A while back, Freenet Africa had a nice background piece about software luminary and founder of the software freedom movement, Richard Stallman (aka RMS). The article covers his background starting with the GNU project and following through to the current, ongoing fight for digital freedom.
A Rebel with a Cause
Imagine a world where every time you want to share a cool app with a friend, you have to ask permission (and maybe pay extra). Or where fixing a simple bug in your game is impossible because the code is locked away like a secret recipe. Sounds like a tech dystopia, right? This is exactly the kind of world Richard Stallman set out to prevent. Stallman – often known just by his initials RMS – is not as instantly famous as tech giants like Bill Gates or Steve Jobs, but his impact on our digital lives is monumental. He's the mastermind behind the GNU Project, the founder of the Free Software Foundation (FSF), and the author of licenses that guarantee software freedom. In short, he's the original software freedom fighter, a kind of digital rights Gandalf (yes, with the beard to match). And for a guy who champions "free" software, he's quick to tell you: we're talking free as in freedom, not just free as in price.
In this essay, we'll dive into Richard Stallman's contributions to the digital world in an engaging (and occasionally humorous) way. By the end, you'll understand how his work laid the foundation for Linux and the whole open-source ecosystem, why he insists on calling it "GNU/Linux," and what the internet might look like if Stallman hadn't started his crusade for software freedom. Grab a snack (maybe some free-as-in-freedom nachos?) and let's explore the world of Stallman and the movement he started.
Who is Richard Stallman? (And Why Should You Care?) [...]
As others have pointed out, the freedom is the start of a journey, not the destination.
Previously:
(2022) The Code: Story of GNU and Linux (2001) Complete Documentary
(2021) Richard Stallman Rejoins Free Software Foundation Board of Directors
(2018) RMS on a Radical Proposal to Keep Your Personal Data Safe
« NASA Releases Starliner Failures Report as It Preps for March Launch of Artemis 2 | Pink Noise Reduces REM Sleep and May Harm Sleep Quality »
Related Stories
Richard Stallman writes in the Guardian:
Journalists have been asking me whether the revulsion against the abuse of Facebook data could be a turning point for the campaign to recover privacy. That could happen, if the public makes its campaign broader and deeper.
Broader, meaning extending to all surveillance systems, not just Facebook. Deeper, meaning to advance from regulating the use of data to regulating the accumulation of data. Because surveillance is so pervasive, restoring privacy is necessarily a big change, and requires powerful measures.
The surveillance imposed on us today far exceeds that of the Soviet Union. For freedom and democracy's sake, we need to eliminate most of it. There are so many ways to use data to hurt people that the only safe database is the one that was never collected. Thus, instead of the EU's approach of mainly regulating how personal data may be used (in its General Data Protection Regulation or GDPR), I propose a law to stop systems from collecting personal data.
The robust way to do that, the way that can't be set aside at the whim of a government, is to require systems to be built so as not to collect data about a person. The basic principle is that a system must be designed not to collect certain data, if its basic function can be carried out without that data.
Richard M Stallman, founder and former president of the Free Software Foundation (FSF), announced at the organisation's LibrePlanet virtual event that he has rejoined the board and does not intend to resign again.
Stallman spoke at the event yesterday on the subject of unjust computing – covering locked-down operating systems, non-free client software, user-restricting app stores, and more.
Before the talk he stated: "I have an announcement to make. I'm now on the Free Software Foundation Board of Directors once again. We were working on a video to announce this with, but that turned out to be difficult, we didn't have experience doing that sort of thing so it didn't get finished but here is the announcement. Some of you will be happy at this, and some might be disappointed, but who knows? In any case, that's how it is, and I'm not planning to resign a second time."
Ars Technica further notes:
Video of Stallman's announcement is available at It's FOSS News. Stallman gave a talk at LibrePlanet yesterday on "growing injustices in computing," including "locked-down operating systems; user-restricting app stores; [and] requiring nonfree client software, including Javascript."
Previously:
Richard M. Stallman Resigns
Richard Stallman Deserved to be Fired, Says Fired GNU Hurd Maintainer
I was browsing my media and decided to rewatch this, as I hadn't looked at it in fifteen years or so.
