There's been some speculation on the genetic background of spirituality, but no conclusion has been reached. My personal take is that spirituality is nothing but an annoying side effect of the otherwise beneficial mutations to the human brain.
I haven't any research to hand, but I've read studies which show a positive correlation between religiosity/spirituality and fertility. With any ability to control fertility, even just a simple recognition of the link between sex and pregnancy, a bias towards having more children quickly creates a biological advantage. As one who is secular, the apparent biological advantage of the religious actually worries me: responsibly controlling population growth when others feel compelled to have more and more children only means we shall become an ever smaller and less influential part of the global population.
Spirituality may also have provided some social/developmental advantages. I'm only speculating here, but I can imagine that people who believed they could control their fate by pleasing the gods would have wanted to (a) figure out what the gods wanted, and (b) pass on this knowledge to others, such as their children. The need for a spiritualist class would also have freed some from manual labour, allowing intellectual development.
Looking at early art and architecture, there is a clear connection between religious/spiritual beliefs and great works. If the people of Egypt, for example, had not believed in the divinity and immortality of their pharaohs, would they have built the pyramids? The answer is almost certainly no. In the modern age, we can strive to improve the state of science and technology, for the benefit of future generations, but in earlier ages, spirituality may have been essential.
BSDL does that. What GPL does is to set up a framework of sharing. There is no conceivable reason to use GPL if one is not concerned about that aspect as BSDL would be more "unrestricted" and this have potentially more "utility" to more people. If that is all Linus wanted, GPL is a disastrous choice. But I suspect he simply changed his mind later when the corporate whining started.
You've misunderstood my point. Whatever licence Linus uses for his code, he has never refused to use software under different licences, nor claimed using GPL or non-GPL software is a matter of principle for him. This is different from the GNU ideology, which is in principle against not only contributing to non-GPL software, but even using it.
As a pragmatist myself, I don't find Linus's behaviour either surprising or inconsistent. You're obviously an ideologue, and have said you find it both surprising and inconsistent. I can only suggest this stems more from your own views than from Linus's.
In any case, you appear rather firmly convinced that Linus is lying about his motivations, and it can't be proved one way or the other, so I don't know if there's any value in discussing it further.
This analogy is flawed in the sense that Linux was the project for Linus. In your analogy you would dedicate all your efforts into public service, say free medical care for people, collect accolades and admiration for that, ending up rallying a large crowd to your cause, and then when a buddy from abroad came to visit you, and he happened to be a for-profit chiropractor, you then decided to force all the users of the very public system you worked on, to "try" his "introductory free offer" as a pre-condition of participating. A word: "hypocrisy" comes close in its definition to that. Equally weak becomes the excuse that you never really were claiming to be "against" for-profit doctors.
Linus doesn't spend all his time on Linux, and I don't pay all my money in tax. As for the rest of your comment, I was trying to explain the difference between pragmatism and ideology, not to rehash the BitKeeper issue in another setting (a complete waste of time in my view).
In simple terms, the fact that I support state ownership of some parts of the economy doesn't mean I believe state ownership is best for all parts of the economy (and the reverse for private ownership). This is similar to the way in which a pragmatic open source developer like Linus might believe the GPL is best for some software, but not for all software. The ideologues, like Marxists or Stallmanists, believe in principle that one model of ownership (communal ownership of the means of production in the case of Marxists, with state ownership during a transition phase, and in the case of Stallmanists, effective communal ownership of software) should be used in all cases.
These are licenses, not colour-coordinated socks we are talking about. Since there is nothing preventing the use of BSDL, should it be considered superior or even equal there would be much wider use of it. "Fashion" has nothing to do with it, both BSD and GPL are very well known and thus BSD is not at an "awarness" disadvantage.
To some people, the colour of their socks is more important than whether or not a particular open source project they use or contribute to is licensed under the GPL, BSDL or some other open source licence, as long as it's open source.
There's an extraordinarily large number of open source licences to choose from. The GPL is by far the most well known (again, in large part because of Linux), and is quite obviously the default choice for anyone writing open source software today. I don't think the suggestion that everyone writing such software reads and understands all the open source licences available (or even all the major ones) before choosing which to use is even remotely realistic. However, to the extent that this can't be proved one way or the other, and you're already repeating yourself without addressing my ar
I recall having heard that Linus chose the GPL as a tribute to gcc (not the GNU philosophy)
This is one of those muddled stances I was referring to.
It's clear enough to me: licensing is not a religious issue for Linus. He wants to give away his software, which the GPL lets him do, but his choice of which software to use is based on the utility it provides, not its licence. Most people choose software for this reason.
To use an analogy, I'm happy to pay tax to fund public services like health, education, pensions, support for the unemployed/poor, et al, but I don't believe there's anything wrong with private industry. This is in contrast to Marxists (there are still a few left) who would say the whole economy should be nationalised (to prevent abuse by those who take but don't give back), and classical capitalists who would say all of the state services should be privatised (to allow the profit motive to drive improvements to them).
One only has to look both at the number of developers and users of GPL vs that of BSD. If BSD was indeed superior, or even equal, the natural process would be to equalize the two.
This simply shows that the GPL is fashionable, which is partly a result of the success of Linux. For anyone writing open source software, the obvious licence choice is the GPL, since it's what Linux (the most famous open source software) uses. The only reason to use a more obscure licence like the BSDL would be if it offered some specific advantage, which in most cases it doesn't.
A lot of early Linux users undoubtedly used Linux instead of BSD for the same reason (the USL lawsuit prevented them using BSD), and after that it was largely a matter of network economics.
If it were true, there would be wide-spread efforts by vast hordes of "viral-GPL-hatin'" developers to "restore order" and go back to pure BSD environment.
No, you don't understand. Most of the people who use Linux or BSD don't care about the licence. It's only the ideologues on either side who are overly concerned about it.
The people, like Linus, who used/developed Linux because the USL lawsuit made BSD unavailable didn't want to use BSD because they cared about its licence. They just wanted a Unix-like OS they could use for free, and some also wanted source code so they could learn about it, develop it, etc. Linux provided this, and became dominant in a network economy, which meant that by the time BSD was available, Linux provided higher utility to those looking for a free Unix-like system.
