Linus Torvalds spoke candidly with Mercury News staff writer Dean Takahashi about the lawsuit from SCO Group versus IBM, on Microsoft and open source development. He also shed light on his decision to leave chip maker Transmeta for a Linux corporate software consortium, the OSDL.
one of the things i love about linus is how is always seems to keep out of controversy and how he’s always so neutral. if he didn’t act this way i have trouble seeing how linux could have become so successful.
>how he’s always so neutral
Linus is neutral? Remember the Itanium and Tanenbaum discussion… I’d rather prefer him making considerate comments. He made a reputation of an ignoramus to me so far.
> Remember the Itanium and Tanenbaum discussion…
what was this discussion about ?
The Tanenbaum flamewar was about micro versus monolithic kernels, with Linus defending monolithic kernels and Tanenbaum defending micro kernels. Time proved they were both wrong.
The itanium thing I don’t know about.
Linux vs. Tanenbaum debate:
http://www.oreilly.com/catalog/opensources/book/appa.html
Linus on Itanium:
http://www.theinquirer.net/?article=7966
Linux & Evolution:
http://kerneltrap.org/node.php?id=11&PHPSESSID=6029955ce0df84627bcf…
Best of Linus BS
I’d rather prefer him making considerate comments. He made a reputation of an ignoramus to me so far.
Pot, meet kettle.
Zzzz, another bad troll. What happened to the Snappy Troll School, did it close down?
Best of Linus BS
Brilliant Statements, I suppose. 😉
No, really, have you read the piece about evolution and Unix vendors in-breeding? Great insight.
An interesting analogy I suppose, but not necessarily the holy grail of software development. I disagree with Linus’ statement that designed software is doomed to failure. Software that isn’t designed will be unlikely to be more efficient than any designed software of the time. Therefore although Linux may survive a long time being a “balanced system”, it will never surpass anything that other niche or specialized systems of the time period can do better at. Hence, this could lead to Linux only being a ‘second-best’ alternative system in all categories it competes in. By trying to do everything, it may excel in nothing.
Another problem is that since Linux software “evolves” through a series of hacks, it becomes difficult for Linux to undergo a major change in the kernel, unlike (for example) a designed system such as FreeBSD.
atici,
Have you even read the Tovalds vs Tannenbaum thread that you linked to? If anything, it shows how driven, focused, and intelligent Linus is when it comes to OS design. Even as a college student 12 years ago Linus showed more knowledge and forethought about computer science than prof. Tannenbaum did.
Tannenbaum spends most of his time early in that discussion talking about how multithreaded filesystems are not necessary and that RISC chips are going to displace CISC so developing for the x86 line is just a waste of time.
Linus, on the other hand, in 1991 is smart enough to realize (and promote) the idea that OS portability shouldn’t be a base design factor. He understands that it’s the programs that need to be portable and that a clean, understandable API will get you that. His comments about how it’s the OS’s job to talk to the hardware so layers upon layers of abstraction just for the sake of portability are wrong is dead-nuts on target. He also understood, even back then, that low cost Intel hardware was going to win out over expensive complex chips.
He defends his monolithic kernel thoughts by showing how a properly written multithreaded monolithic kernel gets you multithreaded file access as a free benefit. Minix, in all of it’s micro-kernel, university research driven glory never achieved that feature, except as a very ugly hack.
He also makes a comment about how code written for Linux in 1991 will be portable to GNU/HURD when it is available, “in the 21st century”. Maybe it’s just serendipity that his pseudo-prediction is 100% true–Or, maybe he’s smarter than you give him credit for.
From what you posted Linus seams smarter then ever. I by no means care for linux. But Linus is it’s shinning spot. He is the one who actully seams in touch with reality. It’s sad that so many that go to the religion of linux hail it as so great but at the same time have veiws completely conficting with linus and the way he sees linux. He’s not a zelot, he’s not crazy, he see’s linux just the way it is. I can’t imagine what it would be like if the collective mindset of the linux community was what he was like. For sure linux would have imploded years ago.
atici, the peices you linked to show how smart he is, you seam to just be clueless.
on slashdot.
Or, maybe he’s smarter than you give him credit for.
Yes I read them. I don’t say Linus is stupid (or not), I just don’t consider him as a good computer scientist based on what he claimed. I completely agree with null’s point about the Evolution case. About microkernel approach, or EPIC architecture I think the case is just too clear. These would be considered by almost any CS academic to be superior than what they are replacing for obvious reasons. Fortunately he didn’t go as far as claiming Transmeta’s approach of interpreting x86 code in hardware is efficient.
Today Linux does not have so much momentum because it’s technically superior (not entirely related but: http://news.com.com/2100-1001-949812.html). And if any software or approach does not get merit purely because of technical superiority (but because of hype) this just points out a fundamental flaw within the mentality of the society. That’s why OS/2 never made it. And Linux gaining ground in the world of unices in which there’re technically stronger (better planned and developed) alternatives is just history repeating itself. Thus I don’t think he has much to boast by putting forth the evolutionary development model of linux. That’s not why linux is where it’s today. The momentum behind Linux is just another whim, it could have been behind MINIX too. Unfortunately those whims hardly follow the tehnically superior, it usually happens by sheer luck.
