FUTO – An independent software lab and grant fund
futo.org
futo.org
They do sort of seem to worry about a lot of the right stuff.
As programmers and engineers we can have an idea and then through some effort just make that thing exist. It's really is tantamount to wizardry.
Sadly it feels like there are more obstacles to this every day, both in terms of literal hurdles like locked down hardware and app stores, but also the (erroneous) notions that you need to host your software in the cloud or use FAANG-scale solutions like kubernetes, or just the Jenga towers of half-baked libraries that make up the technology stacks many developers are stuck with means constant attrition from library churn any time you get anything to work.
And also this weird mentality that's drilled into a lot of inexperienced developers that they need to lower their ambitions, and to stick to building simple things.
I think that sticks with people even as their experience and capacity to produce software grows.
I think it's gotten to the point where many people with the skills to build amazing software just don't understand that this is something they're able and permitted to do.
You get to build software for the sake of software existing and providing a benefit to you, to yourself. If you need something, you get to build it. Software can be used to change your world, the world, it can be an extension of your will.
Programmers obviously get that they can write software, but not that software can be very impactful, it can change lives, even reshape societies.
You don't need to limit yourself to todo apps, text editors and static site generators. You don't need a license or approval or permission to do build crazy ambitious stuff for the sake of crazy and ambitious stuff to exist. Not yet anyway.
Mozilla Foundation has large overhead on management (CEO getting more while people getting fired), and it has that structure of a non-profit within a for-profit of the same name/brand.
For organizations providing grants, I don't think Mozilla even does that? Unless you mean stuff which is brought under the umbrella of Mozilla. Even YC and Google provide grants, its just with a twist (not selfless with the kicker being private equity). For an organization which provides selfless grants, there's for example the Dutch NLnet Foundation [1]. Which also falls under a different jurisdiction which can be a pro or con.
FUTO seems diametrically opposite in that it's ostensibly run by a small number of tech people.
NLnet is cool too. They seem relatively similar to FUTO in terms of goals, but they've been around for decades and it shows. Dunno how to describe them other than that they're very streamlined. Still also very low-nonsense and the people you interact with as a grantee are technologists and not bureaucrats.
His summarized description is "FUTO is an organization that seeks to give users back control of their technology".
Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.
Frank Herbert, DuneThere are humanoids that are assumed to be (at least nearly) as smart as some Minds, be it they number in the low tens, one of the reasons that Minds do keep biological humanoids around is because they're sometimes surprised by their (perhaps irrational) but interesting point of view, there's also a point of pride amongst the Minds as to who can be more emphatic towards the humanoids, so it's not as simple as you say, be it that wasn't my point.
Hope we can do better than that
I'm worried about hardware, though. My wife got a new Dell laptop about six months ago, and despite my 20 years of installing Linux distros on every piece of hardware I can, it still took me hours to jump through all the hoops to get Ubuntu on the machine.
Microsoft is locking stuff down, hard. I'd be willing to bet that when Windows 12 releases, it will require that the PC be a closed platform that does not allow users to install another OS. MS sees people just living with the closed by default mobile ecosystem and wants something similar for PCs.
Hence why free software i.e. Libreboot, free firmware etc. is imo as important as ever.
Libreboot et al are important, but meaningless in the grand scheme of things if you have to be a serious computer geek to be able to install them on your own hardware.
The battleground is hardware manufacturers yielding to Microsoft's requests. Lock down the bootloader, lock down the bios, lock down the firmware, before you know it, it's as hard to install Ubuntu on a Dell as it is to install LineageOS on a carrier locked Samsung Galaxy.
We need manufacturers that say no to Microsoft and keep making unlocked PC hardware.
Plus, it can hold multiple distribution images (including Windows) and let you choose at runtime which one you want to install. You just paste the ISO file onto the filesystem (which is NTFS or exFAT or something, so there's no file size limit). And since that's an ordinary filesystem, you can also use the spare space to move any files between computers as if it was ordinary storage drive.
I think my point still stands, though. Users shouldn't have to work around secure boot. Secure boot is making Linux adoption more difficult for casual users for no good reason.
If anyone has a use case for secure boot outside of an enterprise setting I'd love to hear it.
Agreed. I think the Steam Deck is also showing how important hardware with pre-installed Linux is.
GNU Radio is one of the most popular open source software for communication and the initial funding was provided by John Gilmore [1],[2].
[1] John Gilmore:
https://en.m.wikipedia.org/wiki/John_Gilmore_(activist)
[2] GNU Radio:
I saw your chat option but for privacy reasons, or paranoia if you prefer I tend not to give out my details to third parties, such as my email which you require to use chat.
If they are there and I didn't see them, that's on me. Thank you.
Some interesting companies: https://www.ycombinator.com/companies?nonprofit=true
Did I misunderstand his answer or is this something that changed?
I applied for a grant once but never heard back. My goals may be too ambitious for some though, so no hard feelings :^)
I believe the industry is heading down a dark path with even big companies like Microsoft and Opera hiding behind Chromium instead of continuing to develop their own engines. IMO we need more independent implementations of the web platform to ensure that it remains open and healthy.
Re: too ambitious, it's a commonly held belief that building browsers is impossible without a huge budget and hundreds of staff. We're doing it anyway, it just takes time. :)
He also encouraged me to start the right to repair non-profit and donated $1,000,000 to it 3 years ago.
FUTO is not a company.
(Just in the interest of disclosure, I'm a "Free" software partisan (I don't see the point of "Open" source.) I don't want to argue about it, I'm just being upfront about my own bias.)
It seems to me that we have three great UI paradigms: CLI, GUI, and now LLM. The first is typically seen by normals as inscrutable and esoteric, the domain of computer "wizards". The second is "computer" in the minds of most normals, this is how you get University courses on e.g. how to operate MS office software, etc. The third, speech, is of course learned automatically as children.