While it is true that BSDL does not prevent wide community acceptance, in projects where the code is clearly dominant, you should note that GPL people will contribute to them, if they must, on an exception basis, because they see them as critical to their operation and at the same time they see these projects as being in no danger of being commercially sold or in some other way hijacked by some corporation.
Such people are GNU ideologues, who care a great deal about licensing ideologies. On the opposite side are the BSD ideologues, who also care a great deal about the licensing ideologies.
What I'm saying is that there are also people who don't care much about the licence, but only want free or reasonably priced software that meets their needs. The overwhelming majority of people using Linux, Unix/BSD, MS Windows or whatever operating system don't care about the licence, except as far as it requires them to pay something. These people are rational in the economic sense, and therefore choose the software that provides the greatest net utility (ie the utility of the software minus the utility of any money and/or time they have to give up to use it).
I recall having heard that Linus chose the GPL as a tribute to gcc (not the GNU philosophy), without which he'd not have been able to create Linux. I know he's also expressed concerns about keeping software free (I mean free in the normal sense of no cost, not the GNU sense).
The biggest difference between Linus and Stallman is that Linus doesn't care so much what other people do. If they write good software, and he thinks agreeing to whichever licence they use is a good exchange for the utility that software provides, he'll use it. Stallman will refuse to use it, and will otherwise oppose it, unless they use the GPL, or allow him to relicense it under the GPL (as in the case of BSDL'd software).
A lot of pragmatic developers use the GPL/LGPL without necessarily agreeing with the GNU philosophy, or often even fully understanding it. Some might in any case feel (as Linus does) that the open source development model will lead to superior software in the long run. For those with such a view, the GPL v BSDL argument makes little difference, since they believe open source software will always be better in the long run. It's only those who believe proprietary software can add value open source can't who would insist on the BSDL. (Right now, I would lean more towards the view that both proprietary and open source development can add value, but I don't think using the GPL versus BSDL makes all that much difference.)
I've never read any convincing argument that Linux is more successful than BSD because of its licence. Linus has said he only wrote Linux because a free BSD was unavailable (owing to the USL lawsuit). A lot of early Linux users undoubtedly used Linux instead of BSD for the same reason (the USL lawsuit prevented them using BSD), and after that it was largely a matter of network economics.
Given the broad success of OpenSSH, the X Window System, Mach, BSD sockets, Apache, et al, it's abundantly clear that the BSDL (or similar) doesn't prevent open source software being successful, or prevent corporate use of it. I think licensing ideologues on both sides (GPL and BSDL) tend to overestimate rather severely the importance of the licence. Things like meeting the needs of users and network economics matter a lot more than BSDL v GPL.
It only falls apart if Tridgell was observing the BitKeeper traffic without the permission of one of the licensees. If he had permission, then access to the software (in the sense of the technologies it implements, not the source code, of course) was given to him (ie it was lent, even if copies of the blueprints weren't given out). If he was eavesdropping without the concent of the licensed users (I don't think this was so), the more appropriate analogy would be that Tridgell stole one of the cars.
Actually, a better way of putting it is Tridgell was able to closely observe and record the workings of one of the cars as it was being used by one who had agreed not to copy it, either with or without permission (with, as far as I know), but with the express goal of copying it. As I said before, there is only so much value in an analogy, and I've explained what I meant anyway, so my analogy is redundant.
Thanks for the links. I still don't see anything about per-processor licensing of MS Office (as opposed to MS-DOS and MS Windows), or economic (as opposed to legal) arguments, but it's still helpful background information. At the same time, there are some obvious major problems with the arguments on http://reactor-core.org/in-microsoft-we-trust.html. The first is the absence of economic theory, the second is the lack of hard data to back up the points being made and the third is lack of sources (which in an academic paper would alone be fatal).
I never said Microsoft Office overtook Lotus and Wordperfect unfairly (although there has been comment on Microsoft using hidden APIs in their Office software to their advantage). Liebowitz's analysis is true in this respect. Microsoft just unfairly closed the market once they overtook the competition.
In that case, there was a miscommunication. I tend to agree that Microsoft didn't unfairly overtake 1-2-3 and WordPerfect (based on what I've read), and I don't know enough about the tactics Microsoft used to maintain its dominance in the office-suite market to hold a meaningful opinion about whether or not they were unfair, or likely had a significant economic impact.
It is my understanding that Tridgell was not using BitKeeper in any way but was observing the on-line protocol by tapping traffic between end-points. Which negates your whole premise. Noone has "lent" him a car. In your analogy, Tridgell was standing on a street and took photos of the cars as they went by following which he proceeded to make his own compatible one.
It only falls apart if Tridgell was observing the BitKeeper traffic without the permission of one of the licensees. If he had permission, then access to the software (in the sense of the technologies it implements, not the source code, of course) was given to him (ie it was lent, even if copies of the blueprints weren't given out). If he was eavesdropping without the concent of the licensed users (I don't think this was so), the more appropriate analogy would be that Tridgell stole one of the cars.
In any case, the usefulness of such analogies is limited. The only point is that Larry was not asking Linus or anyone else using his software to pay him, or to otherwise buy 'Larry gas' or anything else from him. He was allowing them to use his software for free, because it was beneficial both to him and to the Linux pragmatists (but not to the GNU ideologues, for ideological, not technical, reasons).
Actually no. It was anything but practical (although it was beneficial to Linus personally). Linus dropped the ball on this one, his personal friendship with Larry and subsequent infatuation with BitKeeper went against the grain of the whole Linux development process and was causing major disruptions and upheavals in the mailing lists and amongst the developers. Linus tried to enforce his personal preference upon everyone else, disregarding a significant portion of the kernel developer's opinions. Add to this the fact that BitKeeper license was one of most attrocious, outrageous and megalomaniac pieces of self-ego-stroking crap masquarading as a license to ever hit the FSF world. Including provisions like demands that noone who uses BitKeeper is allowed to work on any competing product even if they are not using BitKeeper for that purpose.
The software allowed for co-ordinated revision control of the Linux kernel. This is the practical use for which Linus wanted it. If its use led to ideological conflicts between GNU ideologues (like Tridgell) and Linux pragmatists (like Linus), this was an ideological problem, and not related at all to the suitability of BitKeeper for tracking revisions to the Linux kernel.