I don’t see anything in the Linux vs. Tanenbaum debate to be excited about. Unless you hate open discussion, the debate doesn’t really give you any particular reason to dislike either Linus or the professor. The professor sounded real cool, and Linus first post sounded like the hot-headed student he is, which is also cool.
Anonymous coward: I think you are giving too much credit to Linus. Sure, he has turned out to be right in some ways, but these aren’t always because of clever conscious decisions on his part. As always, the real world is too complicated to perfectly fit into any one man’s neat theories.
That’s why competition is good.
>The momentum behind Linux is just another whim, it could have been behind MINIX too
Actually, it couldn’t have been behind MINIX because it’s not open.
>> The momentum behind Linux is just another whim, it could have been behind MINIX too.
>>
So you are saying there was no reason why the momentum grew behind linux???? Why was it not behind minix, then??? Minix was there before linux, but if you go back to that discussion, you will find that most of the posters were heavily inclined to back Linux as against Minix. Read the thread again, I think the reason for that is clearly obvious.
>>Unfortunately those whims hardly follow the tehnically superior, it usually happens by sheer luck.
>>
Whatever, but would you be kind enough to explain why /technical superiority/ should be assigned priority over everthing else? Why should it be the most important factor?
I think you are over-hyping the importance of technical superiority. In the real world, it is only one factor to consider when making choices, and history proves that many times over. Price, to take one example, could be a huge consideration.
>> And Linux gaining ground in the world of unices in which there’re technically stronger (better planned and developed) alternatives is just history repeating itself.
>>
Experience is the best teacher. If history is repeating itself so frequently, maybe its time you paid some attention to it ????
Minix was there before linux, but if you go back to that discussion, you will find that most of the posters were heavily inclined to back Linux as against Minix.
What most of the posters say or how many loud backers Linus got in any of those discussions is irrelevant. If most people claimed there was no evolution would that have made any more sense? As an academic myself, I find Linus’ arguments naive at best, even though it might have been supported by many Linux backers.
Whatever, but would you be kind enough to explain why /technical superiority/ should be assigned priority over everthing else? Why should it be the most important factor?
Because in the software world, the other approach is considered as a hack. Well even if sometimes it’s not purely so, you give credit to the better approach (maybe by saying “we take the simpler one because of financial/timing restrictions”). However Linus’ claims tend in the direction of disparaging the better approach rather than showing respect. Well, Linus is clearly no Tanenbaum. And as a natural result of this his OS development model is more populist than elitist. Although he tries to justify this through his pointless evolution argument.
Fortunately there’re people in the open source community,people in the BSD world, that are more knowledgeable and respectful though less loud. But they are quoted less by the media.
Fortunately there’re people in the open source community,people in the BSD world, that are more knowledgeable and respectful though less loud. But they are quoted less by the media.
And this is not to say that there’re none in Linux world. Or I wasn’t trying to bring the discussion to BSD vs. linux debate. But BSD people are generally more academically oriented and that reflects to the quality of the arguments, the development direction of the OS and the whole society.
Hehehe…funny comments here. I just had a feeling someone will bring up FreeBSD and its flawless architecture. :rollseyes:
I just don’t consider him as a good computer scientist based on what he claimed
Probably true, in the sense that Gregory Chaitin is a good computer scientist. But Linus is an engineer, and plus we can’t necessarily make inferences just from his Linux work.
Fortunately there’re people in the open source community,people in the BSD world, that are more knowledgeable and respectful though less loud. But they are quoted less by the media.
Why are you unhappy? With the right licenses, both approaches can coexist and you’re spoilt for choice. In fact, the licenses are compatible but not identical, so you have even more choice.
Let the masses eat Linux.
“Fortunately there’re people in the open source community,people in the BSD world, that are more knowledgeable and respectful though less loud. But they are quoted less by the media.”
Hey everyone can you spell zealot? Z-E-A-L-O-T. Seriously, get over it. The fact is that GNU was missing a kernel and Linux fit in nicely. The whole thing is GPLed and that makes development much faster. So in the end it doesn’t matter what the best approach is because Linux can change and grow so quickly merely because the design approach. Disagree all you want but the truth is that every piece of code introduced under the GPL can be reproduced in a new program wich greatly reduces the time it takes to create an entirely new program with a similar concept. The bazaar nature of Linux also allows for many different people with many different viewpoints to contribute which can greatly improve development efficiency.
“ver Linus’ claims tend in the direction of disparaging the better approach rather than showing respect. Well, Linus is clearly no Tanenbaum.”
I hope you don’t seriously think that microkernels are the end-all be-all of kernel design. Mach proved to be a flop and kernel design today has been leaning towards nanokernels and exokernels.
PS Some people are lucky that I don’t generalize the nature of BSD people as you tend to generalize the nature of Linux people, or I would think you are all a bunch of arrogant pricks.
Why are you unhappy? With the right licenses, both approaches can coexist and you’re spoilt for choice. In fact, the licenses are compatible but not identical, so you have even more choice.
Let the masses eat Linux.
Absolutely. I’m pro choice, not anti-Linux. I have no problems with Linux OS. I know there is quality code as well as hack code in Linux (and probably quite a lot of it is Linus’) and respectable people investing code in it. It’s just Linus’ impudence and ignorant comments that’s getting on my nerves.