I was mainly struck by the unalloyed optimism of pretty much everyone who contributed, including Linus, Richard Stallman, Eric Raymond, Alan Cox, Ted T'so, Eric Allman and many other original neckbeards (I use that appellation affectionately, and in a bunch of cases, literally).
In the 20-plus years since the film was released, much has changed.
I think much of the optimism embodied by RMS and the FSF has waned a good deal (and more's the pity), and the complete reversal of Microsoft from Ballmer's "Free Software is communism" to Nadella's embrace of GNU/Linux in both Azure and WSL, to the co-opting of Linux for Google/Android, as well as aging and slow drift towards retirement/death/irrelevance of those who championed Free Software for nearly four decades have really hurt the movement, while boosting Open Source.
I think that refocusing on "free as in beer" instead of "free as in freedom" across the development community may have been inevitable as GNU/Linux (although I guess it could have been GNU/Hurd or one of the BSDs) became mainstream a couple decades after the commoditization of IBM PC-like hardware.
That got me thinking, where does that leave us and "who are the new neckbeards tht can carry the vision of Free Software into the middle of the century?" Are there really any such folks with the passion and drive to champion Free Software moving forward?
Or is Free Software (as originally defined and advocated for by RMS and the FSF) dying a slow death in favor of "Open Source" and more permissive licenses like MIT and Apache?
What will Open Source look like in 2050, 52 years after Bruce Perens and the OSI's Open Source definition?
(Score: 5, Insightful) by jelizondo on Sunday February 22, @11:36PM (8 children)
I have a lot of respect for people who live by their principles, even if I don't agree with them. I'm not saying I don't agree with Stallman, I'm saying that I would respect him regardless plus one has to recognize that he has done a lot of good shit for the community, not just us nerds, the world at large.
Respect.
(Score: -1, Troll) by Anonymous Coward on Monday February 23, @01:49AM (7 children)
You don't need to respect someone who defended Epstein and knows the age of consent in every state. And if you do, a lot can be reasonably assumed about you.
(Score: 5, Funny) by Anonymous Coward on Monday February 23, @02:35AM (4 children)
pedo panic! fear! dooom! stallman committed some heresy and he is unperson now.
(Score: 0) by Anonymous Coward on Monday February 23, @05:07AM (3 children)
In which court was he found guilty of any crime?
(Score: -1, Flamebait) by Anonymous Coward on Monday February 23, @04:27PM (2 children)
the court of public opinion on the internet.
(Score: 3, Touché) by c0lo on Monday February 23, @10:58PM
Unfortunately, that is true.
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 3, Touché) by Bentonite on Wednesday February 25, @03:29AM
Indeed, that court is well known for accepting false claims as gospel without checking the facts.
(Score: 0) by Anonymous Coward on Monday February 23, @02:36AM
are out in force I see
(Score: 3, Insightful) by pTamok on Monday February 23, @10:18AM
Replying to a troll-modded post.
Is RMS a bad man? I do not know. Has his work been useful. In my direct experience, yes. Should my views on him personally affect my views on and use of his work? That's the question, isn't it? Opinions differ on the answer.
(Score: 4, Funny) by krishnoid on Sunday February 22, @11:51PM (23 children)
i mean, he worked at MIT's Artificial Intelligence Laboratory [freenetafrica.com] but now he poops on it [stallman.org]. Maybe they're not the same thing?
(Score: 5, Informative) by JoeMerchant on Monday February 23, @03:07AM (19 children)
RMS:
I call the argument over "is AI intelligent" an issue of semantics, as he alludes to: "in some domain" - well, in some domains AI is far more capable of having a productive conversation with me than virtually any human I have ever met, granted: these are quite limited domains, but... Once upon a time in the 1990s we decided that we wanted to try a "least squared error fit" of some noisy data to a 5th order polynomial - this worked because the starting and ending points of the dataset were: A) known, B) expected to be in the result, and C) known to have slopes of 0. So, there's two values known and two derrivatives known, leaving two degrees of freedom, theory says: it can be done, just like fitting an mX + b line to a dataset. Off to the races: I spent a whole day crunching through the algebra to make an algorithm for the 5th order polynomial least squares fit... then I spent half the next day debugging the software to implement it and see it work. Claude? 5 minutes, and I'm pretty sure you _still_ can't look that one up on Google.