Linus made an error and he paid for it by losing a significant portion of his stature and respect people had for him. After the fiasco blew up in his face he managed to recover somewhat by quickly moving to disprove his own earlier claims of insurmountable difficulties of automating kernel source management by creating the "git" system which is by now superior to BitKeeper in most respects.
You agree with the GNU ideologues. A lot of other people disagree with them, so there's no point in claiming one opinion or the other is the right one. The facts are: (a) BitKeeper was technically suitable for revision control of the Linux kernel; (b) it is no longer used because of ideological, not technical or economic, reasons; (c) the development of git has imposed an opportunity cost, in that the time Linus and others have used to develop it could otherwise have been used to improve Linux itself.
But a lot of people will remember this massive screwup which exposed Linus for his lack of principles and will no longer trust him to make the right call as they did before.
Linus has never claimed to be a GNU ideologue, or to believe in Stallman's GNU philosophy. The principles you refer to are the principles of that philosophy, not general principles normal people believe in. Since Linus has never adhered or claimed to adhere to the GNU philosophy, it should not surprise you that his principles are not the same as you
It's also true that if software patents don't spur innovation (and I don't think they do, at least not the current regimes), eliminating them in only some countries won't put those countries at a competitive disadvantage, in the way that destroying nuclear weapons would put them at a military disadvantage. In other words, there is no risk in being the first mover, and in fact there may be an advantage to being such.
On a company level, it would of course be unwise to give them up, but on a national level it could be done without requiring all countries to do it at once, and without any action from companies (since their software patents would simply become invalid).
There were no licensing fees for using BitKeeper, so your suggestion of having to use gas/petrol from Larry isn't so good, since this is normally something that has to be bought.
A better analogy (sorry!;-) ) would be:
Larry let Linux and his friends borrow early models of a new car he was designing, helping them by letting them drive the cars, and helping him because they told him how they liked them, etc. It was agreed that they would not try to copy the car design.
Someone lent one of the cars to another friend, who hadn't agreed not to try to copy its design, and he immediately began to do so. Larry reacted by taking his cars back, and so Linus and his friends had to design and build their own, which is something they hadn't wanted to spend their time doing.
If your perspective is the GNU/FSF ideology, then of course you will consider it better not to use Larry's cars at all. However, not everyone shares that ideology, so to some of the others, it was better to have nice cars for free, without having to bother designing and building them.
I'm sure there are problems with this analogy too, but the basic point is the agreement between Linus and Larry was mutually beneficial (in practical terms, not GNU/FSF ideological terms) until some people who disagreed with it for ideological reasons spoilt things.
Most economists involved in extensively researching such cases are biased towards one side or the other: why else would they take an interest? The question is whether or not their biased arguments stand up to scrutiny. For interested observers, it's a matter of reading arguments from both sides, to see which are more logically coherent.
Being interested in this from the perspective of economics (I'm not interested in software advocacy), I'd like to read the views of opposing economists, if you know of any. Most of the ones I know of have written about MS Windows, not MS Office.
The most striking thing for me in this market is the dominance MS Office achieved on the Apple Mac, when Microsoft was still far behind its competitors on the PC (still DOS in those days). A coherent explanation of why this happened, if Microsoft's products weren't technically superior to the competition, is something I'd really like to read.
I don't have experience of the market in those days, so I'll take your word for it. However, this still doesn't explain why Word and Excel were dominant on the Apple Mac years before they were market leaders on the PC, unless Apple bundled them (I don't know, but this would surprise me if it happened). If it was not simply a matter of being better products, why were WordPerfect, Lotus, et al. unable to compete with Microsoft on the Mac? The most convincing argument I've heard is that they failed to appreciate the importance of the GUI, which is why the transition from DOS to Windows on the PC undermined their business models.
As for the prices, I've read economic analyses which show the prices for Lotus 1-2-3 and WordPerfect were higher than those for Excel and Word, respectively, until 1995. By 1995, Word and Excel had achieved 70-80% market share, so the moderate price reductions by Lotus and WordPerfect (for products which were still more poorly rated than Excel and Word) weren't enough to stop their decline. Steep price falls to 1-2-3 and WordPerfect only came in 1996 and 1997, and they never achieved technical superiority. With tiny market shares and technically inferior products, it's easy to see why low prices weren't enough to save WordPerfect and Lotus, but only starved them of the income needed to develop their declining products.
By the way, can you provide a link documenting a switch by Microsoft from per-processor licensing of Windows to per-processor licensing of Office? The original anti-trust case against Microsoft (over abuse of Windows to undermine Netscape) was filed in 1998, long after WordPerfect and Lotus 1-2-3 had collapsed. A licensing change after 1998 couldn't have had anything to do with Word and Excel gaining market dominance, since this had happened some years earlier: by 1997, both products had more than 90% market share in revenue terms.
It wasn't listed in the invoice? In my experience, computer sellers always try to convince customers buying PCs from them to add software bundles and other things, but except for very cheap software and some operating system, these bundles are always optional items.
Do you think it's viable to investigate every one of your customers to make sure they aren't agents of your competitors? How will you guard against them later deciding to become such agents?
To use a real world example, do you think there is any way for Red Hat to prevent SuSE/Novell buying a single copy of Red Hat's most expensive product, and copying from it any code they find useful?
That's true, but it wouldn't matter. In a GPL world, the important differentiation is in the provision of services, not software itself. Even today, studies that have shown that when companies acquire software, their willingness to pay is most strongly linked to warranties and product support. That product support often involves development of new or customized features. In addition, GPL does not require redistribution of in-house code. So if you create some "secret sauce" that is tailored to a specific company's internal needs, you don't have to make it public. This is no different than in-house software today, which represents a very large portion of the jobs out there.
In this case, investment would be diverted from software development to services development. The end result might be production of a similar amount of overall value, at least initially, but even if this were so, the software would still see less investment and a slower rate of technological development. This would only be offset by a more highly developed services market. Do cheaper services represent technological progress? Would continued development of the services side of the business be sustainable, and lead to a similar level of overall value production to proprietary and BSDL'd software develepment? I don't know the answer to either question.