Yes. Thanks for bringing up FreeBSD, Mystillef.
Yes. Thanks for bringing up Linux in every story on Microsoft, security, Windows, actually, just plain every story. We appreciate it just as much.
>>
Because in the software world, the other approach is considered as a hack.
>>
And that obtuse classification is good enough reason? Is an OS primarily a tool for doing a job, or is it primarily an implementation of an academic theory?
If you see an OS as the later, then I can see your point. But if your focus is getting the job done,which is what most of the world cares for, then design theories would have to do better than just sound good on paper. They would have to work well in practice. And Minix didn’t do that. That’s where Linux took over.
>> What most of the posters say or how many loud backers Linus got in any of those discussions is irrelevant.
>>
And would you care to explain why their views are so irrelevant? A lot of those posters are developers. Some were students. Quite a few might have been teachers too. And all were certainly users. So how can their views be irrelevant? Are you just being over-the-top elitist???
>> If most people claimed there was no evolution would that have made any more sense?
>>
No, it wouldn’t. And if a minority claims that biology was created by in 6 days, as is now the case, that wouldn’t make any more sense either. In itself, being supported by a majority, or by an elite academic minority, is no sure sign that a theory is true or correct.
>> However Linus’ claims tend in the direction of disparaging the better approach rather than showing respect.
>>
Professors have their respectful place, sir, they really do. But imposing dogma and stiffling student efforts isn’t one of those places. I would rather have a system that allows hot-headed students that one in which teachers are high priests.
But hey, if you are concerned with just what Linus learnt on his mother’s knee, say so clearly. I think we can happily dissagree on that one.
I read the links provided earlier on and I realise that even back in the days, Linux was a dream come true as it is for many of today. This is a miracle. Whatever happened to Minix?
Regards,
Mystilleef
@linux_baby:
The whole discussion was the way Linus makes his comments. Because andy said why he likes Linus’ comments as he is always “neutral” and “keeps out of controversy”. Whereas I thought that’s clearly not true. I for one usually ignore his comments because he is generally nothing more than a media doll.
Now if you still want to discuss about other points:
And that obtuse classification is good enough reason? Is an OS primarily a tool for doing a job, or is it primarily an implementation of an academic theory?
Academic theories on OS do not live in an abstract world. They develop because they represent the planned, efficient and knowledgeable approach to a relevant OS problem. Of course we can get the job done in a most sloppy manner, but arguing this way is superior in any sense (which is what Linus proudly does) is blatant naivete. If software development worked in a purely evolutionary sense we’d be randomly modifying bits to simulate the mutations…
And would you care to explain why their views are so irrelevant?
I say the number of supporters is irrelevant because it seemed to me you’re trying to use it as an argument point. It’s not the number but the supporting argument that matters, and I think in the Tanenbaum case it clearly favors him. For instance I can’t think of any reasonable way how one (be it a developer, student or a top academic) could argue EPIC is an inferior architecture to x86.
In itself, being supported by a majority, or by an elite academic minority, is no sure sign that a theory is true or correct.
Yes how reasonably and thoughtfully a theory is supported is important. That’s the basis of academic system anyway. That’s why I think any member of (as you put it) elite academic minority would argue on the basis of good reason rather than his title.
Professors have their respectful place, sir, they really do. But imposing dogma and stiffling student efforts isn’t one of those places.
Dogma is the last thing I want to impose. But some matters bears little point to discussion. And that’s when Linus chooses to strike!
One thing is true, Linux has a great potential to be used as high end server. This is just the matter of time when Linux would kick many of the Unix servers out of the market. And whatever Microsoft tries to do, It cannot touch the High end server Market.
I am not a developer or computer scientist, but I am both a chemist and a biologist with more than a passing interest in evolution.
If you think that major coding projects are designed rather than evolving you are living in cloud cuckoo land.
You might think you are very scientific with your formal CS approach but in the end you are in the same camp as the “creation science” religious loonies and can’t see the real world because of your dogma.
Evolution does not produce the best or most elegant solution – were are stick with leftover junk like an appendix and furthermore we are stuck with estabished ground plans.
Its the same with OS’s as rms points out that the basic ground plan of Unix might not be the best in the current stage of software development but we are stuck with it. The best answer is to have a process that is optimized for rapid evolution like with kernel.org.
Computer code is like DNA as Dawkins puts it is selfish it wants to replicate and be on as many machines as possible (its not really teleological and anthopomorphic like that but it summarizes the basic outcome of evolutionary processes)
hehehe…
http://ars.userfriendly.org/cartoons/?id=20030706
@atici –
The flaw with your view is that CS has little to say about engineering issues; in other words, the actual act of programming.
You call yourself an ‘academic.’ How does that qualify you for talking about real-world projects where you must work with people who often have better ideas than implementation skills? A lot of ivory-tower teachers ruin their students by claiming outdated design techniques as truth. Take Paul Graham, the Common Lisp booster. He argues that his personal style (and that of most people he knows) is hackish — spit out something, spend time debugging. And while someone like you would find this appalling, it’s little different from a mathematician scribbling vague concepts on paper instead of trying to proceed in some axiomatic way like a braindead computer.
I’ve worked in academic settings and nearly everyone works in a hackish manner (which would be called ‘Agile Methods’ in the real world), but everyone’s ashamed to speak up. If they did, academia might have led the software world.
