelligence or computer science. Genetic algorithms are just one of many useful tool of AI, just not a tool for every problem.
People have an extraordinary capability of thinking that what they believe is the Truth. Add to that the ego and hubris and the pressure to publish and get money that is required to be a top scientist, and you have lot of ways that things can go wrong. The scientific method eventually wins out in the end, but it is a typical human mess along the way.
One night some years ago, there was a full moon on Halloween. I and two other teenage hooligans went driving around in the country turning a bunch of mailboxes 180 degrees. [Why? Because they were there.] I assure you that the full moon was more than bright enough to see where you were going.
When you install Linux, normally a large number of other packages and libraries are installed. If I write some code or script, how am I supposed to know if I am violating any licenses? Or that some license would be violated depending on what is distributed with my code? Do we all need lawyers to determine whether the free software we write can be shared freely?
If all free software used a single license, these questions wouldn't be so difficult to answer. But since this ain't gonna happen, it seems to me that the licenses we use should try to avoid this mess, rather than causing it.
I would second the previous recommendations of two books.
Introduction to Algorithms by Cormen, Leiserson, and Rivest, MIT Press
Numerical Recipes in (various languages) by Press et al., Cambridge University Press
The main difficulty with the Cormen textbook is that it is too analytical for a lot of people. You really do need the math in the first part of the book to understand the rest of it (so it's your fault if you skip those chapters). Also, I hate red-black trees; AVL trees are simpler to understand, simpler to implement, and perform about as well. Otherwise, very nice pseudocode and analysis.
The Knuth books are classic, but the choice of MIX for programming is dreadful.
For numerical algorithms, the Numerical Recipes books are superb. I wish the style of analysis in the Cormen book was more like this.
My advice would be to apply to a range of schools and choose the best one that offers you a fellowship/assistantship which includes paying your tuition. US schools are dying to admit good US students so aim high. A high GPA and a high GRE including a high score in the subject GRE test (the special GRE in CS) will get you a long, long ways. If a school sees high numbers here, the reference letters are pretty much ignored. The other way to get attention is to demonstrate research capability (e.g., co-author on some conference paper). The top schools (MIT, CMU, Stanford, Berkeley) are highly competitive and very picky, so don't put all your hopes on getting into these schools. Also send your application to slightly lower tier schools (e.g., UCLA and most other UC schools, Purdue and most other Big 10 schools, and Harvard, Yale, and most other Ivy schools).
Red Hat IPO on E*TRADE only for the Chosen Ones
on
Red Hat IPO Update
·
· Score: 1
I got an E*TRADE account hoping to take advantage of the Red Hat IPO among others.
Anyway, after frustration with their IPO web pages, I poked around and discovered this web page. You need to be a E*TRADE customer to see it. I quote "Red Hat has set aside a certain number of shares in its initial public offering for purchase online through E*TRADE by certain members of the open source community." Another quote, "Red Hat has chosen certain people that the company feels have contributed much to the growth of the company to date. Red Hat wishes to provide these people with the opportunity to purchase shares of stock in the company at the public offering price, should they wish."
Will any of the Chosen Ones speak up? Should the rest of us forget about it?
I think one advantage of the GPL is that many developers believe that the GPL preserves freedom substantially better than other licenses, and this belief gives them greater motivation to develop free software. Whether or not this belief is true [this essay is no help at all], I am happy to use the results of what developers believe.
I think the real issue is less about freedom, and more about whether you are anti-proprietary and/or whether you care about somebody else making money off of your code.
3) Why would they travel the many many light years it would take just to send a message no one is going to receive for centuries and probably won't understand, anyway, when they get it?
I think the postmodern business is silly. Perl has to be a formal language, just like any other programming language.
That said, Perl has some nice features that could be part of any language:
regular expressions and pattern matching of course
automatic memory allocation and garbage collection
relaxed data typing and argument passing
familiar C-like syntax
convenient primitives for files, lists, stacks, queues, and hash tables
eval
For short programs and "exploratory programming", these features are nice because you can avoid tedious details (declarations, memory allocation, pointers) that get in the way of what you want to really want to think about. Larger programs, however, need more constraints to at least ensure widespread agreement on names and types and to limit unexpected behavior (e.g., an eval might do anything).
There are quite a few things I don't like in Perl, e.g., heavy use of $_, uncommented regular expressions, and omitting optional parentheses can make Perl programs look like so much gibberish. It would have been nice if one goal of Perl was to ensure that any program looks semi-sensible to a C programmer, but I guess you can't have everything you want unless you create your own language.
I agree. I also think a big fine (in the billions) is in order with the threat of more fines if Microsoft balks at compliance. The judge also needs to appoint somebody neutral who is technical enough to check compliance.
I think the FBI (and others) might also be worried about the potential damage that a cracker might cause. Although most of the crimes amount to adolescent mischief, it is a little more serious to mess with the government and corporations than with playing tricks on your next-door neighbor.
I agree that these crimes are blown out of proportion, nothing like stealing atomic secrets or genocide. But they are crimes none the less. I don't see any need to romanticize it.
