I'm playing around with it. I self-hosted the Tranquil PDS (personal data server, part of AT Proto), a knot (git repositories), and a spindle (CI runner) on my home server. It was all relatively easy to set up but even in the week that I was bringing it up I found bugs and then saw those bugs get fixed so they're earning the "alpha" badge on their website. I'm not using it for anything load bearing yet but I like the idea at a high-level, though it remains to be seen how well code hosting fits into the AT Proto model.
20GB isn't enough for a 13B parameter model? I thought the 29-31B models could run on a 24GB GTX x090 card?
I'm currently shopping for a local LLM setup and between something like the Framework Desktop with 64-128GB of shared RAM or just adding a 3090 or 4090 to my homelab so I'm very curious what hardware is working well for others.
It's crazy that the systems the best designed for decentralization like git, email, and the internet itself wound up being the most centralized with single points of failure.
I wonder if we could find a way to donate unused tokens or even local compute resources to open-source projects we support. Especially for security auditing where it could probably be somewhat more asynchronous and disconnected than the open-source developers' personal tool choices.
In the auto industry these are usually called a series hybrid and there have been a handful. The Chevy Volt (though it had the ability to directly connect the engine to the wheels at highway speeds), and the BMW i3 and i8, the Fisker Karma/Karma Revero. The new Ram Ramcharger truck and the second gen Ford Lightning will also be series hybrids.
It's a really good drivetrain that was unfortunately made untenable for a long time by a combination of regulation and market forces.
That's a complicated question that unfortunately has quite a bit of "well it depends" in the answer. I worked in the auto industry for a long time - both doing engine development and EVs - so my opinions here are well-informed but not world expert.
From a pure energy efficiency perspective you can't beat economies of scale. A stationary power plant (even ones that are just big gasoline engines) run at a constant load and RPM so they can be optimized for pure efficiency, they rarely have to start, warm up, and shut down, and they can use larger and more expensive exhaust aftertreatment systems. Most energy conversions grow more efficient with scale and this is no different. The locomotive powertrain works for a handful of reasons but one of them is you can build much more efficient engines that are optimized for a single constant speed and load. But most of the advancements in internal combustion engines over the last 20-30 years don't increase peak efficiency but increase the conditions in which they're efficient. Variable valve timing and lift are probably the most underrated and overpowered technologies that have transformed engines from having one narrow regime of high efficiency to running well over a huge range of the map. But turbocharging, variable intake geometries, 7+ speed transmissions, and mild hybrid systems like belt-starter-generators get honorable mentions here. However we're not talking about anything close to EV-levels of efficiency. I think the cutting edge research engines are running in the mid to high 40s for thermal efficiency (percentage of fuel energy captured as useful work), most passenger car engines probably peak in the mid 30s.
So while there is some efficiency to be gained by a more locomotive-style system it's not as much as you would hope. In the industry that's called a series hybrid system, vs a parallel hybrid system where either ICE or EV power can go to the wheels. The benefits of a series system are more emissions and product features. You can get the full torque and power of an EV, you can start and stop the IC engine in a more emissions optimized way, and and you can filter load spikes to use a small engine that meets average not peak load.
From a more pragmatic perspective, with the energy density of gasoline and other liquid fuels it's probably best to use it in applications for which you just can't use full electrification. Planes are currently the best example of this. It's also worth noting that passenger cars benefit massively from strong hybridization because of the uneven load cycles so that's a technology where you can deploy a gasoline engine but then claw back a lot of the efficiency losses with hybrids. That's not always true, for example boats don't really have a regen cycle so hybridization just doesn't get much.
When one gets in the weeds on EVs or ICE cars two things become shockingly clear: internal combustion is hilariously inefficient YET gasoline is hilariously energy dense. Most people's intuition is wrong on both of these points but then they cancel each other out.
Edit: another important point is that the "cost" to acquire gasoline is only the very end of the process. The energy has already been gathered, stored, and most of the processing is complete. Our cost (in money and energy) to "make" gasoline is really just gathering it. This is why the comparison to renewables is often a hard sell, it's just apples to oranges. Gasoline started on third base, renewables are batting from the plate. Some of the internal combustion enthusiasts are holding up e-fuels or synthetic fuels as the solution but then we have to pay for the entire energy gathering and processing pipeline and still be using a conversion method that's not at all efficient. It's the worst of both worlds.