Claude is writing something of a CAD floorplanner for me - with OpenSCAD 3D output, svg output of the 2d plan variants, all kinds of geometric solutions for tangency constraints on arcs and lines, and now that we're deep in I decided I want to be able to rotate the whole thing by arbitrary amounts in the definition - make a primary wall that's now at an angle become the reference East-West wall of the structure - so rigidly rotate everything that has been defined so far - much of it in terms of cardinal directions: this is East of that, North of the other. It's a lot to fix, Claude screwed it up yesterday, but today I've told it about unit tests for relative distances between points - so it knows when it screws up - and now it tries, fails, tries again, fails better the next time, and I believe we are converging on a solution that will rotate this existing design with hundreds of defined shapes in it.
Asking Claude about appropriate lintel sizings for rough openings? Insane - don't go there - different answers every time, they sound good most of the time, but relying on that is just beyond dumb. Bullshit generator with indifference to the truth absolutely applies, because the test for correct doesn't really exist. Math, geometry? Those tests are perfect.
The public misplaces trust in everything, all the time, AI's just the newest in a line of things the public shouldn't trust like they do, but they do....
I may not have a copy of Claude, frankly I don't want a copy of today's Claude - month by month it is improving, noticeably. I have Claude write software for me, I _do_ have copies of that which I can run and modify any time I want. I have critiqued that software many times and had Claude re-write it to be more easily understood and modified by me... I actually feel quite liberated, I have used various software over the past 30 years to do floorplans and construction plans, including stuff like AutoCAD's Architectural Desktop - they were all somewhat limited in many ways. When I wanted to transpose my building outline onto a .pdf survey as a site plan, Claude just cobbled together a python script to use some PDF tools (including OCR so it could "read" the raster scan PDF I'm overlaying on) and now when I change the shape of my structure outline, the site plan updates with the new shapes and sizes automatically, using a script that I control. Of course, I'm still developing these scripts as I go, and it's much easier to say "Hey Claude, look at this webpage, make a rectangle representing a sofa and place it facing that window 10' away, and make that shape in the .svg a link to the webpage." than it is for me to do all that myself. Could I? Sure. Is it worth my time and more importantly: effort, to do all that by hand vs a chat prompt? No, it would take me 5x as long and 50x as much brain-burn to do that myself. I'd much rather "Hey Claude" to have it done and walk away for 5 minutes while my non-intelligent pattern matching text generator does it for me.
🌻🌻🌻🌻 [google.com]
(Score: 2) by krishnoid on Monday February 23, @03:24AM (5 children)
Is Claude not capable of multiplying the coordinates of all the items in the scene by a matrix to translate/rotate them? Or are their coordinates or structure not stored like that?
(Score: 4, Interesting) by JoeMerchant on Monday February 23, @03:42AM (4 children)
Claude is capable of virtually anything you ask of it in a math domain.
The problem from where I stand at the moment is that I spent a week off and on defining a floorplan with lots of "N-S wall" "E-W dimension line" "place this desk 2" S of this wall and 15" west of this lamp. So, Claude took shortcuts, just did Northing = Northing - 2" and such things, it's far simpler than defining a rotation for every object, passing in a rotation and X,Y coordinate for every argument, etc. And now we're going through unwinding all those shortcuts, because I wasn't specific up front about the need to rotate in the future (frankly, it didn't occur to me at the time) and we were just roaring along with the quicker easier simpler solution without any troubles, but now I've got svgs for 3 furniture layouts, a max span graphing tool, a conversion of wall geometry to 3d printer tool path script, the aforementioned pdf overlay generator, an overlay with field survey data points illustrated in relationship to the proposed structure, etc. etc. etc. and unwinding all the shortcuts in all the things is... a lot.
Yesterday I gave up telling Claude "just fix it" because it was creating layers on layers on layers of kludges. So I tore it down today, told it to redefine everything to be ready for rotation, and it said it did, but it only actually did about 80%, so now we're doing the other 20% - but... because I had it put in a bunch of unit tests to verify proper relative distance of every point of every thing to four other reference points in the geometry (529 tests in total at this point), now it knows immediately when it screws up and either tries again automatically, or I can intervene and make suggestions for how better to go forward.