I think the real hypocrisy here is that the government is working hard both to crack down on the crackers and to keep our computers less secure by restricting cryptography.
Most (almost all?) PhD dissertations are quickly put to eternal rest on dusty library shelves, never again to see the light of day. In contrast, Linus has made a visible and significant contribution. Several of you have complained that it is not original. Linus is the developer/coordinator of a free OS kernel that has put fear into Microsoft and all other commercial OS companies. No particular piece is original [note: no piece of a program is original if you look at small-enough pieces, e.g., all programs are formed out of machine instructions], but the combination forming the Linux kernel clearly is original by observing the results. You naysayers can prove me wrong by programming your own world-class OS that gets used by millions of people.
By the way, I don't mean to belittle all the hard work that all you PhDs have done, but unfortunately, an original contribution does not equate to an important contribution. And I also think the *BSD OSes are significant contributions, certainly >= PhD quality.
Can anyone give a timeline on this issue?
on
GNU Inside?
·
· Score: 3
It is my impression that "GNU/Linux" did not become an issue until Linux became popular. I've only been using Linux since 2.0.28, so I can't say much about the history. Any help here? When did RMS decide that a name change was in order?
Also, the point has been made many times that Linux would not be what it is today without GNU. However, I think it is safe to say that GNU would not be what it is today without Linux. Linux has given GNU a lot of vitality. Maybe Linux needs GNU more than GNU needs Linux, but this is nitpicking.
Anyway, "GNU Inside" sounds ok to me. It's not very innovative, so I guess the slogan fits.
I would add John von Neumann (sure is convenient to store your program in memory) and whoever should be credited for the transitor and its continuous miniaturization.
No two nodes on a network can communicate without some sort of "agreement" (i.e., protocol). If you want "metadata", almost any configuration file provides that. The X window system and Sendmail comes to mind as more complex examples.
Can anyone explain the lawyerly mumbo-jumbo? What exactly is supposed to be novel? How does a Netscape cookie (for example) not qualify?
AI is not close to passing the Turing test, or creating any "thinking" machines at a level comparable to humans or any intelligent animal. No one has the foggiest idea of what thought or consciousness is in the first place. Much of what we have are philosophy arguments, which can be fun, but also shows that there is a lot of hot air and very, very little substance (i.e., implementation). Most all of us in AI are trying to get relatively simple things working (simple at least compared to people).
The logic of emergent behavior implies intelligent behavior is crap. Any algorithm is emergent behavior. Just a few simple kinds of instructions are enough to implement any algorithm, so the behavior of any algorithm in some sense "emerges" from simple elements.
As for those of you worried about AI, make sure that you don't improve any program because that will just make the program smarter and one more small step to when computers enslave us.
Slashdot Mirror
I agree.
I think the ./ elite need to remember that are many, many users that do not continuously download the latest security fixes. Only people who are full-time admins or don't have a life can keep up with this stuff. The rest of us at best manage to update our system with the latest version of our favorite distribution.
Ideally, the system should update itself, but that has its own problems. Maybe this is available already?
A manmade gene/bacteria or a nanomachine might start replicating itself out of control. At some point, it will start running out of whatever resources are needed. Depending on the resources, we have an event anywhere from a blip to a catastrophe.
A smart enough robot will believe that it knows what decisions are best for you. It could become all too easy to simply accept whatever the robot recommends or to let the robot make "minor" decisions without getting your permission. It is not too far off from that point to loss of human control. Whether or not we can achieve perfect AI is a non-issue; you just need enough AI to automate your finances, read your mail, operate your applicances, decide what articles are interesting, etc. This future is not so far off.
Will we have the foresight to design and enforce the checks and balances that are needed? I think Bill Joy is right that we need to do this before we cause some disaster.
Maybe they meant free beer.
Regarding open source management, if some company wants a change RIGHT NOW, the simple solution is that the company puts some money where its mouth is. If the company (or any user) is not willing to pay the price (or help coding), then they don't have any hold on the open source developer.
This is a reasonable commonsense definition, but I don't think that many philosophers will accept this formulation. I recall that my parents and teachers would react negatively when I acted according to my nature and would apply stimuli and external forces in an attempt to change my nature, with some amount of success. So much for my free will?
Amen!
Maybe you want to give up or relax the "physical law" axiom. That's fine, but there's no science to back it up, only our desire to believe that we are not slaves to Physics.
Also, there are still good reasons to punish wrongdoers if there is no free will. Among other things, we are still very sophisticated learning machines, and negative feedback for bad actions is a good learning principle.
It's easy. Just implement the stack as a linked list instead of an array.
For example, computer programmers are faced with debugging programs every day, and debugging is very easily shown to be impossible to do all the time. Nevertheless, many programs are successfully debugged, and many, many programs are debugged enough to be useful, like the Netscape browser I am using right now.
I wouldn't get too excited about genetic algorithms, neural networks, or anything else being the magic bullet of artificial intelligence or computer science. Genetic algorithms are just one of many useful tool of AI, just not a tool for every problem.
