tomo,
Rooftop hydro might have yielded a bit of energy in Sydney's recent rather wet years. Bit of a nuisance though, having to hold dinner up until the storm is at its peak.

With "floating wind", I wonder if anybody has considered the merits of mixing in tidal energy to make a doubly unreliable hybrid generator.

DaveS,
Like the old Mastercard ads, Hydrogen fuelled crematorium? £1,168,500. Bragging rights with other local councils? Priceless.

I gather Mr Trump has been found guilty on all counts. No great surprise, but it will be interesting to see the result of the result.

Latest econtalk is with Vinay Prasad on the COVID vaccines. No huge revelations to any COVID sceptics, but uncomfortable for the believers. Perhaps the most interesting observation was how the initial efficacy tests were redesigned on-the-fly so that they wouldn't come to a conclusion too quickly. This was so Biden would be announcing their success rather than Trump. That makes it pretty clear that, even in those early days, the experts didn't think the disease was so bad that it should have priority over politics.

May 30, 2024 at 12:01 AM | Robert Swan

If I read your link correctly then taxpayers have stumped up £1,168,500 for a four-week trial. Nice work for the various consultants involved. But if it helps alleviate Worthing's "climate crisis" then money well spent, I'm sure...

Superman saves the day (also from Jo Nova's).

tomo,
Maybe "towing to Norway" is just a cover story and the real plan is to tow them to an unobtrusive spot and scuttle the useless things. Can blame it on a North Sea storm.

A comment at Jo Nova's this morning revealed this brilliant initiative for a crematorium to use hydrogen instead of natural gas for the main event. The next step in sustainability might be to pop the bodies onto the conveyor belt at Drax. Still plenty of ash to give the bereaved at the other end of the process.

tomo,
I'm not exactly carried along by the recent surge in enthusiasm for AI, but outside of the hype, it might have some useful applications.

The guiding principle of the LLM is: given the "words" currently on the page, what is the most suitable "word" to come next, the "most suitable" evaluation being based on existing training.

This fits some tasks better than others. It can write a nice piece of creative writing (though it's up to *you* to decide this after reading the output), but it won't get you far solving a differential equation that's not already in its training material. I likewise don't fancy its chances driving cars or whatever. But I think it *could* be trained on daily/hourly/whatever weather data and how the weather actually played out afterwards. Instead of predicting the next chunk of text, it's the next chunk of weather, and the training material is solid empirical data giving the before and after.

In comparison to weather models, this more or less parallels the pragmatism of engineers who apply physics formulae when they can, and experimental results for problems too complicated for physics. Typically engineers distill such results down to look-up tables, but the AI methods might help for weather prediction due to the complexity of search.

Good point on Vennells: I didn't do it, and if you go on accusing me, I'll show everyone how you helped me do it.

On the el Nino video, I am an unbeliever in the significance of the whole ENSO apparatus. Certainly the patterns are there, but my view is that they're an emergent thing, like Jupiter's Great Red Spot: we know where it is and how it behaves in the main, but it's very unpredictable at the edges.

Other than timescale, do you think el Nino is really any different from the totally unremarkable way we look at the pattern of sunny day/wet day? We have "neutral" days too. It seems to be a natural consequence of the chaotic nature of weather: there are "attractors" at all timescales. I shake my head at ENSO being treated as some external thing and that's why our weather models got it wrong. Their weather models get it wrong because it's not *possible* to accurately model.

You had me worried with the Briggs tweet and the one about nobbling ad blockers. The Heath Ledger snippet in the first froze after a couple of frames, so I thought the second might apply to me. But no, at least for now, my YouTube remains ad-free. Long may it continue.

Still catching up with the John Anderson podcasts. I'm gradually working my way backwards through his archive, and simultaneously getting his new ones. Sound quality has improved over time, but his older interviews had a better standard of conversation. The latest one with Peter Hitchens was particularly poor. It was like Anderson wasn't hearing what Hitchens said, and just proceeded down his list of questions. Particularly bad with Hitchens, since he likes to give unexpected answers.