Look at the Scheme examples you’re taught in school. Is Scheme better than Common Lisp because it’s cleaner? If you think so, it’s because you’ve cut your teeth on toy programs that won’t scale.
I looked at that link you provided with Tanenbaum vs. Torvalds. Tanenbaum should have been a lot more respectful to someone with Torvalds’ mind.
🙂 Nice…I got a good laugh out of that one.
“Academic theories on OS do not live in an abstract world. They develop because they represent the planned, efficient and knowledgeable approach to a relevant OS problem. ”
I don’t think so. I believe they develop because a certain type of person has a tidy mind and likes to look for rules that will lay down the “correct” way to do something.
Exactly the same approach has been followed for a couple of centuries by academics in music. All kinds of rules were worked out detailing the correct way to write a symphony or a string quartet. The academics do make an honest attempt to understand how the music that has already been written works.
Their theories are of almost no help for writing new music – real composers simply do not and cannot follow the rules.
Tanenbaum:
# I think it is a gross error to design an OS
# for any specific architecture, since that is
# not going to be around all that long.
🙂
Well, the current processors are more like RISC in design, but they are still compatible. Besides, Linux has been ported to a number of platforms.
Linux and the open source projects that are based on that platform seem to depend more upon open standards than closed software. This is a victory for the computer industry because open standards are objective, and they support unbiased competition to take place, and a level of compatibility exists regardless of the platform. It solves a lot of compatibility headaches. Not all software has to be based on open standards but web browser technology is certainly an example where the benefits of standards can be recognized.
Microsoft is a dirty company and they can not find peace or victory without freedom. I believe that a microkernel is able to handle more complexity however that complexity must be sigificant otherwise there is no value added by using a decentralized model. It might even be less efficent until complexity reaches a broader level.
Corporations are focusing on object oriented technology yet they have not discovered many designs or paradims. There are many designs that are documented, but not implemented in full scale because the developers are not experienced enough to do the work, however the designs exist, and the benefits of their use are apparent in smaller scale libraries.
I think that Linux will be very effective in taking market share away from Microsoft. The contrast between these two choices is clear. Linux represents fredom, hence victory, while Microsoft represents bondage.
The reason why I support Linux is because it is worth supporting. The computer industry seems to be very unstable, because of the many impending lawsuits and violence. I wouldn’t invest much on it, that’s for certain, yet I’ll buy a used computer every five years and always learn more. It’s amazing what you can get if you are conservative. I’d much rather have my platform written by machines than humans just because there are so many lines of code, literally several million for any application of depth. I have no reason not to support Linus Torvalds and others who work on open source technology. I think that their work is much more important than the work of any vendor.
So, which BSD fanatic wants to take time out of this argument and actually provide some info about the OS. I have a few spare x86 boxes sitting around, and am frankly sick of Linux. I have it running on 2 machines, but want to look into BSD. As I understand it there are several BSD distros as well – anyone wanna give a BSD newbie a push in the right direction? Net? Open? Free? Iknow, Iknow, do your own legwork, but I figured if we were going to be arguing about the finer points of the OS, maybe someone might want to make a convert in the fracas.
Chemicalscum wrote:
I am not a developer or computer scientist, but I am both a chemist and a biologist with more than a passing interest in evolution.
In other words, your observations and conclusions regarding software development or computer science should be taken as a layman’s opinion.
If you think that major coding projects are designed rather than evolving you are living in cloud cuckoo land.
You might think you are very scientific with your formal CS approach but in the end you are in the same camp as the “creation science” religious loonies and can’t see the real world because of your dogma.
If you think that major coding projects in the commercial world (which many posters here seem to love to call the “real” world) are not formally designed, you are living in cloud cuckoo land yourself. If a manager of any significant software project doesn’t insist on a proper design and instead allows his team to “just randomly flip some bits until it works” (which would be proper “evolution”, I’m sure he won’t be leading another project for his company.
There needs to be a good design. That doesn’t equal “dogma”. If there turns out to be an unexpected issue (there always does), the team can deviate from the design – as long as it’s clear that they’re doing that, and why.
Linus certainly has strong opinions on technical issues,
but I think the original poster was referring to Linus’ reluctance to get into *political* discussions. This is something i really like about him too. It would be very easy for him to continually rant about MS or SCO or whoever being evil, like so many of the other self-righteous wankers in the open source world, but he just doesn’t bother.
Look at this SCO nonsense. If i was in his situation i would be foaming-at-the-mouth angry by now, but he seems to be taking a very detached, mature view of the situation.
In other words, your observations and conclusions regarding software development or computer science should be taken as a layman’s opinion.
In other words, your like alici and think I should show proper “respect” for academic authority in computer science. As as biologist I find it enlightening that Linus as a software engineer should be talking about evolution – and I don’t start objecting to him commenting about a biological field.
I might not be a developer but like most scientists at times I generate junk code to get projects working. The most practical and real world of these was in regular production use for laboratory instrumental analysis in the company where I had worked, for two years after I had left, with the only support being two telephone calls. I know how that code developed in an evolutionary manner.
If a manager of any significant software project doesn’t insist on a proper design and instead allows his team to “just randomly flip some bits until it works” (which would be proper “evolution”, I’m sure he won’t be leading another project for his company.
