> I'll be comparing this to the Ford Maverick.

This truck might just steal the thunder from an EV Maverick, and Ford can't release that soon enough.

This is even more damning because it means the maintainers want their MIT-licensed projects to be used by for-profit companies, but bellyache when certain big-tech companies fulfill the maintainer's vision.

> What's good about being "permissive"?

They want widespread usage of their project, but always decry not like that when Amazon or Microsoft is responsible for the usage.

> Whereas on local LLM, I watch it painstakingly prints preambles that I don't care about, and get what I actually need after 20 seconds.

You may need to "right-size" the models you use to match your hardware, model, and TPS expectations, which may involve using a smaller version of the model with faster TPS, upgrading your jardware, or paying for hosted models.

Alternatively, if you can use agentic workflows or tools like Aider, you don't have to watch the model work slowly with large modles locally. Instead you queue work for it, go to sleep, or eat, or do other work, and then much later look over the Pull Requests whenever it completes them.

I have a 4070 super for gaming, and used it to play with LLM a few times. It is by no means a bad card, but I realize that unless I want to get 4090 or new Macs that I don't have any other use for, I can only use it to run smaller models. However, most smaller models aren't satisfactory and are still slower than hosted LLMs. I haven't found a model that I am happy with for my hardware.

Regarding agentic workflows -- sounds nice but I am too scared to try it out, based on my experience with standard LLMs like GPT or Claude for writing code. Small snippets or filling in missing unit tests, fine, anything more complicated? Has been a disaster for me.

As I understand it, these models are limited on GPU memory far more than GPU compute. You’d be better off with dual 4070s than with a single 4090 unless the 4090 has more RAM than the other two combined.

I have never found any agent able to put together sensible pull requests without constant hand holding. I shudder to think of what those repositories must look like.

Not if you have ever signed up for a service that uses Plaid. Then it's your bank and Plaid.

I've been surprised at how many people will happily hand over their detailed financial data to Plaid. I can understand how many people might not know how sophisticated malicious actors can be with it, but everyone has heard of identity theft at this point and having someone taking out credit cards in your name is a huge pain.

> I figure it's only a matter of time before America invites "trusted commercial partners" to "analyze citizen data" for a "more efficient" auditing process

No need to partner with the IRS, the DOGE dudes have an external drives in a sock drawer at home, ready for their next Fintech startup. For reasons attributed to "world class algorithms" the Fintech will be preternaturally good at targeting financial services to individuals, and assessing credit-worthiness given nothing but a name and zip code...

I too would work for "free" to gain access to that valuable data. If I didn't have a half-decent moral compass.

> These small models only make sense with "thinking" enabled

This entirely depends on your use-cases.

...and the "Disaster" in "Disaster recovery" may have been localized and extensive (fire, flooding, major earthquake, brownouts due to a faulty transformer, building collapse, a solvent tanker driving through the wall into the server room, a massive sinkhole, etc)

Yes, the dreaded fiber vs. backhoe. But if your distributed file system is geographically redundant, you're not exposed to that, at least from an integrity POV. It sucks that 1/3 or 1/5 or whatever of your serving fleet just disappeared, but backup won't help with that.

> But if your distributed file system is geographically redundant

Redundancy and backups are not the same thing! There's some overlap, but treating them as interchangeable will occasionally result in terrible outcomes, like when a config change that results in all 5/5 datacenters fragmenting and failing to create a quorum, then finding out your services have circular dependencies when you are trying to bootstrap foundational services. Local backups would solve this, each DC would load last known good config, but rebuilding consensus necessary for redundancy requires coordination from now-unreachable hosts.

> It'll sell us supplements and personalized media and politicians and we'll feel enormously emotionally satisfied the whole time.

Which is why we'll need to acquire the drug gland technology before AGI - no mind can sell me anything if I can feel content on demand.

Alternatively they can sell you anything if they can make you feel content or euphoric on their command. Get your new drug gland today, free of charge, sponsored by Blackrock

> "kill it, we don't need this"

"We are paying MITRE how much? Bigballs and co will write a better ststem in 1 week and have it integrated with xAI. How hard could it be? Send out a first draft of an xAI contract to our DHS contact"