You're still looking at the situation from the traditional perspective. You can't look at it as two competing software companies A and B, which are reluctant to invest large chunks of effort because the other will free-ride. Instead, consider hundreds of firms that use a particular piece of GPL software while each contribute small portions back to the shared codebase. The incentive is not to profit on software itself but to meet their own needs. Sharing improvements just makes the process easier and ultimately cheaper. All of those contributions add up over time and the software evolves through user-driven development. How that development occurs is variable. Some of it may be in-house work. Some of it may be contracted through core project leaders. Some of it may be a side-effect of hired support services.
It doesn't matter how many firms there are, or whether or not software production is their primary business. If the benefits of investment accrue to all users, those who fund the investment will be put at a competitive disadvantage, and the market will push down levels of investment. An example of this is the PC market, where Dell has gained its position by explicitly doing as little investment in development of the PC platform as possible. This frees up resources for manufacturing, marketing and distribution.
That's a reasonable consideration because Open Source does, indeed, largely eliminate the potential for monopoly profits. (services markets can be monopolized as well.. it's just much harder.) On the other hand, I do not believe that new software technology is primarily produced by large investors. Software, like science, is too evolutionary of a field. It's hard to force progress by simply throwing money at any one area -- in fact this can often result in "solutions in search of problems." In the early days of computing, large investment was needed to jump-start the industry; today, there are so many players that innovation happens automatically amidst the mundane. As a developer, I don't care about monopoly profits. I only care about making a reasonable living while doing what I enjoy. I am paid to meet needs but I innovate in the process because it makes the jobs more fun.
It's not a monopoly situation, but rather monopolistic competition, which means each producer attempts to differentiate its products from others by adding value not available in them. This is the norm for most goods markets. Perfectly competitive markets (or their closest real-world approximations) are found in areas like agriculture, where individual producers have no power to influence prices.
If the services market allows for differentiation, then this could o
This is only a guess, but I expect they have a lot of standard forms, with automated processes for finding the right forms, filling them in, saving them to the right places, sending copies to the right people and so on. A word processor can be used for much more than just expressing text.
Apart from word processing, there are also a lot of things I can imagine they would use Excel for, and even PowerPoint. They could also be using Access, but there are much better databases available, so I don't think they would be using MS Office for that.
I don't much like Word, but Excel is something I use a lot, and no other spreadsheet I've tried is good enough to convince me to use it instead. Besides, everyone I know uses Excel, so why bother with some other spreadsheet?
I've never seen MS Office bundled for 'free' with a PC. It's usually available for a price much lower than retail (ie an OEM price) if you buy it with a PC, but this is true of a lot of software, including competitors to MS Office, and has long been so.
Stan Liebowitz (a professor of economics at the University of Texas) makes a fairly convincing case that Word and Excel succeeded because they were better than the competing products. Both were market leaders on Mac before PC, so those who think Microsoft cheated have to come up with an explanation of how it did so on Mac.
Although the "innovation" approach works somewhat even in the free software world -- consider RedHat, who makes large investments in Linux and therefore reaps the rewards instead of the "stagnant" Debian et al.
I find this interesting, since economic theory would predict that investment in Linux made by Red Hat would benefit all Linux producers, and therefore leave Red Hat at a competitive disadvantage. Two possibilities that come to mind are:
1) Funding development of Linux allows Red Hat to provide better support than competitors who do not fund its development.
2) Red Hat is benefiting from its investment in the short run, but as other firms catch on, this will be eroded and its levels of investment will have to fall to maintain profitability.
I don't know which it is, but if it's the first one, that suggests a GPL-dominated software environment can still leave room for differentiation. That's good in the sense that it makes investment attractive, but if it is the case, it also means the GPL won't prevent domination of a given market by a single firm.
No, I'm suggesting that the software industry should be viewed as a labor / services market rather than an artificial goods market. Until machines program themselves, software labor will never be commoditized because developers will always have their specialties. So yes, all software itself can be commoditized in the sense that the code itself is not seen as the source of value. This is different than commoditizing the software industry. There's a huge difference.
You're playing a game of semantics. If I say the the RAM industry has been commoditised, it means exactly the same thing as saying that RAM has become a commodity. What this means is that there is no meaningful differentiation between RAM chips from different producers. In an entirely GPL world, there would be no meaningful differentiation between software from different producers.
Investment in software itself does not need to be profitable on its own for the incentive to exist. Yes, we may eventually see the end of traditional software companies where all development is productized and centralized under one roof. Regardless, software companies will continue to become more diverse and fewer will take on the whole development load by themselves. Investment in software itself will be more driven by real user needs, and meeting those needs properly will always be profitable.
Without a sufficient return on investment, new investment will not be made. If firm A invests in software to meet customer needs, but firm B can benefit from that investment too, it is obvious that firm B will have a competitive advantage, and thus the trend within the sector will be towards ever lower levels of investment. After all, there will always be alternatives to invest in if software development becomes unattractive.
This is pure speculation. First, the total amount of investment in software is not what matters; the rate of meaningful production is what counts. Today's software industry is incredibly inefficient. If Open Source causes a massive increase in efficiency, due to code sharing, it is expected that investment in software itself may drop; but the total useful output could actually increase greatly. At the same time, Open Source dramatically reduces barriers of entry into the market and this has the opposite effect on investment. Likewise there is no reason whatsoever to believe that technological progress would likely slow. Most of the investment in today's software industry does not go into improving the state of the art; it goes into reinventing wheels. In terms of resource allocation, services and support are intimately tied to software development within the Open Source model. Companies that have tried to separate the two have generally failed.
Open source with a GPL-like licence moves the software market from a market structure resembling monopolistic competition to one resembling perfect competition. Perfectly competitive markets are more efficient, yes, but they are also not profitable in the long run (which is why, for example, agriculture is subsidised). Under market structures more resembling monopoly, there is the possibility for profit in the long run, and this profit can be used to invest in new technology. One obvious example of the effects of monopoly investment is the development of Unix at AT&T.
GPL open source will lead to greater efficiency, yes, but this greater efficiency will reduce the opportunity for profit, and will drive many producers out of the market, especially those which invest heavily, since the benefits of such investment accrue to all producers, not only the ones funding the investment. The expected result, according to economic theory, will be cheaper software, but also lower rate technological progress.
