shubb

I think it’s because the new post training approach that has enabled a huge improvement on long running tasks and tool / computer use and the underlying latent reasoning applies to verifiable problems that can be solved en mass and repeatedly in isolation.

Generating text humans like, empathising and cold reading the user’s intent, observing something interesting if second hand ally the human experience - none of these can be verified without a human. So they rely on larger pre/train and a larger human curated post train corpus. These are not scalable in the same way.

If the ROI was there, they could do more. Startups like sesame and previously pi showed that a more personable agent than today’s frontier models can be achieved with a sane budget. That’s another unverifiable objective.

But I do think this technology that explains the jagged leaps forward this year does not apply to this problem.

On the face of it, product management and service management are the big duty donors.

Project managers and real executives manage a budget in dollars, not man days or tokens. Historically businesses are pretty reluctant to devolve that. I don’t think a big corp vibe coder is going to be able to pull out a company credit card and buy a marketing campaign any time soon.

So you have those other 2, managing product market fit and availability.

One person startup you do everything I guess

A lot of folks want to run tests off a GitHub action I.e. on a server somewhere. Ideally you want your test stack to fit in a docker image. So this does suck for developers in the respect and you could imagine apple releasing a special docker image that just ran safari if they wanted to really make it easy to develop for thier platform.

However, I imagine someone will fill a server rack with cheap old macs and offer and safari mcp as a service…

There are pre-ipo tokens, which are a bet on the price at ipo and proved accurate fit SpaceX (lots of caveats here). Anyway we can see it as a prediction market where the price may have information because insiders can trade without SEC oversight or something?

OpenAI and anthropic tokens are down 30% in the start of the month but flat on the week, possibly meaning that market front ran this sell off, possibly not.

Are we post peak?

Everyone is worried about AI for good reason but if he's the promise is true then we see significant productivity improvements.

The pie might be shared out worse than even today but there would be more pie.

We are in a period of deglobalisation, but also a period of reorganisation. Today's supply chain is less efficient than 2021s. We are materially poorer as a result. But after the dust settles, it will be a lot more efficient than it is right now. Even with no ai.

Potentially we are in a dip. Stuff for worse, but it can get better.

Net worth is a funny metric.

Joe has a 300k house with 100k equity and 200k mortgage. He has 100k in stocks in a 401k. Net worth negative 100k.

Pete has $300 in his cheques account, and isn't eligible for loans or mortgage. Net worth positive $300

Obviously Joe is richer than Pete though.

Take a Damm Small Linux. Package manager (apt) optional.

I do think a lot of what you are rejecting (automatic updates, centralised package signing, permissions) are solutions to security problems that you might actually have.

Something to be said for requiring the human to do something related like press go every time the vehicle starts from a complete stop, or annotate "threats" detected in the 3d view.

This is a safety problem that needs solving. Just detecting when they sleep isn't a good solution.

I think citymapper tried to execu this as a pivot. They had an idea to do it in London and other countries and did trial it for a while. Not sure why it (presumably) failed.

I'd note that startup money of the is much harder to get in London, so a US startup might be able to force the idea from experiment to profitability.

We live in a very complex system, beyond any one persons comprehension. Some people think devolved decision making allocating resources to things like, advertising better, is the most efficient way of allocating resources. The invisible hand. How much is bullshit and how much is just beyond your awareness? If you were king and allocating so the work, would it be better? For who? I'm doubtful about bullshit jobs.

Seems like ofcom picked thier targets to create precedent fur using ISP blocks against non compliant companies.

Case law is important in the UK because laws are vaguely written and thier specifics are established in court.

By going after entities that can't comply and don't have a big legal budget like 4chan, they can go through the motions and establish that ISP level blocks against non complying companies are okay.

They can then hit progressively more difficult targets until they get to X and tiktok.

I think it needs a water gun. If the diaper was a spray on layered rubber, like a sponge then an impermeable layer, and then you sprayed a solvent to clear the old diaper and poop and then spray on a new one. You'd just need to slot them into styrups briefly or some socks on strings to move the legs into a good position.

But can this be done with baby skin and lung safe chemicals at a reasonable temperature?

Point being humanoid designs for robots that manipulate objects designed for humans are an artificially hard problem we have decided to fail at solving.

Sorry, typo, I meant to ask 'what'. Agree there are tropes in llm outputs, specific and strangely not specific to a vendor. Very useful for recognising generated content. I was hoping you were aware of a good article to find more in.

Sorry, but wear is the recursive protocol trope? Also do you know a good list of these tropes for llm spotting?