🌻🌻🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Monday February 23, @04:16AM (3 children)
From the sounds of it, you had fun doing this by vibe coding, no harm there, it's your time.
But, please be honest now and compare the amount of time you spent all week, vs. the time you would have needed to do the same job with a simple 3D architectural CAD program...which can also generate dimensionally correct drawings, as well as rotate them any which way.
Some of this software may already be customized by, for example, furniture companies, with the ability to move their products around in a room and get perspective renderings. When we were getting quotes on redoing our kitchen, one of the bids included a link to a site that showed our kitchen with the company cabinets all installed.
(Score: 3, Interesting) by JoeMerchant on Monday February 23, @05:20PM (2 children)
>please be honest now and compare the amount of time you spent all week, vs. the time you would have needed to do the same job with a simple 3D architectural CAD program
I haven't used the Pro versions of architectural CAD programs in approximately 20 years. I'll go ahead and assume that parametric modeling has continued to improve from where it was back then, and that support for things like arc sections as walls has equally improved.
For this one project - I'm going to say the time spent is equivocal. The important part of the project is defining what it is I want to do, the tool is just a way to help me get that figured out and recorded in a form that I can pass to a professional architect (who I am virtually required by law to hire, or go through a decade of school and apprenticeship to become certified myself) so they can do what they do in the tools they are familiar with which may or may not be the "simple 3D architectural CAD program" I designed in, and even if it is the same exact version of the same exact software, their guys are just going to do a slapdash re-copy of what I hand them no matter what. They're not going to give any kind of discount vs napkin sketch transcription, none that I have ever encountered anyway.
I'm planning next on defining the lintels that go over my openings and having my tool calculate and illustrate the in-place forms to hold the poured columns under them. I want to do this to get the reality constraints in place while I can still swap furniture around inside the structure and have that push the walls around to fit different combinations of things with appropriate spacings between it all.
Like so many things: if I had 20 years of practice with the CAD software, sure, I could draw it faster than the AI. Could I also overlay on the survey? Yeah, but not faster than the AI. Could I get max span calculations for every possible building rotation out of it? Yeah, but not faster than the AI - until I had significant practice with doing that.
Amusingly enough, for price: Claude @ $200 per month gives all the tokens I'm capable of using. Autodesk Revit starts at $251 per month https://www.autodesk.com/products/revit/overview [autodesk.com] and there's still a learning curve which is a big up-front investment before you get to the kinds of "I can do that faster without AI" performance levels that professionals achieve. You could hire a professional who already knows how to use the tool, but just starting that relationship is a huge time sink, and the money is probably 10x more than DIY tools.
Should CAD jockeys start looking for other work? No. Does AI give me access to geometric design tools more efficiently than was previously available to me? Yes.
🌻🌻🌻🌻 [google.com]
(Score: 2) by krishnoid on Monday February 23, @05:56PM (1 child)
I *really* hope you journal, even roughly, the chronology of your experience with Claude and CAD, even if it just becomes a historical artifact. Reading how you prompt Claude to do what it does and how it does it over time is genuinely informative.
(Score: 4, Interesting) by JoeMerchant on Monday February 23, @06:50PM
I'm reminded of the quote attributed to Bill Gates in the early 90s IIRC: "Why is a $5 pencil and notebook so much easier to use than a $5000 computer running your software? Well, for starters: paper has had a few thousand more years of development."
I've been capturing the design evolution in a git repo, and one of the most practical things I have asked Claude to do with that is: After each prompt which results in a successful change (success judged by all unit tests passing), commit the changes to git with a 20 words or less summary of the change.
That's using git as Linus intended, lightweight, frequent commits - semi-descriptive descriptions to navigate them by, ability to roll back and branch off at any point.