I am nor am I sure is Linus suggesting “just randomly flip some bits until it works” although such an approach has shown it value in both software and mechanical engineering through the use of genetic algorithms for the optimization of the design of aicraft wings, jet engines, automobiles and so on. Rather programmers are are making a lot of purely empirical changes in the code, and what code that is commited to a mojor project is made on a selection process that pretty much mirrors biological evolution. There is a dangerous myth about management that managers actually manage.
I think it can be argued that much of human thought processes occur by evolutionary processes in the brain as the cognitive philosopher Dan Dennnett does in his book “Consciousness Explained” and this of course would be important in understanding the basis on which a developer really produces code.
I wonder why some of the kernel hackers willfully misunderstood what Linus said about evolution. From his perspective, there are all sorts of evolutionary forces at work leading to patches. Each patch may be “designed,” but from his mile-high perspective it’s mutations in his trunk. There is no overall Hand of God design, as you would find in most corporations.
In fact, it reminds me of when people totally misunderstand the reasoning behind Eugenia’s polls.
They don’t wish to understand.
“In other words, your like alici and think I should show proper “respect” for academic authority in computer science. As as biologist I find it enlightening that Linus as a software engineer should be talking about evolution – and I don’t start objecting to him commenting about a biological field.”
Not necessarily; you obviously have bias towards Linus’ evolution view due mostly to the fact that he argues his point using a parallel metaphor that you happen to be able to relate to. I’m not saying that this is necessarily a bad thing, but we should know that models aren’t perfect, and comparing software development to biological evolution is no exception.
Sure, while it may be _possible_ to perform software development in an “evolutionary” manner, it is not the most efficient way to do so. Software design is about deciding on the quickest path from A to B, and as Linus even points out himself, it is a good way to go if you know what your destination is to be. In most software projects, the development team _does_ know what the destination is to be, and designing the software before implementing it will save a lot of extra time and resources in the end. For most projects, developer resources is typically limited (and in the case of companies, costs money). Linux happens to be in a special position where the amount of manpower developing Linux is immense, since it basically scales with the size of the Internet. Most software projects, even other open-source projects, do _NOT_ have such manpower at their disposal, and doing it the “evolutionary” way takes longer than the “design” way.
“Rather programmers are are making a lot of purely empirical changes in the code, and what code that is commited to a mojor project is made on a selection process that pretty much mirrors biological evolution. ……….. I think it can be argued that much of human thought processes occur by evolutionary processes in the brain as the cognitive philosopher Dan Dennnett does in his book ‘Consciousness Explained’ and this of course would be important in understanding the basis on which a developer really produces code.”
This may be true in a sense, but the portions of committed code actually follows a predetermined design guideline, to ensure that the committed code is all with the same focus and goals in mind. The individual code changes may appear to mimic evolutionary design, but the actual content of the code is not pulled into random directions as Linus seems to be advocating.
Another point worth mentioning is that some aspects of software development simply /do not work out well/ with the “evolutionary” development model. Games, for instance, must be designed or it will be a flop. Also, as we have all seen, open source attempts at pleasant and usable GUIs are terrible compared to well designed and thought out commercial GUIs.
Where I work, we have a PHd(Who makes a s-load of money and has a PHd in computer science no less) and a recent grad working for him. This PHd has ABSOLUTELY NO IDEA how databases work. It’s as if he never took one class on databases. I should state that the discussion involved doing esitmates on Oracle. Some acronym threw the guy. Something simple like SQL. Actually, I think the guy didn’t now what a “querry”(sp?) was.
We pay the guy a ton of cash and he is a useless pencil pusher. Not to mention, our work environment doesn’t need PHd’s to ge the jobs done, so why he was hired is beyond me.
Anyways, my point is that in my experience as software engineer, anyone with good grades and Master’s degrees and all that BS are useless 90% of the time. Someone, how they nlearned to think for themselves. Professors, alot of the time are also useless. Their lack of real world experience is absolutely horrible. They are theorists and nothing more.
AS a note, the smartest person I’ve worked with so far, has nothing more than a bachelors degree and is 40. He got promoted up VERY fast because he knows what he is doing and is the best software lead I have met thus far.
Also, a masters in computer science makes absolutely no sense. Or a PHd for that matter. In 5 years, everything you leanred is obsolete. Don’t believe me. One word: JAVA. How about this… Oracle. It’s not like mechanical engineering where steal has been made out of Iron for hte past 100 years. Yesterday was C, then C++ and now JAVA is getting a good foothold.
RANTING IS FUN
linux_baby,
>> I don’t see anything in the Linux vs. Tanenbaum debate
>> to be excited about.
The best thing to laugh about is PHds. Totally USELESS! Linus based his decision on logic and used personal experince. The Teacher only has what others have done and books as his guide, thus the failure.
Sure, my comment is brash, but the debate just shows what is true 90% of the time.
I wonder why some of the kernel hackers willfully misunderstood what Linus said about evolution. From his perspective, there are all sorts of evolutionary forces at work leading to patches. Each patch may be “designed,” but from his mile-high perspective it’s mutations in his trunk. There is no overall Hand of God design, as you would find in most corporations.
