Stallman — 20 Years of Explaining Free Software

H4x0r Jim Duggan writes "The first recorded talk by Richard Stallman on free software was in 1986, so I've picked from the 2006 recordings and have made a transcript of a recent talk: The Free Software Movement and the Future of Freedom. Those two are the only transcripts of his general free software talk. Others that exist are on specific topics such as patents, GPLv3, copyright, etc. For those who've been reading Slashdot during the gradual evolution of Stallman's pronouncements, it's interesting to see what has changed over 20 years."

Comments

Re:Hard to explain

[kfg]

You see when people adopted the monkier of open source software. . .

So that Microsoft could exploit the ambiguity of the word "open" to claim that their software is open source? I'm afraid the word "open" is just as open to interpretaion as any other non-technical word.

Of course RMS provided a technical defintion of what he meant by "Free Software."

The reason a lot of people prefer to use "Open Source" isn't because the term "free" is ambiguous (although I recognize the existence of the "libre" crowd); it's because they the disagree with the specificity of the term. The definition of "Open Source" is more, ummmmmmmmm, "Open."

ESR, for example.

KFG

Re:Bleh

[Dik Zak]

what has HE done? He developed the original Emacs, GNU Emacs, the GNU Compiler Collection, and the GNU Debugger. That's a pretty serious contribution you know.

No

[orasio]

He will never get over it.
When mswindows 95 appeared, it wasn't called "the DOS system". It was the Windows system, running on DOS. Okay, that's too much of a stretch.
mswindows nt/2000 was not the "kernel32.exe".
OSX is not "mach + some apple stuff".

An operating system is a lot more than a kernel, in the same way that a car is a lot more than its engine, even when it doesn't work without it. The user doesn't get to interact with the engine, and the car would be the same car, if the engine is replaced. That happens the same way with Operating Systems and kernels. Debian is not there yet, but they have several GNU distributions with varying kernels.

Linux is a good kernel, and plays an important role for the success of free software. Aside from that, when you get for example, Ubuntu, there is a lot more GNU than Linux included in the CD. And the platform is defined by the GNU system, not the Linux kernel.
When people say they know "Linux", for example the "Linux" console, they are talking about bash. When talking about "Linux" programming, it's usually GCC, the "Linux" desktop might be Gnome or KDE, of course, but it's not Linux either.

The guy will never get over it, because, in that particular issue, he is right, and the people who think different from him are just wrong. There's no way he will change his opinion on that issue.

Re:No

[a.d.trick]

The guy will never get over it, because, in that particular issue, he is right, and the people who think different from him are just wrong. There's no way he will change his opinion on that issue.

I beg to differ. The term 'Linux' has gained a second meaning as a short form for 'an OS that uses the Linux kernel' which is almost always the GNU system with a Linux kernel. Language and words change so we can talk more efficiently. It happens all over the place in our language: 'refrigerator' became 'fridge', 'windows' instead of 'Microsoft Windows', even the notorious "where's the internet" is short for "where's the icon to open my web browser". Of course, it causes ambiguity and confusion sometimes, I have a hard time talking to new people about windows as in that box your graphical apps open up in, but that's the price we pay for shortening our language. In the end, it's all about efficiency.

I understand that RMS wants the extra publicity, and I think they really deserve it. Unfortunately, it's not going to happen unless you turn GNU/Linux into a two syllable word: people are too lazy.

Re:No

[swillden]

If Linux Torvalds hadn't got involved in software RMS would have a following of academic lisp gurus numbering nearly in three digits.

I doubt it. If Linus hadn't done what he did, I think there would have been another kernel by the mid-90s. Perhaps it would have HURD (I think the availability of Linux slowed HURD development), or perhaps it would have been BSD, or perhaps it would have been something else. Linus' contribution was important, but the kernel is one of the smaller components in a full operating system.

If Richard Stallman hadn't got involved in software Linux would have a different compiler.

And a different license. If RMS hadn't started GNU, Linux would have had a BSD user environment, and probably a BSD license. It's hard to say what the impact of that would have been. It seems clear that a BSD-licensed Linux wouldn't have gotten all of the corporate participation that the GPL-licensed Linux has.

Without GNU, I also think Linux would have been delayed for a few years, because it would have been necessary to either write all the user space tools or wait for the BSD settlement to legitimize the BSD stuff.