Ironically, what you described is precisely where the software industry was before Open Source entered the scene and started shaking things up. Thus far, Open Source has done nothing but stimulate progress and innovation. The only thing that remains is full-on
Yes, but surely you can see that helping those who would prevent you from being alturistic to others in the future is an inherently condradictory stance?
If I give some of my code to someone, under even public domain, how can they stop me either: (a) giving that same code to someone else; (b) giving away any other code I write? The answer is they can't! They cannot stop me being altruistic, even if they're entirely selfish.
You're confusing altriusme with a philosophy of achieving a 'greater good' through not being altruistic. If Richard Stallman had been motivated by altruism, he would not have been bothered that James Gosling benefited from his emacs work without sharing his own code, and would therefore have never written the GPL.
I prefer the BSDL to the GPL because I think it provides greater utility to a greater number of people: even users of closed systems can benefit if the BSDL'd code is good enough to be integrated into them. Supporters of the GPL argue that it will lead to greater utility in the long run because it will destroy proprietary software (which the believe provides less utility than open software). This is a position I disagree with (primarily because I don't believe the utility of software in general is a function only of its licence), but a sound argument can certainly be made in favour of it.
A competitor has only to buy one copy of the given software, in order to get full rights to the source code, and thus the right to copy anything valuable into its own product. In other words, any competitor that wants it can easily get it, so a firm is effectively required to give all changes to its own software immediately to all its competitors.
If your motive is altruism, why use the GPL rather than the BSDL or public domain? With the GPL, you are only willing to help others if they agree to help you. This is not altruism.
What you are suggesting is effectively commoditisation of the software industry. What this would lead to is a situation in which investment in software itself would not generate a sufficient return to be generally profitable. As a result, investment in software would steadily decline, and technological progress would slow. Resources would be shifted from software development to profitable endeavours such as marketing, services and support.
Because of the way markets work, a situation such as the one above would likely be broken by innovation in one place or another. It might be innovation in new proprietary hardware, or it might be innovation in new proprietary software. One way or the other, the steady decline in investment in commodity software would leave a stagnant market, ripe to be overturned by investment in alternatives in one area or another.
Most successful proprietary software firms make their money by providing features their customers want. Those that follow strategies such as you're suggesting typically end up being marginalised, as their customers migrate to alternatives from firms who respond to their requirements.
For firms whose business is based on meeting customer needs more rapidly than their competitors, the GPL is not viable, because under the GPL, any improvements made to satisfy customers must also be immediately made available to all competitors. Those who invest in R&D and technical development therefore lose out to those who invest only in marketing.
Slashdot Mirror
Spirituality may also have provided some social/developmental advantages. I'm only speculating here, but I can imagine that people who believed they could control their fate by pleasing the gods would have wanted to (a) figure out what the gods wanted, and (b) pass on this knowledge to others, such as their children. The need for a spiritualist class would also have freed some from manual labour, allowing intellectual development.
Looking at early art and architecture, there is a clear connection between religious/spiritual beliefs and great works. If the people of Egypt, for example, had not believed in the divinity and immortality of their pharaohs, would they have built the pyramids? The answer is almost certainly no. In the modern age, we can strive to improve the state of science and technology, for the benefit of future generations, but in earlier ages, spirituality may have been essential.
You've misunderstood my point. Whatever licence Linus uses for his code, he has never refused to use software under different licences, nor claimed using GPL or non-GPL software is a matter of principle for him. This is different from the GNU ideology, which is in principle against not only contributing to non-GPL software, but even using it.
As a pragmatist myself, I don't find Linus's behaviour either surprising or inconsistent. You're obviously an ideologue, and have said you find it both surprising and inconsistent. I can only suggest this stems more from your own views than from Linus's.
In any case, you appear rather firmly convinced that Linus is lying about his motivations, and it can't be proved one way or the other, so I don't know if there's any value in discussing it further.
This analogy is flawed in the sense that Linux was the project for Linus. In your analogy you would dedicate all your efforts into public service, say free medical care for people, collect accolades and admiration for that, ending up rallying a large crowd to your cause, and then when a buddy from abroad came to visit you, and he happened to be a for-profit chiropractor, you then decided to force all the users of the very public system you worked on, to "try" his "introductory free offer" as a pre-condition of participating. A word: "hypocrisy" comes close in its definition to that. Equally weak becomes the excuse that you never really were claiming to be "against" for-profit doctors.
Linus doesn't spend all his time on Linux, and I don't pay all my money in tax. As for the rest of your comment, I was trying to explain the difference between pragmatism and ideology, not to rehash the BitKeeper issue in another setting (a complete waste of time in my view).
In simple terms, the fact that I support state ownership of some parts of the economy doesn't mean I believe state ownership is best for all parts of the economy (and the reverse for private ownership). This is similar to the way in which a pragmatic open source developer like Linus might believe the GPL is best for some software, but not for all software. The ideologues, like Marxists or Stallmanists, believe in principle that one model of ownership (communal ownership of the means of production in the case of Marxists, with state ownership during a transition phase, and in the case of Stallmanists, effective communal ownership of software) should be used in all cases.
These are licenses, not colour-coordinated socks we are talking about. Since there is nothing preventing the use of BSDL, should it be considered superior or even equal there would be much wider use of it. "Fashion" has nothing to do with it, both BSD and GPL are very well known and thus BSD is not at an "awarness" disadvantage.
To some people, the colour of their socks is more important than whether or not a particular open source project they use or contribute to is licensed under the GPL, BSDL or some other open source licence, as long as it's open source.
There's an extraordinarily large number of open source licences to choose from. The GPL is by far the most well known (again, in large part because of Linux), and is quite obviously the default choice for anyone writing open source software today. I don't think the suggestion that everyone writing such software reads and understands all the open source licences available (or even all the major ones) before choosing which to use is even remotely realistic. However, to the extent that this can't be proved one way or the other, and you're already repeating yourself without addressing my ar
This is one of those muddled stances I was referring to.
It's clear enough to me: licensing is not a religious issue for Linus. He wants to give away his software, which the GPL lets him do, but his choice of which software to use is based on the utility it provides, not its licence. Most people choose software for this reason.