Yes – this is what the physicist and cosmologist Max Tegmark calls the “frogs eye view and the birds eye view” though he referring to an even higher level perspective – a theory of everything (TOE).
BTW he uses Linux as his OS to run CMBFAST the cosmic microwave background simulation program in his cosmological research.
The argument rages on, yet not a single person has bothered to address my nooB question. Figures. This has honestly been my experience with *BSD users for a while now. Fsck it, I’ll just load SuSE 7.3 on those boxes and cluster them – that should be a fun project.
The argument rages on, yet not a single person has bothered to address my nooB question. Figures. This has honestly been my experience with *BSD users for a while now. Fsck it, I’ll just load SuSE 7.3 on those boxes and cluster them – that should be a fun project.
I didn’t answer it because it wasn’t related to the discussion and you could’ve found more about it if you did a google search… Check this out:
http://www.extremetech.com/print_article/0,3998,a=20072,00.asp
BSD flavors are not distros. They’re based on different kernel. I’d suggest you stick to FreeBSD if you feel like a nooB.
Thanks, that’s perfect.
And if any software or approach does not get merit purely because of technical superiority (but because of hype) this just points out a fundamental flaw within the mentality of the society.
<p>
BetaMax was superior from a technological standpoint. It lost out because Sony wanted to control what was shown on that medium. They wouldn’t allow porn. Phillips let VHS run free, the porn movies got distributed on it, and that’s why we all have VCRs and not BetaMax machines.
The Wright brothers patented wing-warping as a way to control the roll of an aircraft. Curtis Aircraft, instead of licensing the patent, invented ailerons. To this date there are still strong arguments showing that wing-warping is a superior technology. NASA and the Air Force are even studying it again. But, a superior technology lost out.
Tesla’s plan for DC power grids was (is) arguably better from a technological standpoint than Edison’s AC initiative. Yet, we’re all using AC because of Edison’s political lobbying.
Superior technology loses out all the time and it has nothing to do with the technology or the mentality of society. Sometimes the superior technology is too hard to use (making the term “superior” be objective. Sometimes it gets mired in patents. Sometimes it gets mired in politics.
Linux is winning out because it is superior in certain ways. Those ways may not be technological, but there isn’t the only measure of something that needs to be taken into account.
“>>The momentum behind Linux is just another whim, it could have been behind MINIX too
>Actually, it couldn’t have been behind MINIX because it’s not open. ”
and because some hardware architectures (namely x86) make context switching between kernel and userland very costly, otherwise no one would give a rats ass about monolithic kernels, maybe this is the reason Linus dislikes the Itanium, who knows?
Yes, Richard Gabriel (founder of Lucid, author of worse-is-better) wrote a brilliant book called Patterns of Software, about why Lucid failed.
http://www.amazon.com/exec/obidos/tg/detail/-/0195121236/104-214988…
One of the themes he talks about all the time is how technically worse software has evolutionary advantages over superior software.
http://www.jwz.org/doc/worse-is-better.html
“”So you are saying there was no reason why the momentum grew behind linux????”” – Linux baby
Minix is a _teaching_ OS, Tanenbaum has made strenuous efforts to make sure it is clearly written and doesn’t expand much. Comparisons with current Linux are monumentally unfair.
So into why Linux is here, and now.
Linus couldn’t make the kernel a worthwhile Unix substitute on his own. He needed more developers, so he made it open source to get developers. The Linux zealot would see it as an inspired move, the FSF zealot would see it as some kind of epiphany, and the cynic would see it as a fairly cold blooded move to push development beyond what he could achieve himself.
Combine making it open source with the sluggish nature of Hurd development and the notoriety (And therefore geek cool factor) gained from the Tanenbaum discussions, stir for 10 years and you get what we have today. If we’re accepting Linus as being a smart guy then we shouldn’t fail to credit him with some excellent manipulation of people and situations.
Guess Bill and Linus finally have something in common, they can both program people.
“Tesla’s plan for DC power grids was (is) arguably better from a technological standpoint than Edison’s AC initiative. Yet, we’re all using AC because of Edison’s political lobbying.”
Edison was the DC man, Tesla was AC. Edison lost big time because Tesla and westinghouse had the superior tech and won. There was lobying on Edison’s part, and in the end he fail. DC power grid was a horrible idea, where they were installed were quickly removed and replaced with AC. DC is only good for local leve stuff, IE in a machine, or maybe as a subsystem in a house.
And as a natural result of this his OS development model is more populist than elitist.
Which is exactly why it was so successful. As another said: technical superiority is nice, but it’s hardly the only factor of success. I can truly believe that you’re an academic, seeing how you lack that “real-world” insight.
I wondered if someone was going to bring this essay up. I was going to but I was too lazy to fetch the link. When I read the original article and comments here, worse-is-better was the first thing that popped into my mind. Of course, it doesn’t hurt that I’ve re-read it recently. Very relevant to the current situation and discussion.
The interesting thing in my view is that both Linux and Windows fit the “Worse is Better” mold.
aright lemme put my points across i have slogged through three years of internet computing which is a hybrid of stuff.
in the last two years i made some bad module choices aswell as being forced to do a lot of design namely uml.
