If going back to X is "the biggest arbitrage opportunity" of my career, then I have but truly messed up.

By that logic people *would* trust a no-filter politician who....

Oh, never mind.

There are problems with the open source supply chain, and yes, sustainability is a problem. And don't get me sorted on phone apps.

But imagine going back to the 90s and telling them that paying for software would be a deprecated business model within decades.

If they set up a party HQ at a Nando's in order to remain close to the people, I might take them more seriously than the OG at this rate.

"...the organisers were found guilty of eight charges of misleading the public, and jailed for 13 months each."

I have a recurring nightmare that some well-meaning policy maker will look at the whole-of-life carbon data and go: we need to get rid of all the pets. Cats, dogs, rabbits, outlaw 'em all.

And so set back progress on actually achievable changes by decades.

South Africa's intelligence apparatus is dreadful, but there is no excuse for the state not noticing that would stand up diplomatically. There were financial transactions, weird travel patterns, plenty of signals to pick up on.

Just as well the US and Russia are now allies too.

AI people are bad at talking about this because they're already sounding insane on other fronts and don't need a new one, and they don't want investors to hold back until efficiencies are production-ready.

They also, I reckon, believe they can soak up infinite electricity, so build baby build.

Meanwhile – and this is just today's proof point – Samsung is showing power-efficiency gains in the order of 90%, for a type of hardware that happens to lend itself to brain-like structures.

Not marginal gains, not slightly more efficient, but 9x%.

www.thestack.technology/samsung-says...

The vendor hype also helps obscure the reality that a lot of today's AI hardware is basically gaming stuff with a few extra bits duct-taped on. And the software side is still mostly of the same generation of literally the first thing that actually worked, just multiplied by a lot.

AI is turning into a great natural experiment for two questions:

* Does spending more on top leadership show an appreciable return?
* Is there a difference between, say, bricks-and-mortar companies and high-transformation, early-stages industries in that return?

By that definition – high return on equity, stable earnings growth, and low debt levels – the time to buy quality stocks is always.

There's an inevitable relationship between TAU and the IDF, and there will inevitably be some kind of war-AI scandal that traces back to the university, and possibly the Google funding.

But then again, war AI will run on Google Cloud anyway, so ¯\_(ツ)_/¯

Remember the good old days, when it was all about freaky dog-like quadrupeds that looked like they wanted to murder you and we didn't have to worry about Uncanny Valley Droids?

Man, I miss 2024.

(Also, dibs on UVDs, totally going to get that into print.)