To use an analogy, I'm happy to pay tax to fund public services like health, education, pensions, support for the unemployed/poor, et al, but I don't believe there's anything wrong with private industry. This is in contrast to Marxists (there are still a few left) who would say the whole economy should be nationalised (to prevent abuse by those who take but don't give back), and classical capitalists who would say all of the state services should be privatised (to allow the profit motive to drive improvements to them).
One only has to look both at the number of developers and users of GPL vs that of BSD. If BSD was indeed superior, or even equal, the natural process would be to equalize the two.
This simply shows that the GPL is fashionable, which is partly a result of the success of Linux. For anyone writing open source software, the obvious licence choice is the GPL, since it's what Linux (the most famous open source software) uses. The only reason to use a more obscure licence like the BSDL would be if it offered some specific advantage, which in most cases it doesn't.
A lot of early Linux users undoubtedly used Linux instead of BSD for the same reason (the USL lawsuit prevented them using BSD), and after that it was largely a matter of network economics.
If it were true, there would be wide-spread efforts by vast hordes of "viral-GPL-hatin'" developers to "restore order" and go back to pure BSD environment.
No, you don't understand. Most of the people who use Linux or BSD don't care about the licence. It's only the ideologues on either side who are overly concerned about it.
The people, like Linus, who used/developed Linux because the USL lawsuit made BSD unavailable didn't want to use BSD because they cared about its licence. They just wanted a Unix-like OS they could use for free, and some also wanted source code so they could learn about it, develop it, etc. Linux provided this, and became dominant in a network economy, which meant that by the time BSD was available, Linux provided higher utility to those looking for a free Unix-like system.
While it is true that BSDL does not prevent wide community acceptance, in projects where the code is clearly dominant, you should note that GPL people will contribute to them, if they must, on an exception basis, because they see them as critical to their operation and at the same time they see these projects as being in no danger of being commercially sold or in some other way hijacked by some corporation.
Such people are GNU ideologues, who care a great deal about licensing ideologies. On the opposite side are the BSD ideologues, who also care a great deal about the licensing ideologies.
What I'm saying is that there are also people who don't care much about the licence, but only want free or reasonably priced software that meets their needs. The overwhelming majority of people using Linux, Unix/BSD, MS Windows or whatever operating system don't care about the licence, except as far as it requires them to pay something. These people are rational in the economic sense, and therefore choose the software that provides the greatest net utility (ie the utility of the software minus the utility of any money and/or time they have to give up to use it).
The biggest difference between Linus and Stallman is that Linus doesn't care so much what other people do. If they write good software, and he thinks agreeing to whichever licence they use is a good exchange for the utility that software provides, he'll use it. Stallman will refuse to use it, and will otherwise oppose it, unless they use the GPL, or allow him to relicense it under the GPL (as in the case of BSDL'd software).
A lot of pragmatic developers use the GPL/LGPL without necessarily agreeing with the GNU philosophy, or often even fully understanding it. Some might in any case feel (as Linus does) that the open source development model will lead to superior software in the long run. For those with such a view, the GPL v BSDL argument makes little difference, since they believe open source software will always be better in the long run. It's only those who believe proprietary software can add value open source can't who would insist on the BSDL. (Right now, I would lean more towards the view that both proprietary and open source development can add value, but I don't think using the GPL versus BSDL makes all that much difference.)
I've never read any convincing argument that Linux is more successful than BSD because of its licence. Linus has said he only wrote Linux because a free BSD was unavailable (owing to the USL lawsuit). A lot of early Linux users undoubtedly used Linux instead of BSD for the same reason (the USL lawsuit prevented them using BSD), and after that it was largely a matter of network economics.
Given the broad success of OpenSSH, the X Window System, Mach, BSD sockets, Apache, et al, it's abundantly clear that the BSDL (or similar) doesn't prevent open source software being successful, or prevent corporate use of it. I think licensing ideologues on both sides (GPL and BSDL) tend to overestimate rather severely the importance of the licence. Things like meeting the needs of users and network economics matter a lot more than BSDL v GPL.
Actually, a better way of putting it is Tridgell was able to closely observe and record the workings of one of the cars as it was being used by one who had agreed not to copy it, either with or without permission (with, as far as I know), but with the express goal of copying it. As I said before, there is only so much value in an analogy, and I've explained what I meant anyway, so my analogy is redundant.
I never said Microsoft Office overtook Lotus and Wordperfect unfairly (although there has been comment on Microsoft using hidden APIs in their Office software to their advantage). Liebowitz's analysis is true in this respect. Microsoft just unfairly closed the market once they overtook the competition.
In that case, there was a miscommunication. I tend to agree that Microsoft didn't unfairly overtake 1-2-3 and WordPerfect (based on what I've read), and I don't know enough about the tactics Microsoft used to maintain its dominance in the office-suite market to hold a meaningful opinion about whether or not they were unfair, or likely had a significant economic impact.
It is my understanding that Tridgell was not using BitKeeper in any way but was observing the on-line protocol by tapping traffic between end-points. Which negates your whole premise. Noone has "lent" him a car. In your analogy, Tridgell was standing on a street and took photos of the cars as they went by following which he proceeded to make his own compatible one.
It only falls apart if Tridgell was observing the BitKeeper traffic without the permission of one of the licensees. If he had permission, then access to the software (in the sense of the technologies it implements, not the source code, of course) was given to him (ie it was lent, even if copies of the blueprints weren't given out). If he was eavesdropping without the concent of the licensed users (I don't think this was so), the more appropriate analogy would be that Tridgell stole one of the cars.
In any case, the usefulness of such analogies is limited. The only point is that Larry was not asking Linus or anyone else using his software to pay him, or to otherwise buy 'Larry gas' or anything else from him. He was allowing them to use his software for free, because it was beneficial both to him and to the Linux pragmatists (but not to the GNU ideologues, for ideological, not technical, reasons).
Actually no. It was anything but practical (although it was beneficial to Linus personally). Linus dropped the ball on this one, his personal friendship with Larry and subsequent infatuation with BitKeeper went against the grain of the whole Linux development process and was causing major disruptions and upheavals in the mailing lists and amongst the developers. Linus tried to enforce his personal preference upon everyone else, disregarding a significant portion of the kernel developer's opinions. Add to this the fact that BitKeeper license was one of most attrocious, outrageous and megalomaniac pieces of self-ego-stroking crap masquarading as a license to ever hit the FSF world. Including provisions like demands that noone who uses BitKeeper is allowed to work on any competing product even if they are not using BitKeeper for that purpose.
