Russian intelligence had agents in Kiev who had assured them that they could stage a coup and take over the country. But the agents mostly turned out to be grifters and double-agents so that plan failed and a lot of high-level Russian intelligence folks got sacked.

They then progressed to forcing Ukraine to the negotiating table, which took a few weeks. Ukraine would have had no choice but to capitulate until Boris Johnson told them NATO would use them as mercenaries against Russia and provide them with weapons and intel.

That then led to the war between NATO and Russia that's been going on for the last four years and is approaching its end. Either NATO sends in troops soon or Ukraine is done because Russia will continue destroying their infrastructure all winter long.

It's a war. Things get blown up.

As I understand it, the Russian military doesn't use civilian data centers. The Ukrainian military does. So Ukraine will just be hit harder in response.

Remember how Russia was defeated only a few weeks ago because Ukraine was blowing up the Russian version of Amazon? How'd that turn out?

The general plan appears to be to tell the AI to rewrite it from scratch after a while.

We have an AI review all code changes for security holes because it's been programmed to be very good at finding ways to break security that make perfect sense but I would never have thought of. But it's purely advisory, I'd never let it have the final say on whether a change is safe.

That's why you have tests to ensure they don't.

Otherwise you pass it a hostname instead of an IP address, the low-level code happily accepts it in code paths that have never been tested and before you know it something explodes.

That is the way we're heading. So don't build your life around relying on software working because nothing is going to be reliable in future.

One of our problems is that we can't make major changes to the software because it invalidates over a decade of use in the field and massive amounts of testing required to ensure it doesn't kill people. Sooner or later that will be replaced with AI testing, so the code will be developed, tested and installed without a human ever checking whether the AI did it right or just hallucinated a lot of crap and then pretended the tests passed.

Turns out it takes longer to review code that no-one understands.

The solution, obviously, is to have the AI review the code too so humans don't need to understand it.

There is zero benefit to responding to a journo who plans to twist your words to fit their Narrative.

It's not the tariffs, it's the people generating funny cat videos all day.

Yes. In 2023 I was paying $500 for 8TB SSDs. Last week I paid $900 for a 4TB SSD and it doesn't even have DRAM.

So we have a double hit because lots of PC purchases were brought forward last year by Windows 11 refusing to run on perfectly good computers and the price of many components has doubled since then. I wouldn't have replaced my laptop if the old one wasn't literally falling apart.

Overall I'm guessing it's been a net negative for humanity, but for me it's equivalent to having a competent junior developer who works 24 hours a day. There are a lot of things we've wanted to fix in our software which have been put on the 'if we ever have time' list and now I can just leave it implementing them when I'm doing other things.

We're looking for a new car. Two we're considering are the Toyota RAV4 and the Subaru Forester.

You can't buy a non-hybrid RAV4 any more and apparently pretty much all 2027 Foresters will be hybrid. The RAV4 is one of the most popular new vehicles in the US so everyone buying one will be getting a hybrid whether they want it or not (and paying a few extra thousand dollars whether they want it or not).

So to a large extent this is just showing that manufacturers aren't offering the option of an ICE car any more. Hybrids probably have higher profit margins.

Sure, but you can use it to create a plan for software development and then farm that work out to Chinese AI on local servers. So they have no real compelling case for being worth trillions of dollars unless they can ban local AI or make hardware so expensive that no-one can afford to run those local Chinese models.

Also Big AI models are increasingly wrapped in 'guardrails' which burn up lots of tokens deciding they're allowed to do what you're asking them to do or to refuse to do what you ask them to do. The cheap Chinese models either don't give a crap or can easily be modded to not give a crap.

If a journalist asks you to comment on a denigratory story they're doing so because they want a comment from you that they can spin into support for their Narrative.

This is why smart people don't respond. It can only make things worse.

My guess is that SpaceX is the only one who will get away with it. They hit the peak of AI mania and now people are getting tired of it.

Not to mention that open-weight models are competitive with Big AI in many areas; I can do things at home with Chinese AI that I would have had to pay Big AI for only a few months ago.

I just don't see it being hot and shiny enough for more trillion-dollar IPOs at this point.