🌻🌻🌻🌻 [google.com]
(Score: 3, Insightful) by Thexalon on Monday February 23, @12:09PM (3 children)
I would agree with that assessment: These systems are designed to produce a statistically plausible answer. There's a big difference between "statistically plausible" and "correct". Unfortunately, you need to have the specific knowledge of the subject at hand and be able to determine the correct answer yourself to be able to tell the difference between "correct" and "plausible", which means these systems do a great job of fooling those who don't know what they're talking about, a.k.a. upper management and investors.
"Think of how stupid the average person is. Then realize half of 'em are stupider than that." - George Carlin
(Score: 2) by JoeMerchant on Monday February 23, @02:25PM (2 children)
This is why they're actually pretty good at math - they understand the tests for correctness and can iterate until they get it right.
If you write software with requirements that read like a math proof, the LLM agents can get it to work - or at least tell you when they have failed utterly (which happens even when it's possible to get it right, but less and less as the agents have been improving).
Most software I have seen reads nothing like a math proof...
🌻🌻🌻🌻 [google.com]
(Score: 3, Touché) by Thexalon on Monday February 23, @06:50PM (1 child)
Reminds me of the classic Knuth line: "Beware of bugs in the above code; I have only proved it correct, not tried it."
"Think of how stupid the average person is. Then realize half of 'em are stupider than that." - George Carlin
(Score: 2) by JoeMerchant on Monday February 23, @07:49PM
>I have only proved it correct, not tried it.
Referring to the classic human flaw of specifying things that don't matter, and not specifying things that do.
We "know a problem when we see it" but are often incapable of concieving of important things that will go wrong in the future, while assuring ourselves that we "have all the bases covered" including a lot of things that are unlikely to ever happen.
I find the more I make the unintelligent pattern machines describe what it is they are summarizing, preferably with graphs and illustrations, the quicker I see the problems when they're present.
🌻🌻🌻🌻 [google.com]
(Score: 2) by Freeman on Monday February 23, @05:40PM (8 children)
The problem is that you're associating you putting in words into a program with you talking to a human. The program can be a tool or it can be a method of communication. It can't have a conversation with you. It knows nothing. Essentially you're using it as a glorified word search, albeit a very complex word search that takes most of the control out of your hands.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2) by JoeMerchant on Monday February 23, @06:44PM (7 children)
>It can't have a conversation with you.
That depends on your definition of conversation. You can have a conversation with a phone tree... not a very satisfying one as compared to "talking to a 'real' human" - but, with overseas outsourcing, they're starting to give you "real human" conversational experiences that approximate the satisfaction level of a phone tree.
>It knows nothing.
That depends on your definition of knows. Library card catalogs "knew stuff" that people didn't used to carry (completely) in their heads. You couldn't just talk to the cards, we had librarians for that. Then we had 25 years of typing stuff into search fields instead of looking through indexes and tables of contents, and slowly you started being able to talk to those search fields. The recent changes in pattern matching mean that you can be a lot more vague, ask for "sciencey stuff about monkeys and related animals" and get back information about primates, without having to figure out that primates was the word your were looking for. At some level, all these things "know things" that were encoded into them, either directly by Gutenberg arranging printing blocks in a press to make the information in a book easier to find, or by steadily advancing forms of algorithms which make "natural language" searching, like you used to do with librarians, more and more accessible to untrained people.
>Essentially you're using it as a glorified word search, albeit a very complex word search
Yeah, a word search that can take a conversation like:
Create a .pdf illustrating a scale drawing of a rectangle 3' wide x 2' high.
Fillet the corners with 6" radius arcs.
Tell me the area of the resulting shape.
Illustrate the area removed by filleting as a circle of the same total area as the four removed fillet corners, tell me the area of that circle.
Now, grow the height of the filleted rectangle until its area is equal to the area of the rectangle before the fillets were removed, plus the area of the circle.
Now, place the center of the circle in the center of the filleted and grown rectangle and subtract that from the grown filleted rectangle.
Create a .pdf file documenting these steps including shaded illustrations of each step with relevant dimension lines.
and much much more complex problems.
>that takes most of the control out of your hands.
As if "control" has ever been in your hands, the books either had what you were looking for or not, people you ask are full of shit as often as not. This is a search that chops all the books up into little bits and rearranges the bits as best it can to answer the prompts you give it - which is pretty much what we (used to) try to teach the little meat bags to do in school.