The software allowed for co-ordinated revision control of the Linux kernel. This is the practical use for which Linus wanted it. If its use led to ideological conflicts between GNU ideologues (like Tridgell) and Linux pragmatists (like Linus), this was an ideological problem, and not related at all to the suitability of BitKeeper for tracking revisions to the Linux kernel.
Linus made an error and he paid for it by losing a significant portion of his stature and respect people had for him. After the fiasco blew up in his face he managed to recover somewhat by quickly moving to disprove his own earlier claims of insurmountable difficulties of automating kernel source management by creating the "git" system which is by now superior to BitKeeper in most respects.
You agree with the GNU ideologues. A lot of other people disagree with them, so there's no point in claiming one opinion or the other is the right one. The facts are: (a) BitKeeper was technically suitable for revision control of the Linux kernel; (b) it is no longer used because of ideological, not technical or economic, reasons; (c) the development of git has imposed an opportunity cost, in that the time Linus and others have used to develop it could otherwise have been used to improve Linux itself.
But a lot of people will remember this massive screwup which exposed Linus for his lack of principles and will no longer trust him to make the right call as they did before.
Linus has never claimed to be a GNU ideologue, or to believe in Stallman's GNU philosophy. The principles you refer to are the principles of that philosophy, not general principles normal people believe in. Since Linus has never adhered or claimed to adhere to the GNU philosophy, it should not surprise you that his principles are not the same as you
On a company level, it would of course be unwise to give them up, but on a national level it could be done without requiring all countries to do it at once, and without any action from companies (since their software patents would simply become invalid).
A better analogy (sorry! ;-) ) would be:
Larry let Linux and his friends borrow early models of a new car he was designing, helping them by letting them drive the cars, and helping him because they told him how they liked them, etc. It was agreed that they would not try to copy the car design.
Someone lent one of the cars to another friend, who hadn't agreed not to try to copy its design, and he immediately began to do so. Larry reacted by taking his cars back, and so Linus and his friends had to design and build their own, which is something they hadn't wanted to spend their time doing.
If your perspective is the GNU/FSF ideology, then of course you will consider it better not to use Larry's cars at all. However, not everyone shares that ideology, so to some of the others, it was better to have nice cars for free, without having to bother designing and building them.
I'm sure there are problems with this analogy too, but the basic point is the agreement between Linus and Larry was mutually beneficial (in practical terms, not GNU/FSF ideological terms) until some people who disagreed with it for ideological reasons spoilt things.
Being interested in this from the perspective of economics (I'm not interested in software advocacy), I'd like to read the views of opposing economists, if you know of any. Most of the ones I know of have written about MS Windows, not MS Office.
The most striking thing for me in this market is the dominance MS Office achieved on the Apple Mac, when Microsoft was still far behind its competitors on the PC (still DOS in those days). A coherent explanation of why this happened, if Microsoft's products weren't technically superior to the competition, is something I'd really like to read.
As for the prices, I've read economic analyses which show the prices for Lotus 1-2-3 and WordPerfect were higher than those for Excel and Word, respectively, until 1995. By 1995, Word and Excel had achieved 70-80% market share, so the moderate price reductions by Lotus and WordPerfect (for products which were still more poorly rated than Excel and Word) weren't enough to stop their decline. Steep price falls to 1-2-3 and WordPerfect only came in 1996 and 1997, and they never achieved technical superiority. With tiny market shares and technically inferior products, it's easy to see why low prices weren't enough to save WordPerfect and Lotus, but only starved them of the income needed to develop their declining products.
By the way, can you provide a link documenting a switch by Microsoft from per-processor licensing of Windows to per-processor licensing of Office? The original anti-trust case against Microsoft (over abuse of Windows to undermine Netscape) was filed in 1998, long after WordPerfect and Lotus 1-2-3 had collapsed. A licensing change after 1998 couldn't have had anything to do with Word and Excel gaining market dominance, since this had happened some years earlier: by 1997, both products had more than 90% market share in revenue terms.
It wasn't listed in the invoice? In my experience, computer sellers always try to convince customers buying PCs from them to add software bundles and other things, but except for very cheap software and some operating system, these bundles are always optional items.
To use a real world example, do you think there is any way for Red Hat to prevent SuSE/Novell buying a single copy of Red Hat's most expensive product, and copying from it any code they find useful?
In this case, investment would be diverted from software development to services development. The end result might be production of a similar amount of overall value, at least initially, but even if this were so, the software would still see less investment and a slower rate of technological development. This would only be offset by a more highly developed services market. Do cheaper services represent technological progress? Would continued development of the services side of the business be sustainable, and lead to a similar level of overall value production to proprietary and BSDL'd software develepment? I don't know the answer to either question.
You're still looking at the situation from the traditional perspective. You can't look at it as two competing software companies A and B, which are reluctant to invest large chunks of effort because the other will free-ride. Instead, consider hundreds of firms that use a particular piece of GPL software while each contribute small portions back to the shared codebase. The incentive is not to profit on software itself but to meet their own needs. Sharing improvements just makes the process easier and ultimately cheaper. All of those contributions add up over time and the software evolves through user-driven development. How that development occurs is variable. Some of it may be in-house work. Some of it may be contracted through core project leaders. Some of it may be a side-effect of hired support services.
It doesn't matter how many firms there are, or whether or not software production is their primary business. If the benefits of investment accrue to all users, those who fund the investment will be put at a competitive disadvantage, and the market will push down levels of investment. An example of this is the PC market, where Dell has gained its position by explicitly doing as little investment in development of the PC platform as possible. This frees up resources for manufacturing, marketing and distribution.
That's a reasonable consideration because Open Source does, indeed, largely eliminate the potential for monopoly profits. (services markets can be monopolized as well.. it's just much harder.) On the other hand, I do not believe that new software technology is primarily produced by large investors. Software, like science, is too evolutionary of a field. It's hard to force progress by simply throwing money at any one area -- in fact this can often result in "solutions in search of problems." In the early days of computing, large investment was needed to jump-start the industry; today, there are so many players that innovation happens automatically amidst the mundane. As a developer, I don't care about monopoly profits. I only care about making a reasonable living while doing what I enjoy. I am paid to meet needs but I innovate in the process because it makes the jobs more fun.