From now on the kind of programming we know (and love, some of us) is obsolete. No one wants to operate symbolic calculators via janky teletype machines. No one really wants to learn boring video games like Excel or Word. We want to tell the computer what to do and have it do that. Well, now we can.
But that means that we, the hackers and computer nerds, are now playing with model trains. (That's my metaphor: am I writing code people will actually use? If not, then I'm playing with my model trains. Nothing wrong with that! Fantastic hobby! But I'm not starting a shipping company with N gauge trains, yeah?)
It seems to me that now, in 2023, most needful software has already been written, just as most needful technology has already been invented, and the focus of an organization like FUTO should be on simple open hardware, an open library of algorithms and DBs, and a simple LLM UI, all on top of something like Glamorous Toolkit. Plug that into an open IoT, declare victory.
The rest of the job is marketing.
Bicycles, and their near-optimal designs, are still a $68B industry.
> Automating "correct" is still an unsolved problem.
That's our job.
Cf. "Augmenting Human Intellect: A Conceptual Framework" SRI Summary Report AFOSR-3223 By Douglas C. Engelbart October 1962 https://dougengelbart.org/pubs/augment-3906.html
And "An Introduction to Cybernetics" by W. Ross Ashby http://pespmc1.vub.ac.be/ASHBBOOK.html PDF: http://pespmc1.vub.ac.be/books/IntroCyb.pdf Specifically "Amplifying intelligence" which comes at the end so you kinda have to read the whole book to get it. It's worth reading anyway.
> Bicycles, and their near-optimal designs, are still a $68B industry.
What does that have to do with anything?
Natural language is packed with ambiguities and unspoken/unwritten context, and is thus a terrible programming language. People will end up spilling paragraphs of prose into text models, to do poorly and unpredictably, what a couple lines of bash would have done perfectly. For "approximate", natural language is fine, and possibly even optimal, but for precision, for "correct", it is unsuitable. It is why every profession develops its own domain-specific vocabulary and re-definition of phrases: they need more precision. Don't get me wrong. Those paragraphs will most certainly be spilled, but this is not the demise of programming. This is the next degeneration of programming! The Netflixification of programming; the union of hacks and technology, generating infinite low-quality content for insatiable consumers who have lost all faculties of discrimination.
Perhaps the "most needful" software has already been written. But how much of it is any good? Last I checked, everyone's bug tracker is still overflowing. Some are easy fixes. Others will be decade-long slogs. I wish we were as close to done as you think, because I think with these new models, we're actually going to start slipping further away.
> What does that have to do with anything?
If the market size serves an estimate of work the world wants performed, and something as "done" as a bicycle still pulls-in $68B/yr, while software still earns trillions, then by the money-metric we have quite a ways to go before we are anywhere near finished.
> thought that your epitaph was premature.
It's possible. I gave up prognostication when I didn't predict Twitter. I recently checked Craigslist for jobs in my line and it was empty...
> the symbolic calculator via teletype is still getting me to the point of "correct" faster than ChatGPT.
I prefer it myself. But we're not the mass audience, eh? My entire career is built on the fact that most people won't even learn Python or JS (let alone bash.)
When we connect these models to empirical feedback devices they'll be able to tell us correct things.
Schmidhuber says his goal is to "build an automatic scientist and retire."
> Natural language is packed with ambiguities and unspoken/unwritten context, and is thus a terrible programming language. People will end up spilling paragraphs of prose into text models, to do poorly and unpredictably, what a couple lines of bash would have done perfectly.
But you have to learn bash first. Now we can offload all that overhead to (fast powerful) external processors.
> For "approximate", natural language is fine, and possibly even optimal, but for precision, for "correct", it is unsuitable.
Nyet. Read the links I gave you. Our task is selection from the choices the machine generates. That's how you build an intelligence amplifier. The machines are capable of arbitrary precision.
> consumers who have lost all faculties of discrimination.
That's the only faculty the machine can't replace. That's the purpose of humanity: to answer the question "What are people for?"
I don't know what's going to happen, but I am pretty sure of the overall envelope of possibility. We might become the slobs in Wall-E, I don't know.
- - - -
> If the market size serves an estimate of work the world wants performed, and something as "done" as a bicycle still pulls-in $68B/yr, while software still earns trillions, then by the money-metric we have quite a ways to go before we are anywhere near finished.
I'm still not sure I'm getting what you mean? What's "finished" mean here? People buy bikes, yes, but I don't see the connection to my thesis that talking computers will rapidly make obsolete things we programmers take for granted now: Programming Languages, Bug Trackers, Bugs, Frameworks, Libraries, Apps, etc. Just as the smartphone subsumes many previously-separate devices, these "Talking UI" systems will subsume most of what we think of as programming.
I'm already familiar with the Engelbart paper, but this is what people have already been doing with Google; with copy-paste from Stack Overflow. More often than not, all of the generated choices are either partially or wholly incorrect, and in need of either a cross-breeding or a euthanization. In this respect, LLMs are no different from the answer generators which preceded them. Only the priesthood of RTFM knows what to breed, what to gene-edit, and what to put down.
Excellent! I hope I didn't come off too patronizing?
It's hard to argue this point (that the machines will rapidly become empirical, aka scientists) because I don't know how it will happen, it's more than a gut feeling, but less than an equation or a working example. I'm certain it will, but less certain how swiftly.
In a sense, I'm trying to get out ahead of things to the extent that's even possible. FWIW, I would bet the inflection point comes within five years, the computers become automatic scientists, the options they generate will all be non-hallucinatory, and the only job left for us is "be human".