You can run something like SmolLM-135M on an Esp32. It's super slow and it's a pretty crap model, but it probably outperforms GPT2 at some tasks. But some devices (like my tv) have an order of magnitude more compute power than that. Could an ai build a worm that could seek out hosts for it to run on? I still think the answer is no, but it's not as much of a "no" as it was a couple of years ago.

One of the cool things with open source is that we get to see that play out in a real world environment. My sense is that the open source projects who figure out how to use AI effectively to maintain a consistent vision and direction will probably beat out everything else. Maybe the purist human ones will land in the middle and the ones that allow a bunch of slop in will be below that - but maybe the all human ones will stay on top, idk.

I kind of feel like the code contribution path is sort of the wrong direction. What we've been doing at work is having non-technical people use claude to write what are effectively gold-standard change requests - they have detailed steps for whats wrong, risks, criteria for acceptance etc - then i feed those into my claude who converts them into code with all of our required standards and conventions and I can answer any missing pieces as they make sense for the broader project.

I can't help but feeling the human purists will end up looking more like Richard Stallman with his perfect ivory tower. Ideologically pure but moving so slowly by comparison that they are left behind.

Yeah it feels weird. It's like microsoft going after someone that uses Visual Studio to make their own IDE. As long as the Chinese customers are paying for access to the models than this seems like entirely fair game. I'm also rather confused as supposedly Kimi K3 was distilled from Fable/Mythos but they only had about 10 days of access to do that? If it's possible to learn the secret recipe to an AI model in that short of a time, it hardly seems like a secret.

You could even pitch it as a ground-based satellite You could make a box that has compute, batteries, starlink and a few acres of solar panels. Pack the whole thing onto a semi truck and you could plonk it down anywhere there's available land. There are loads of places in the US where you could do that, and I think most the anti-datacenter crowd would be placated if the facility didn't need water and generated its own clean energy.

That seems like the obvious compromise. You can drop POTS when you have FTTH provisioned to an address

GPT-OSS-20B gives me 34 tokens/sec on a 4 yr old gaming laptop with a 3080 RTX - smaller models come in faster still.

But that's an externalized cost - the business isn't paying it and the homeowner isn't directly paying it either. So it's really a win-win.

It would be interesting if you could bring it into your basement in the winter and heat your house for free - that does start to make a little more sense.

tbf that's kind of the definition of a good salesman!

To me the hoops that smoothbrains will jump through to avoid IPv6 and stay on legacy IPv4, especially when hosting, is pathetic. NAT, port forwarding, tunnels, blah blah blah blah.

I have something like ~1.2 trillion times the number of routable addresses that the entire IPv4 space has. Not all are reachable, of course, just the services that need incoming access and they're each on their own isolated DMZ.

Forgot to add: if you enable IPv6, be sure to use the browser extension IPvFoo. It shows a small 4 or 6 in your URI bar to let you know what protocol you're connecting to. Once enabled, you'll see just how much stuff out there actually is IPv6.

Started the move about 18 months ago when I decided to get off my lazy ass. My ISP gives out a /56 prefix, so that lets me run 256 /64 subnets/VLANs in the house, currently there are ~10 in use. Everything get a GUA through SLAAC and I use RAs (Router Advertisements) to give ULAs to everything. Any external facing services get their own VLAN and /64 for the system(s) as needed. Firewall blocks all incoming as they usually do by default and I punch a hole for the external-facing systems. They can't reach back into the network, they only answer the phone. All the systems update DNS dynamically if the prefix or full address ever change.

I have an SSH bastion set up. In all this time there has not been a single SSH attempt from the internet. On IPv4 it was constant background noice.
For those legacy IPv4-only systems on the internet, I set up NAT64. I have an IoT VLAN and IoT 2.4 GHz wireless network that are only IPv4 because a lot of IoT network stacks are junk.

I'm still farting around with it, but man oh man, there's no way I'd go back to IPv4. It was one of the best moves I've done in ages.

I suspect it forces them to consider that you could apply that same logic to the US if gas prices were to rise. The country has cheap ubiquitous gasoline and electricity, but one of them is a lot more prone to market shocks

But a lot of the time these companies do already run some systems.

My employer runs mostly it's own in-house coded system that's been built over 25 years. I'm effectively one of two developers maintaining it and for a long time that was patching critical bugs and adding a few key features each year. I had the question raised a year ago if we could added asset tracking capabilities - since we already had records for employees and a location based inventory system it seemed like a really natural fit but we didn't have the resources to develop it so operations picked some random saas project and i think it works fine.

However if someone brought me that question today, i could spend 30 minutes and outline a really solid project definition for claude and have it back the next morning. It's still a lot more challenging to get claude to work on the highly specialized tasks that are deeply tied to our data but for cloning a $10/user/month saas app, it really hits the ball out of the park.

They are going to have a hard time expanding into scotland https://www.urbandictionary.co...

Also this is pretty rational government behavior. You've got a country that produces copious green energy (and has the potential for much more) and where oil is expensive. It's exactly the right thing to do.