It's not a monopoly situation, but rather monopolistic competition, which means each producer attempts to differentiate its products from others by adding value not available in them. This is the norm for most goods markets. Perfectly competitive markets (or their closest real-world approximations) are found in areas like agriculture, where individual producers have no power to influence prices.
If the services market allows for differentiation, then this could o
Apart from word processing, there are also a lot of things I can imagine they would use Excel for, and even PowerPoint. They could also be using Access, but there are much better databases available, so I don't think they would be using MS Office for that.
I don't much like Word, but Excel is something I use a lot, and no other spreadsheet I've tried is good enough to convince me to use it instead. Besides, everyone I know uses Excel, so why bother with some other spreadsheet?
Stan Liebowitz (a professor of economics at the University of Texas) makes a fairly convincing case that Word and Excel succeeded because they were better than the competing products. Both were market leaders on Mac before PC, so those who think Microsoft cheated have to come up with an explanation of how it did so on Mac.
I think we shall have to agree to disagree on what altruism means. :-)
I find this interesting, since economic theory would predict that investment in Linux made by Red Hat would benefit all Linux producers, and therefore leave Red Hat at a competitive disadvantage. Two possibilities that come to mind are:
1) Funding development of Linux allows Red Hat to provide better support than competitors who do not fund its development.
2) Red Hat is benefiting from its investment in the short run, but as other firms catch on, this will be eroded and its levels of investment will have to fall to maintain profitability.
I don't know which it is, but if it's the first one, that suggests a GPL-dominated software environment can still leave room for differentiation. That's good in the sense that it makes investment attractive, but if it is the case, it also means the GPL won't prevent domination of a given market by a single firm.
You're playing a game of semantics. If I say the the RAM industry has been commoditised, it means exactly the same thing as saying that RAM has become a commodity. What this means is that there is no meaningful differentiation between RAM chips from different producers. In an entirely GPL world, there would be no meaningful differentiation between software from different producers.
Investment in software itself does not need to be profitable on its own for the incentive to exist. Yes, we may eventually see the end of traditional software companies where all development is productized and centralized under one roof. Regardless, software companies will continue to become more diverse and fewer will take on the whole development load by themselves. Investment in software itself will be more driven by real user needs, and meeting those needs properly will always be profitable.
Without a sufficient return on investment, new investment will not be made. If firm A invests in software to meet customer needs, but firm B can benefit from that investment too, it is obvious that firm B will have a competitive advantage, and thus the trend within the sector will be towards ever lower levels of investment. After all, there will always be alternatives to invest in if software development becomes unattractive.
This is pure speculation. First, the total amount of investment in software is not what matters; the rate of meaningful production is what counts. Today's software industry is incredibly inefficient. If Open Source causes a massive increase in efficiency, due to code sharing, it is expected that investment in software itself may drop; but the total useful output could actually increase greatly. At the same time, Open Source dramatically reduces barriers of entry into the market and this has the opposite effect on investment. Likewise there is no reason whatsoever to believe that technological progress would likely slow. Most of the investment in today's software industry does not go into improving the state of the art; it goes into reinventing wheels. In terms of resource allocation, services and support are intimately tied to software development within the Open Source model. Companies that have tried to separate the two have generally failed.
Open source with a GPL-like licence moves the software market from a market structure resembling monopolistic competition to one resembling perfect competition. Perfectly competitive markets are more efficient, yes, but they are also not profitable in the long run (which is why, for example, agriculture is subsidised). Under market structures more resembling monopoly, there is the possibility for profit in the long run, and this profit can be used to invest in new technology. One obvious example of the effects of monopoly investment is the development of Unix at AT&T.
GPL open source will lead to greater efficiency, yes, but this greater efficiency will reduce the opportunity for profit, and will drive many producers out of the market, especially those which invest heavily, since the benefits of such investment accrue to all producers, not only the ones funding the investment. The expected result, according to economic theory, will be cheaper software, but also lower rate technological progress.
Ironically, what you described is precisely where the software industry was before Open Source entered the scene and started shaking things up. Thus far, Open Source has done nothing but stimulate progress and innovation. The only thing that remains is full-on
If I give some of my code to someone, under even public domain, how can they stop me either: (a) giving that same code to someone else; (b) giving away any other code I write? The answer is they can't! They cannot stop me being altruistic, even if they're entirely selfish.
You're confusing altriusme with a philosophy of achieving a 'greater good' through not being altruistic. If Richard Stallman had been motivated by altruism, he would not have been bothered that James Gosling benefited from his emacs work without sharing his own code, and would therefore have never written the GPL.
I prefer the BSDL to the GPL because I think it provides greater utility to a greater number of people: even users of closed systems can benefit if the BSDL'd code is good enough to be integrated into them. Supporters of the GPL argue that it will lead to greater utility in the long run because it will destroy proprietary software (which the believe provides less utility than open software). This is a position I disagree with (primarily because I don't believe the utility of software in general is a function only of its licence), but a sound argument can certainly be made in favour of it.
A competitor has only to buy one copy of the given software, in order to get full rights to the source code, and thus the right to copy anything valuable into its own product. In other words, any competitor that wants it can easily get it, so a firm is effectively required to give all changes to its own software immediately to all its competitors.
altruism: willingness to do things which benefit other people, even if it results in disadvantage for yourself
If your motive is altruism, why use the GPL rather than the BSDL or public domain? With the GPL, you are only willing to help others if they agree to help you. This is not altruism.
Because of the way markets work, a situation such as the one above would likely be broken by innovation in one place or another. It might be innovation in new proprietary hardware, or it might be innovation in new proprietary software. One way or the other, the steady decline in investment in commodity software would leave a stagnant market, ripe to be overturned by investment in alternatives in one area or another.
For firms whose business is based on meeting customer needs more rapidly than their competitors, the GPL is not viable, because under the GPL, any improvements made to satisfy customers must also be immediately made available to all competitors. Those who invest in R&D and technical development therefore lose out to those who invest only in marketing.