From a real world perspective i can tell you this straight out. Design never fucking works.
you can design for every possiblity but its bullshit still.. its not till you actually sit down and do it do you notice the problems with implementation.
as for uml it is the biggest fucking joke i have ever had to do. the most analy retentive idea known to man. only way of getting anything useful out of uml is doing the shitty uml diagrams after you have done the work and to be honest i have no fucking patience sitting down and going through all them boxes with the shitty rational rose working out the bullshit relationships between classes.. fuck all of that.. u do a simple flow diagram showing this class does this goes to that that does this gives it to this class and done fuck all the other bullshit. (sorry for the language but uml really boils my blood)
I have to say this after 3 years of this degree i have come to near abouts the same conclusions as what linus said in the thread about design. I didnt even know his views but i read alticis links and was quite shocked in the same apparent conclusions he has about design.
one of our semesters was actually operating systems where we had to buy tanenbaums book and play around with minix.
minix is very rough and ready.
To explain to all that dont know basically tanenbaum was being a very big prick with his comments basically saying i know better than you. Time has shown how wrong he was. same with a lot of people in universities they have their heads so far up their arse they rarely come out and smell the fresh air..
But tanenbaums whole point of minix was it should be used as an education tool piece and saying what i said linux would have never been created if linus didnt have access to minix code.. Tanenbaum refused to add stuff to minix so that it would be easier to teach with.. he wanted students to have an easier method of understanding the internals of a kernel and an operating system for that he should be commended, but the dismay of all of the people that wanted to add stuff caused the rise in popularity of linux.
what he could have done was keep a small and light version of the core of minix for teaching and allow minix to be developed on.. then minix would prabably be in the place of linux today.
as for the micro-kernel to monolithic.
monolithic is 1 huge chunk. micro-kernel is dynamic loading of different chunks from what i understand of it put in laymans terms. and from what i know windows kernel is monolithic aswell.
i have to say my course leader was one person that i still have a lot of respect for he was the one teaching operating systems who set up the whole course and the most knowledgable and most willing to accept different methods and he is one who from experience came out with how bullshit software lifecycle is, infact most of our lectures with him was him ripping the piss out of information systems students working at mcdonalds at the end and most of the software lifecycle as being crap, and how they would never work in operating systems. which i really cant disagree with.
The problems with planning are very straight forward as linus says u cant say right this is what im gonna do and this is how ill get it all to work cos when u actually implement it nothing ever is as easy as it seems. thats the truth of the matter.
take the church method of microsoft i gaurantee everything is planned and replanned and replanned.. its taken them how long to get a half decent os namely windows xp.. and that operating system is a joke..
ive heard so many people praise windows xp with how stable it is. thats a joke in itself go out and buy an amd xp 2800+ with a mobo with the nvidia nforce2 chipset and a 180 gig hdd (abit nf7-s mobo) yeah the drivers arent digitally signed but that shouldnt matter nvidia are known for how good they are for driver writing install the service pack the amount of headaches and blue screens i saw was a nightmare.. it cant handle new hardware at all.
it cant handle hardrives over 136 gigs lba 48 bit extension is turned off and u have to download a patch to enable it which also comes with sp1.. infact they want u to install sp1 before instlling the patch but as soon as you install sp1 it blue screens on boot up cos it has problems with the nforce drivers etc.. as for serial ata and getting serial ata drive to be noticed that was a damn big headache aswell.. if it was my machine i woulda just installed redhat or debian and done.. by the way redhat installed fine detected the drive size fine. infact the space it didnt detect i originally installed redhat for my friend instead of wasting the space then winxp wouldnt boot because it noticed on space that it didnt relaise went into console mode trying to fix prob something like fix-mbr is the command now instead of fdisk /mbr that did nothing i just ran fdisk afterwards to see if i could just delete the redhat partition and instead it decides to wipe the whole partition table. and just give back 1 big partition i go back to redhat to see if disk druid can fix it but i lost all of it and had to restart again i wasnt happy.
right thats what you call 1 bullshittly massive tangent that had absolutely nothing to do with the discussion to much caffeine and no sleep.. ahh shit lol. but you get the point.
chemicalscum wrote:
In other words, your like alici and think I should show proper “respect” for academic authority in computer science.
No, that’s not what I meant. What I meant was that your observations regarding computer science topics were probably anecdotal.
As as biologist I find it enlightening that Linus as a software engineer should be talking about evolution – and I don’t start objecting to him commenting about a biological field.
You probably would, though, if he was proclaiming doubtful “truths” about that field on the basis of some personal observations.
I might not be a developer but like most scientists at times I generate junk code to get projects working. The most practical and real world of these was in regular production use for laboratory instrumental analysis in the company where I had worked, for two years after I had left, with the only support being two telephone calls. I know how that code developed in an evolutionary manner.
This is exactly where your background shows. I am a physicist myself and I long held the belief that CS was a load of bollocks because any “real” scientist could get any project of any complexity working by “just typing in the code” (in FORTRAN, of course). Yes, that works in many cases, but there is a certain level of complexity above which that Just Doesn’t Work.
I took Software Engineering classes during my physics study and used to scoff at the professor because he claimed that that level of complexity was at around 5000 lines of code. Since I single-handedly designed, built, and marketed an application 20 times that size, he was clearly wrong.
But I found out that it simply doesn’t extrapolate all the way up to infinity.