People have an extraordinary capability of thinking that what they believe is the Truth. Add to that the ego and hubris and the pressure to publish and get money that is required to be a top scientist, and you have lot of ways that things can go wrong. The scientific method eventually wins out in the end, but it is a typical human mess along the way.
One night some years ago, there was a full moon on Halloween. I and two other teenage hooligans went driving around in the country turning a bunch of mailboxes 180 degrees. [Why? Because they were there.] I assure you that the full moon was more than bright enough to see where you were going.
If all free software used a single license, these questions wouldn't be so difficult to answer. But since this ain't gonna happen, it seems to me that the licenses we use should try to avoid this mess, rather than causing it.
Introduction to Algorithms by Cormen, Leiserson, and Rivest, MIT Press
Numerical Recipes in (various languages) by Press et al., Cambridge University Press
The main difficulty with the Cormen textbook is that it is too analytical for a lot of people. You really do need the math in the first part of the book to understand the rest of it (so it's your fault if you skip those chapters). Also, I hate red-black trees; AVL trees are simpler to understand, simpler to implement, and perform about as well. Otherwise, very nice pseudocode and analysis.
The Knuth books are classic, but the choice of MIX for programming is dreadful.
For numerical algorithms, the Numerical Recipes books are superb. I wish the style of analysis in the Cormen book was more like this.
My advice would be to apply to a range of schools and choose the best one that offers you a fellowship/assistantship which includes paying your tuition. US schools are dying to admit good US students so aim high. A high GPA and a high GRE including a high score in the subject GRE test (the special GRE in CS) will get you a long, long ways. If a school sees high numbers here, the reference letters are pretty much ignored. The other way to get attention is to demonstrate research capability (e.g., co-author on some conference paper). The top schools (MIT, CMU, Stanford, Berkeley) are highly competitive and very picky, so don't put all your hopes on getting into these schools. Also send your application to slightly lower tier schools (e.g., UCLA and most other UC schools, Purdue and most other Big 10 schools, and Harvard, Yale, and most other Ivy schools).
Anyway, after frustration with their IPO web pages, I poked around and discovered this web page. You need to be a E*TRADE customer to see it. I quote "Red Hat has set aside a certain number of shares in its initial public offering for purchase online through E*TRADE by certain members of the open source community." Another quote, "Red Hat has chosen certain people that the company feels have contributed much to the growth of the company to date. Red Hat wishes to provide these people with the opportunity to purchase shares of stock in the company at the public offering price, should they wish."
Will any of the Chosen Ones speak up? Should the rest of us forget about it?
I think the real issue is less about freedom, and more about whether you are anti-proprietary and/or whether you care about somebody else making money off of your code.
Maybe because they are French?
That said, Perl has some nice features that could be part of any language:
- regular expressions and pattern matching of course
- automatic memory allocation and garbage collection
- relaxed data typing and argument passing
- familiar C-like syntax
- convenient primitives for files, lists, stacks, queues, and hash tables
- eval
For short programs and "exploratory programming", these features are nice because you can avoid tedious details (declarations, memory allocation, pointers) that get in the way of what you want to really want to think about. Larger programs, however, need more constraints to at least ensure widespread agreement on names and types and to limit unexpected behavior (e.g., an eval might do anything).There are quite a few things I don't like in Perl, e.g., heavy use of $_, uncommented regular expressions, and omitting optional parentheses can make Perl programs look like so much gibberish. It would have been nice if one goal of Perl was to ensure that any program looks semi-sensible to a C programmer, but I guess you can't have everything you want unless you create your own language.
I agree. I also think a big fine (in the billions) is in order with the threat of more fines if Microsoft balks at compliance. The judge also needs to appoint somebody neutral who is technical enough to check compliance.
I agree that these crimes are blown out of proportion, nothing like stealing atomic secrets or genocide. But they are crimes none the less. I don't see any need to romanticize it.
I think the real hypocrisy here is that the government is working hard both to crack down on the crackers and to keep our computers less secure by restricting cryptography.
By the way, I don't mean to belittle all the hard work that all you PhDs have done, but unfortunately, an original contribution does not equate to an important contribution. And I also think the *BSD OSes are significant contributions, certainly >= PhD quality.
Also, the point has been made many times that Linux would not be what it is today without GNU. However, I think it is safe to say that GNU would not be what it is today without Linux. Linux has given GNU a lot of vitality. Maybe Linux needs GNU more than GNU needs Linux, but this is nitpicking.
Anyway, "GNU Inside" sounds ok to me. It's not very innovative, so I guess the slogan fits.
I would add John von Neumann (sure is convenient to store your program in memory) and whoever should be credited for the transitor and its continuous miniaturization.
Can anyone explain the lawyerly mumbo-jumbo? What exactly is supposed to be novel? How does a Netscape cookie (for example) not qualify?
The logic of emergent behavior implies intelligent behavior is crap. Any algorithm is emergent behavior. Just a few simple kinds of instructions are enough to implement any algorithm, so the behavior of any algorithm in some sense "emerges" from simple elements.
As for those of you worried about AI, make sure that you don't improve any program because that will just make the program smarter and one more small step to when computers enslave us.