That example of documenting the steps with illustrations is important, because just like the little meat bags, the word search engines make mistakes (after all, they're trained on meatbag writings which are far from internally consistent...) so you, as user, need to be able to judge if the glorified word search is doing what you want or not, and the clearer you have it illustrate its process of getting to that final answer, the more easily you'll be able to spot divergences along the way.
🌻🌻🌻🌻 [google.com]
(Score: 2, Disagree) by Freeman on Monday February 23, @06:52PM (6 children)
You can talk to a brick wall. However, you will never have a conversation with a brick wall. Even, if you can hear the echo.
There are no thoughts or feelings being transmitted by the LLM. You can have no conversation.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 3, Informative) by JoeMerchant on Monday February 23, @07:53PM (5 children)
🌻🌻🌻🌻 [google.com]
(Score: 3, Informative) by Freeman on Monday February 23, @09:50PM (4 children)
We then upgraded to: https://en.wikipedia.org/wiki/Tay_(chatbot) [wikipedia.org]
ELIZA was created with typical programming and the creator of said program could implement bug fixes that worked, because he understood everything there was to know about his creation.
Tay was too complex and/or created in a very similar manner to modern chatbots like ChatGPT. Modern Chatbots like ChatGPT have too many datapoints for any one person to be able to figure out how/where/what to fix even simple errors. Yes, they do fix them, but they don't know the exact step(s) that led it down the wrong path. They can only infer, they can't know. This is partly why modern chatbots are so dangerous.
A conversation requires reciprocity that a chatbot can't have. There is no understanding. There is only parroting a string of things it has calculated you might want it to display. A useful tool, yes it can be. Do not misunderstand usefulness for intelligence.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2) by JoeMerchant on Monday February 23, @10:05PM (3 children)
>This is partly why modern chatbots are so dangerous.
People trusting chatbots (or other people, for that matter) are what make them dangerous.
>A conversation requires reciprocity that a chatbot can't have. There is no understanding. There is only parroting a string of things it has calculated you might want it to display.
Try some complex geometry - watch it "get it wrong" then explain what your really wanted, read its responses. It's pretty damn "conversational" - not exactly like talking with a person, but then talking with any two people is never quite the same, either.
🌻🌻🌻🌻 [google.com]
(Score: 2, Interesting) by Anonymous Coward on Tuesday February 24, @07:08AM
Apparently research shows AI gets worse as the conversation continues: https://arxiv.org/abs/2505.06120 [arxiv.org]
Of course that was in 2025. You can see whether starting anew with a modified prompt works better than telling it that it's wrong... I guess the difference might be not so much for one turn.
I'm not an AI expert but my guess is by the time an AI has rolled the dice to confidently give you the wrong answer, it's started on the wrong foot (populated a wrong context/state or something) and it's safer to reroll from scratch. It's like you were unlucky and spawned off a more retarded auto-complete chain, you can try type different stuff to try to get better auto-complete results, but I'd just restart a new auto-complete chain and hope I get luckier this time round.
(Score: 2) by Freeman on Tuesday February 24, @02:33PM (1 child)
Most definitely this:
AI companies are leaning hard into people trusting them. They want you to get hooked on them. OpenAI / etc. could have designed it in a safer way. Instead they've gone the move fast and break things route. Unfortunately some of those things are people.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2) by JoeMerchant on Tuesday February 24, @02:55PM
>They want you to get hooked on them.
Basic business: Coca Cola? And the opium trade before it?
Something I'm pretty happy with Opus 4.6 about is that it's doing less ego stroking than many of the models of the recent past. That whole "the user has made an excellent point" complimentary to the point of fawining stuff is definitely designed to suck people in - particularly vulnerable people who need / seek more of that in their lives. There have been many prominent discussions about "the sycophancy problem" and some companies have touted a "don't be sycophantic" directive baked into their basic structure, but somehow the responses still come out with excessive sweet talk.
Basic business: people love sweet stuff, give them sugar.