Getting back to the question of the compiler, I wonder what Linus would have used if GCC weren't available. What were the options for a poor college student in 1991? I was a student at the time, and I know that the compilers available to me were Borland's Turbo C and compilers from OS vendors, including Microsoft, Sun, HP and DEC. Borland's was the the most accessible to students, because of their education prices, but neither it nor Microsoft's compiler would have run on Linus' fledgling new OS, unless it provided a DOS-like kernel interface. The others were really expensive. The BSD and Minix compilers were around, but I'm not sure if he could have used either of them legally.

Perhaps Linus would have had to write a C compiler as he was writing his kernel? I really don't know the answer to these questions.

Speculating about how Free Software history would have changed with either RMS or Linus removed from it is complex and difficult. There were a lot of interrelated factors.

Re:Will Stallman ever get over this?

[AlanS2002]

GNU however could be replaced with something else. e.g. the BSD userland/libraries. Would we then be obliged to call the operating system BSD/Linux?

Re:Will Stallman ever get over this?

[AlanS2002]

Or you could just do what most of the human population does and call it something simpler. i.e. Linux

RMS' rationale condensed

[FallLine]

The idea of owning information is harmful in three different levels. Materially harmful on three different levels, and each kind of material harm has a corresponding spiritual harm.

|SNIP|

The first level is just that it discourages the use of the program, it causes fewer people to use the program, but in fact it takes no less work to make a program for fewer people to use.

|SNIP|

The second level of harm comes when people want to change the program, because no program is really right for all the people who would like to use it. Just as people like to vary recipes, putting in less salt say, or maybe they like to add some green peppers, so people also need to change programs in order to get the effects that they need.

|SNIP|

The third level of harm is in the interaction between software developers themselves. Because any field of knowledge advance most when people can build on the work of others, but ownership of information is explicitly designed to prevent anyone else to doing that.

That is it folks. In other words, his argument is closed source software is wrong on pragmatic grounds because:

A) fewer people will use the software (because it tries to prevent people from using w/o paying)

B) the software is less useful to people because they can't modify the original program

C) proprietary software is less valuable because other developers in lateral areas can't learn from it.

It seems pretty clear to me that his arguments failed on these pragmatic grounds and that he's had to shift his anti-ownership rational to far more nebulous and entirely philosophical arguments about "freedom" for its own sake.

The facts are:

A) Contrary to his "first level" of harm: proprietary software has vastly outcompeted open software despite its barriers.

B) Contrary to his "second level" of harm: that most users still prefer closed source software despite the fact that they can't tinker with it and despite the fact that it costs more/has more barriers.

C) Contrary to his "third level" of harm: that proprietary software still appeals more to its end users despite the fact that proprietary developers benefit little from the pool of open source code. This despite the fact that open source developers supposedly have a huge advantage over proprietary developers because they can exploit the GPL and other copyleft code to a level that their counterparts cannot.

In short, he's given up on his pragmatic rationale since they've been proven almost entirely wrong. I'll concede that there is something to be said for the sharing of code in some cases, but we're to choose rationally between no ownership vs choice of ownership (the status quo) that the latter is the only sensible and pragmatic choice given his own (old) arguments and the empirical evidence (or lack thereof) from his so-called copyleft movement.

Re:RMS' rationale condensed

[daigu]

Let's do a thought experiment shall we? Let's assume your rendering of his argument is correct and let's change "software" to "information" - as a concrete example, newspaper information available in sources such as the New York Times (NYT), Wall Street Journal (WSJ) and free (no cost) versions available through Yahoo or other services.

Fewer people do use the WSJ versus the NYT. It costs money to get the WSJ. NYT requires registration. Now compare Yahoo and other sources that have no cost and no barriers such as registration. What gets used more? What is a more competitive product?

It is clear that WSJ is referred to and used less than free or registration only services - which basically supports his first point. The competitiveness - however you define it - is besides the point.

I think you can make the argument that the Wikipedia, the trend for online publications to provide discussion forums attached to specific articles and so forth basically supports Stallman's second point. These resources are more useful because they can be updated in a timely fashion and errors and corrections can be made. Compare that to the old newspaper model - which works much like proprietary software and where the publisher can publish bug fixes in the form of "corrections" on a page no one sees.

