One of the resident trolls at Jo Nova's responded to one of my comments with an interesting link: GraphCast, AI-based weather forecasting. Looks like they've done the sort of thing I've been suggesting as a worthwhile use for AI: training on historical weather data and using that to predict upcoming weather. Done right, I think that should well outperform anything equation-based.

Might be our government meteorologists won't rush to it. The last thing they want is for a large and cheap advance in forecasting.

Was unsatisfied with Jo's response on that comment's parent thread. I'm not usually a stickler for semantics, but it really is time that climate got nailed down to a rigorous definition.

tomo,
We're often singing the praises of engineering here, and it's well deserved. Where would we be without it? But engineers are a different story. They definitely run from gooduns to baduns. In the 1980s, Sydney's Water Board was under the influence of the latter (IMO). They just loved big engineering projects, the bigger the budget, the better they were pleased. Perhaps, alongside the various strands of engineer: mechanical, chemical, electrical, civil, ...., we should include the empire building engineer. Might be that the Western Link is from that discipline.

That HVDC stuff does look impressive. The schematics don't look so fancy — just a handful or two of components — but I guess they're pretty chunky, and the cooling system has to visit all parts when gigawatts are moving (hopefully not dissipating too many of them).

.,
I suppose Musk favours Eric Raymond's bazaar over the cathedral. It is a lot noisier, but better than having high priests running the show.

To me, whole Twitter thing is a strange country. An e-pub, where you meet and exchange views with a few mates is easy enough to picture, but with famous people it's more of an e-stadium. What sort of conversation can be held between many thousands of people, each knowing that his view is the most important, and all blethering away at once?

On TIKhistory: I looked through the Stalingrad series, but didn't see where TIK was broken. Looks like it comes to a conclusion too, so maybe he unbroke himself and completed the series.

Rather than start on Stalingrad, I took pot-luck on the clickbaitiest title I spotted: Churchill Was an Idiot. He did a decent job with it, and was magnanimous enough in the comments section to allow that the world was probably better off for Churchill having led (idiocy and all), but I think he's missed the more important point: Churchill's idiocy (or genius) didn't matter. The great men of history are caricatures. The world is far more complicated than the impressionistic painting of history.

The funny thing is that, in the handful of TIKhistory videos I've watched, the primary recurring theme is how planning by the military leader is crushed by the complexities of the real world. I think TIK knew all along that Churchill's capabilities didn't matter very much; it really was just clickbait.

I don't mind having been sucked in. It was enjoyable.

I see X community notes are being used by lefty eco twits to attack critics of EVs...

and around it goes

DaveS,
I haven't yet dipped into the Stalingrad series because it looked pretty daunting. Might jump right in at Tik's crisis since that'll probably give good enough coverage of what has broken him. Thanks for the pointer.

Yesterday Jo Nova told of Swedes and Norwegians not too happy about the European grid. Apparently not that keen on the umbilical to the UK either. IMO this whole "open market" idea is idiocy. Each country should aim for energy self-sufficiency, paying a premium to use interconnectors for rebalancing. Having massive flows across interconnectors at market prices is bound to make the whole thing fragile.

And today, Jo's on American climate scientists feeling threatened by Trump's return.

The underlying Guardian article (well up to Guardian standards) refers to a Pew Poll which found that public trust in scientists is down 10% on pre-pandemic levels.

I was wondering how that percentage was measured. Clicking on the Pew Poll link, and digging further down to the (rather complicated) questionnaire itself, it turns out the question is:

Scientists. How much confidence do you have in them to act in the best interests of the public?

FWIW, this is on the heels of asking the same question about elected officials, journalists, the military, religious leaders and business leaders.

The 10% drop is in percentage of respondents with positive responses. The offered responses seem calculated to herd people to one place. Here they are:

1 A great deal of confidence
2 A fair amount of confidence
3 Not too much confidence
4 No confidence at all

#2 and #3 are magnets for everyone and, given a trade-off between not being "fair" and having "too much" confidence, #2 is going to attract a lot of people. If I were wording such a survey (and didn't want to skew it), I'd have symmetrical questions using trust and distrust:

1 strongly trust
2 somewhat trust
3 somewhat distrust
4 strongly distrust

But I suppose I'm not a pollster.

TiKHistory is a good channel. He was more into military history but his highly detailed narrative of Stalingrad over countless episodes broke him (somewhere in his playlist you'll find a 'I'm a broken man' video) after which he had a bit of a reset and now tends to post more on historical-political-economic themes. But hopefully he'll complete the epic Stalingrad series one day.

Dave Jones did a 3-hour interview with Lee Felsenstein on the Amp Hour. Sound quality was so-so, but the guest was interesting having been in computers since the late 50s. (Perhaps his best known achievement was creating the Osborne portable computer in the '80s)

Not included in the show notes was his take on AI (would have been in the third hour, and it looks like they lost interest in making notes at two hours). Rather unoriginally, he refers to AI as artificial stupidity, but he was quite confident it would make a satisfactory replacement for a large number of jobs. Specifically "bullshit jobs" like the people who pad out documentation with boilerplate stuff. AI will do a worse job of it, but it doesn't matter and it'll be cheaper.

Listened to a few more TikHistory videos. Good listening, and his economics "lens" gives a different take on supposedly settled stories.

Did some rationalising on the podcasts I'm suscribed to. Kicked out an ABC show that used to be 1/3 worth listening to, but lately might be 1/12. Replaced it with Josh Szeps's Uncomfortable Conversations. He's a pretty firm leftie and seems a little more self-regarding than is healthy, but he's intelligent, and interviews interesting people. Confident it'll be a step up from the ABC stuff I ejected.