Hannah Ritchie is a quite well reputed writer and data scientist squarely in the climate field. She's written two books on climate and I found the one I read (Not The End of the World) was quite good and data-driven.
I had the exact same thought while reading the above comment, as helpful and generous as it is. Google's entire business model is to help people find things on the internet. They're an insanely well resourced company with all kinds of smart programmers. They have a moral and financial incentive to direct people to canonical sources of information. And STILL it's on this open-source dev to do all the steps outlined just to get the situation corrected?
I was in engineering school back in ~2012 when Google Glass came out. One of my classmates got hold of a pair when they were still quite uncommon and wore them to an extracurricular club meeting. Within minutes someone made a comment about him wearing the "creeper" glasses and asked if he was filming. He never wore them to the club again.
I just don't see a world where that doesn't happen with Meta glasses.
My non-technical mother recently texted the family group chat to try to get us to use Signal. The winds are shifting towards privacy in a broader sense than ever before. This type of counter argument ("that doesn't sell [product") is usually a bad argument when the market doesn't offer anything that actually sells on privacy. It becomes a self-fulfilling prophecy.
The iPad and MacBook teams have been in market competition for nearly a decade now while clearly Apple corporate strategy has been trying to nerf each line to prevent them from actually competing. It's an artificial tension that gets more pronounced as the devices get more capable in other ways and the induced limitations are more glaringly obvious.
It could absolutely be false nostalgia since I was a Mac fan from about 2005 to 2009 with an MBP that had the breathing feature but this really felt like a high point for computer hardware. It felt slick and performant but still serviceable.
Ironically, after a series of uninspiring Windows machines the next laptop that made me feel any level of enthusiasm like that Mac is my current Framework 13 running COSMIC on NixOS. Quite an about face!
I hope that the EU becomes a real innovation center of decentralized tech initiatives. There are all these tech movements like local-first apps, atproto/activitypub, and self-hosting that could be absolutely supercharged by both the user and developer base of Europe flat out rejecting big tech cloud platforming.
I love this. I spent my holidays hearing non-technical family members complain about their ever deteriorating Windows experiences, issues that make me righteously angry at Microsoft.
IMO the next important unblocker for Linux adoption is the Adobe suite. In a post-mobile world one can use a tablet or phone for almost any media consumption. But production is still in the realm of the desktop UX and photo/video/creative work is the most common form of output. An Adobe CC Linux option would enable that set of "power users". And regardless of their actual percentage of desktop users, just about ever YouTuber or streamer talking about technology is by definition a content creator so opening Linux up to them would have a big effect on adoption.
And yes I've tried most of the Linux alternatives, like GIMP, Inkscape, DaVinci, RawTherapee, etc. They're mostly /fine/ but it's one of the weaker software categories in FOSS-alternatives IMO. It also adds an unnecessary learning curve. Gamers would laugh if they were told that Linux gaming was great, they just have to learn and play an entirely different set of games.
Like any opposition party, the anti big tech crowd is actually a loose coalition of different goals and interests. I've noticed that as these platforms get through the earlier stages of "will it even work" the differences in values are becoming more pronounced and controversial. The primary two groups seem to be those who value federation and see centralized control and algorithms as the threat and those who value encryption and see surveillance as the threat. Obviously these two things aren't mutually exclusive and we all want to see new platforms that can solve for both. But there's a quite distinct difference in the primary priority and consequent technical decisions.
I hope maybe if we can be aware that this is a broad set of technologies being driven by a broad set of goals then we can be a bit more gracious when a project isn't perfectly aligning with our personal values and find the common ground and values.
I'd go a few layers even broader than this article and say that the modern tech industry has an abysmal track record when building tools for non-software technical fields. Tech builds either their own software-oriented workflows or the most dumbed down consumer-oriented workflow they can. Law is an excellent example of a field with a very high degree of fidelity, philosophy, and process yet it can only ever have partial crossover with software development methodologies. Tech often treats someone like a lawyer as either a substandard developer or an advanced consumer without making a real effort to understand the context and needs of highly complicated yet non-software professions.
https://tangled.org/tranquil.farm/tranquil-pds
https://docs.tangled.org/knot-self-hosting-guide
https://docs.tangled.org/spindles#self-hosting-guide
edit: TIL HN doesn't support markdown, why does this surprise me every time.