However, I think his strongest point is the third one. Proprietary information is less valuable because people in lateral areas can't learn from it. The best example for these and newspapers is the ability to aggregate them. Let's say you are doing research on a topic and want to be able to do a search across the NYT, WSJ, Yahoo free services like AP Newswire and so forth. Right now, there is only one service that provides this capability - Factiva - which owns the WSJ. If you used a service like Nexis, you would not be able to search the Wall Street Journal as well. Factiva itself has troubles keeping other sources in their database like the Financial Times.

I can tell you that this has negative effects on the business decision making ability of organizations because they cannot look at all the relevent press coverage on a topic. The ability to do this kind of search is contingent on companies being willing to license their aggregate content so that it can be searched through one source. The more restrictive and proprietary the information becomes because the companies that own it won't license it, the less useful these aggregating search database becomes and I would argue it has a negative impact on business overall.

Now, I think you can basically make many of the same arguments when you change "information" back to "software". I think your premise that proprietary software has outcompeted free software is questionable at best. Based on what metric, presence on the desktop? Code quality? Anywhere you look you don't have a particularly strong argument. I also think that you that advantages of free software are ones that are realized over time that you are not accounting for.

You second point about users is also a bit dubious. His argument is geared toward developers. Do most developers prefer closed source software and how has that tracked over time? I'd argue that it has increased, but I don't have a source handy that supports me. I also think that as companies attempt to assert more control over the desktop, you will see users making choices about using products that don't unnecessary restrict them over those that do. Again, this is something that will manifest itself over time.

I'm not sure I'm following the last part of your argument. I think you are using users when you should be talking about developers. The ability to reuse code mostly impacts the development cycle. Users will ultimately follow a development cycle that gives them tools that enable them to use their computer the way they want to use it.

You can see the impact today in product releases like IE7. You think IE7 would look and work the way it does without Firefox blazing the trail? You think this might b

If only he could count

[gr8_phk]

I appreciate most of what RMS says. I strongly disagree with his numbering scheme for the 4 essential software freedoms. Read people count starting at 1. It's stupid to have the leader of a movement use an inside joke when giving a public talk about something so important. Freedom zero.... How stupid.

Hey Richard, how many freedoms are there?
Four.
What's the fourth one?
There isn't one... Only a zeroth through third.

This nonsense has got to stop. The GPL is fairly readable, but this stupid geekism right there mixed in with the fundamental freedoms is IMHO just adding confusion where none needs to be. I would hope this renumbering will make it into GPLv3.

Just a few counterpoints.

[Spaceman40]

A) Contrary to his "first level" of harm: proprietary software has vastly [out-competed] open software despite its barriers.

Thought experiment: if somehow, suddenly, Linux closed all the sources and took a non-free license, would they gain or lose users? If somehow, suddenly, Microsoft opened the sources of Vista under the GPL (or BSD, or whatever), would they gain or lose users? Correlation vs. causation and all that.

B) Contrary to his "second level" of harm: that most users still prefer closed source software despite the fact that they can't tinker with it and despite the fact that it costs more/has more barriers.

See the thought experiment. Take Photoshop: if you offered a user the choice to take Photoshop with no access to the source vs. complete access to the source, what do you think that most people would choose, all other things (including price) being equal?

C) Contrary to his "third level" of harm: that proprietary software still appeals more to its end users despite the fact that proprietary developers benefit little from the pool of open source code. This despite the fact that open source developers supposedly have a huge advantage over proprietary developers because they can exploit the GPL and other copyleft code to a level that their counterparts cannot.

The third level of harm doesn't have much to do with the end users, anyway. However, to continue the thought experiment: in an office environment, would you rather have open-source printer drivers you could get tech support to fix on-site (or vendor patches if already fixed), or closed-source printer drivers that require vendor support?

I think that your arguments focus on the wrong side of the point. Proprietary software is popular, true. That doesn't mean that open sourcing it would make it less popular.

Re:Hard to explain

[TheRaven64]

The reason a lot of people prefer to use "Open Source" isn't because the term "free" is ambiguous Unfortunately, so is 'open.' Look at the different definitions of 'open specification' you have:

1. Anyone may read this specification and implement it.
2. Anyone may read this specification, but there are conditions on implementations.
3. Anyone who gives us a load of money can implement this specification.

The same is true of 'open source.' There is a legal definition on the OSI site, which is very long, and far less concise that the FSF's four freedoms, but without it the term is highly ambiguous. Is Microsoft's Shared Source initiative 'open source?' You and I might know that it isn't, but anyone can download the source and read it, so it certainly sounds open.