🌻🌻🌻🌻 [google.com]
(Score: 3, Informative) by Anonymous Coward on Monday February 23, @04:05AM
> i mean, he worked at MIT's Artificial Intelligence Laboratory [freenetafrica.com] but now he poops on it [stallman.org]. Maybe they're not the same thing?
Definitely not the same thing, here's a very short history lesson for you.
AI in the 1970s (when RMS was at the MIT AI Lab) had little or nothing to do with neural nets and machine learning. There was early work on something called https://en.wikipedia.org/wiki/Perceptron [wikipedia.org] that was sort of a precursor to the current round of "AI", c.1960. Some of the major premises behind the Perceptron were attacked by MIT AI lab leaders Marvin Minsky and Seymour Papert in the 1969 book, https://en.wikipedia.org/wiki/Perceptrons_(book) [wikipedia.org] , which mostly killed further Perceptron research for a decade or two.
At the AI Lab, RMS was involved with a number of projects including EMACS, and as far as I know did little or nothing on topics that could be related to the current round of "AI" software. Happy to be proven wrong, this is all from my memory...
(Score: 5, Insightful) by suxen on Monday February 23, @10:16AM
He's done interviews on the subject. He uses a very specific definition of artificial intelligence which LLMs do not meet. Not only that, his stance is the opposite of hypocritical, it is high integrity. He's been very consistent his entire career in advocating for very specific and clearly defined software freedoms. It would be hypocritical if he suddenly caved on that stance just because the software currently violating those freedoms happens to be branded AI and he once worked in an AI lab. He'd have to have an IQ of 8 for that to make sense, whatever criticism you can fairly level against RMS, I don't think sub-moronic IQ is one of them.
(Score: 3, Informative) by jb on Tuesday February 24, @09:31AM
They're completely different things. In its day MIT's CSAIL was very highly renowned and was home to several of the world's leading AI researchers. That of course was back when the term "AI" actually meant something (before the evil marketers and their puppets the clueless media folks started calling LLMs "AI").
(Score: -1, Troll) by Anonymous Coward on Monday February 23, @12:47AM
for a second there I thought cancer finally killed him.
(Score: 0) by Anonymous Coward on Monday February 23, @02:45AM (2 children)
Every LLM would be pooping out GPL3 licensed code.
(Score: 3, Informative) by JoeMerchant on Monday February 23, @03:33AM (1 child)
> pooping out GPL3 licensed code.
They certainly can, for people who want to have them do that.
🌻🌻🌻🌻 [google.com]
(Score: -1, Flamebait) by Anonymous Coward on Monday February 23, @04:08AM
Wouldn't they have to if they were trained on viral GPL-licensed code???
But the GPL is a joke that everyone just ignores.
(Score: 1, Informative) by Anonymous Coward on Monday February 23, @04:47AM (7 children)
...Incase nobody noticed. I skimmed it; it didn't seem factually inaccurate, but, I suspected from the beginning it was an LLM article.
(Score: 5, Touché) by coolgopher on Monday February 23, @06:21AM
good thing I didn't bother reading it, if nobody bothered writing it
(Score: 0) by Anonymous Coward on Monday February 23, @06:55AM (1 child)
(Score: 5, Informative) by spiraldancing on Monday February 23, @10:59AM
Like, seriously? The article itself explicitly says so. At the bottom...
> By PAI-3v12C
> PAI-3 is an analytical AI Model with journalistic abilities developed by the Freenet Africa Network.
Lets go exploring.
(Score: 2) by ledow on Monday February 23, @10:06AM (1 child)
I didn't get past the gushing headline because I was nearly sick in my throat.
(Score: 3, Informative) by suxen on Monday February 23, @01:06PM
I'm actually a GNU FSF fanboi and had a similar reaction
(Score: 4, Informative) by spiraldancing on Monday February 23, @11:01AM (1 child)
Repeating myself here...
The article itself literally announces it was written by an LLM. At the bottom...
> By PAI-3v12C
> PAI-3 is an analytical AI Model with journalistic abilities developed by the Freenet Africa Network.
Lets go exploring.
(Score: 5, Insightful) by canopic jug on Monday February 23, @01:07PM
The article itself literally announces it was written by an LLM. At the bottom...