I’m not trying to attack you personally, please don’t get me wrong. As a biologist, you should know that anecdotal evidence from someone outside the field does not constitute a solid basis for discussion.
You are missing my point further on in my post when you talk about the maximum size of a project “single-handedly designed, built,”, there I am talking about large projects and code selection from code submited by multiple developers being similar to biological evolution.
As a Linux user I have for some time been thinking about the development of the Linux kernel and FOSS in general as being a form of evolution. Consequently I was pleased to discover that Linus’s views on the subject were virtually identical to my own.
Dawkins concept of the meme and the burgeoning science of mimetics is also very relevent to the question. Particularily the concept of the second replicator. Software can be seen as as new sort of meme, or even as a third replicator which makes a fundamental change the the dynamics of cultural evolution. To me the implications of this are enormous.
On other but related points some time ago I came across the worse is better article while browsing jwz’s site and found it intriguing. As an aside does anyone know if the famous MIT AI lab programmer mentioned in the anecdote is rms (BTW Sander I agree with you about anecdotal evidence but introspection is an accepted tool in psychology, further more for much of the theory in CS to be real science it needs to be empirically testable if it goes beyond being a strict mathematical proof). Finally to come to the point both Linus and rms are opposed to backward binary compatability for the kernel and compiler respectively which suggests they want to make the trade off of consistency for rapid evolution at the price of rpm hell etc (though from rms’s position it also creates extra problems for supporting proprietary software to run on GNU/Linux).
To finish this rambling comment – Ximian’s choice of name for its key product suggests that they too see an evolutionary based model to software development (as does their original name of Helix software – presumably a reference to the structure of DNA).
chemicalscum wrote:
“there I am talking about large projects and code selection from code submited by multiple developers being similar to biological evolution.”
This may show resemblance to biological evolution, but my point as stated earlier is that if you take a step backwards to see a bigger picture, there actually *is* a design that guides which code submissions actually will get integrated into the main source tree. Although Linus maintains that Linux was never designed, it actually was, in a sense. If you take a look at which patches he rejects and accepts into the kernel, it is not completely random. Torvalds *does* have some idea of how he wants things to be done (and not to be done) in his tree. If he hadn’t, he would have accepted any kind of patch and leave it entirely to the massive userbase to determine which patches would “survive.”
What many people fail to realize is that no good scientist would make the argument that evolution is superior to design. Scientists are only interested in that it happenned. In fact, I would personally say evolution is inferior. It is a process that is doomed to constant failures, and hence is forced to keep many rollback versions. Look at how many species have gone extinct. Look at the dead end the dinosaurs were. If evolution was to keep one DNA code base (the Linux kernel) for everything, it would have eventually hit a snag and at that point it would have failed almost immediately. It took evolution about 3.5 billion years to make a human. When humans actually try to make a superior being, it will take them less than a .0000001% of that time.
Another thing most people fail to realize is that survival of the fittest applies to everything, both designed and evolutionary based products. Also, fit is a very relative term. For example, in terms of Betamax versus VHS, even though Betamax was better visually, VHS was more fit because it was open. In the current environment, Linux is proving to be more “fit”. It does many things “well enough” while being extremely cheap. But if the environment changes, some other products might prove to win out.
“What many people fail to realize is that no good scientist would make the argument that evolution is superior to design.”
You miss the point I am not saying that evolution is better than design but that from the point of view of mimetics design is an illusion.
“What many people fail to realize is that no good scientist would make the argument that evolution is superior to design.”
You miss the point I am not saying that evolution is better than design but that from the point of view of mimetics design is an illusion.
With regard to null’s comment the situation is analagous to the selective breeding of livestock ( a topic which Darwin covers in detail in the Origin of the Species) as Linus points out in his interview.
Quite frankly, Sun is doomed.
Ah, I love it. I agree 100%
Oh, and…
About microkernel approach, or EPIC architecture I think the case is just too clear. These would be considered by almost any CS academic to be superior than what they are replacing for obvious reasons.
and how much real-world experience do those CS academics have?
Its so frustrating to listen to people who think they know it all.
And if any software or approach does not get merit purely because of technical superiority (but because of hype) this just points out a fundamental flaw within the mentality of the society.
Hey, nobody ever said capitalist society is sane. But look at what Linux does have going for it. Its technically equivelent, open and free, and costs next to nothing. The monetary advantages alone are more than enough for complete and ruthless monopolistic adoption within such a sick society. And I’m all for it.
.!..
If software development worked in a purely evolutionary sense we’d be randomly modifying bits to simulate the mutations…
Ever heard of genetic programming?
Only most Linux developer’s brains are slightly more powerful that the supercomputers we throw at these genetic algorithms. In time I’m sure that will change.
chemicalscum wrote:
“With regard to null’s comment the situation is analagous to the selective breeding of livestock ( a topic which Darwin covers in detail in the Origin of the Species) as Linus points out in his interview.”
If Linus did point this out, it may be somewhat obfuscated… I’m afraid I could not identify this in the interview’s context. (could you show the quote he says please?) With regard to selective breeding, the less-than-worthy mutations are discarded _after_ they undergo a testing phase — the organism with the mutated gene ceases to exist to pass on further offspring. In software design, we cut out nonworthy “mutations” _before_ we even bother to waste our time testing. This, in my view, is the distinction between design and evolution.