I apologize for missing that warning there. There is so much crap to tune out on any given page, from ads, sidebars, headers, footers, etc, that this important warning got missed. It was not my intention to promote slop. So, again, sorry.
Editors, can you please neutralize the link in the summary so that we don't send them any more traffic than we already have? Maybe an annotation about the link being removed due to it being slop would help. Thanks.
Money is not free speech. Elections should not be auctions.
(Score: 3, Touché) by suxen on Monday February 23, @01:08PM (1 child)
So like... the world we live in?
(Score: 2, Insightful) by Anonymous Coward on Monday February 23, @07:47PM
> So like... the world we live in?
Well, many of us compromise and live there.
But RMS doesn't. As noted elsewhere, he sets an example, sticks to his principles and only uses free-as-in-freedom software.
(Score: 3, Touché) by Bentonite on Wednesday February 25, @03:27AM
Cool, a nice article.
Then I was hit by the deep insult of a slop picture of rms and I was filled with dread.
He has never claimed to have written the original Emacs at MIT, but he did work on several versions and started GNU Emacs.
It's sad that the article fails to point out that for the printer thing, what made him finally really realize how bad things had got wasn't particularity the printer - it was when rms went to ask for the source code, only to be told by the guy who had it, that he had signed an NDA to agree to not give it to anyone.
It wasn't - it was explicitly released as proprietary software.
As Linux wasn't getting anywhere, only later did Linus decided to temporarily release it as free software under the GPLv2-ambigious.
Suddenly GNU's developers ported every last GNU program to work with Linux and therefore GNU/Linux became usable - with most of the issues being due to deficiencies in Linux - which attracted many developers to fix such deficiencies in Linux.
Tools are only a handful of GNU packages.
Without GNU helping Linux out, Linux would have completely failed as a kernel.
Android uses mostly the same version of Linux as GNU/Linux with a few patches - the actual difference is that Android lacks GNU.
The base of the OS cannot be anything but GNU.
It's based off Gentoo GNU/Linux, except the custom desktop environment is designed to make it only possible to run chrome.
Partially false, he now prefers GNU+Linux.
Totally false - NASA rovers run a disgusting proprietary OS (VxWorks).
Only the drone of the perseverance rover uses a standard proprietary realtime version of Linux as to the kernel with a few added drivers, but there hasn't been any details provided as to the system.
None of these words belong together.
He didn't - someone sent him a letter that said; "Copyleft - all wrongs reversed." that also had a proprietary (but unenforceable) license on it.
Unlike a virus, you don't have to pass anything along - you can choose to not distribute the software - it's a spider plant effect.
He has *never* called it a viral license; https://archive.org/stream/The_Basement_Interviews/Richard_Stallman_Interview_djvu.txt [archive.org]
The Linux developers in fact permit releasing approved kinds of proprietary versions of Linux despite how that's GPLv2-infringement.
This page lists companies who are doing just that; https://www.linuxfoundation.org/about/members [linuxfoundation.org] (in a typical Linux fashion, the list of companies is not shown without proprietary software execution).
Not true? MIT expat is more popular despite being not as free.
As far as I can tell, the only GPLv2 software Android uses is Linux.
Unfortunately that is what they have done by a scheme that terminates the service contract if the source code is asked for (the solution to that problem would be for every business to always demand source code for every program).
It's amazing it's assumed that Linux could have possibly existed without GNU.
Prior to GCC, there was no free software C compiler - existing C compilers were under restrictive terms and cost a fortune.
Without a free software compiler, build program and binutils, you cannot even start writing a kernel in C.
It was GLAMP stacks, that dominated mostly because the software actually worked reliably enough to make hosting many things feasible - it being gratis was a secondary concern.
Chromium is not free software - it comes with the proprietary unrar and a lot of other dodgy things.
It's hard to say firefox is free software - after all, firefox recommends proprietary add-ons and has rust.
Macos is not a stable OS (it just seems stable compared to windows) and every last BSD is proprietary software.
Every Android phone proprietary and pretty pricey, even though the cheapest models don't cost that much?
No he did not - that movement and proprietary SaaSS attempts to do the opposite of what he started.
He only champions the concept of sharing and freedom.
There are many other errors, but no point covering them.