[–] 5 points 3 months ago* (last edited 3 months ago) (1 child)

I find this analysis is a useful starting point for discussion, although there are plenty of details one might adjust.
Personally I'm using (inter alia) claude to help me refine an interactive climate model (example here - although that's last year's version pre-ai-help ). So I care about these things.

As my own life also has an energy cost - even just sitting at a desk with computers and some heat light and food. I reckoned by my own crude calculations that my 'human' energy cost per hour was considerably higher than that of my AI assistant, which certainly helps me progress faster, so the net effect was less energy per 'task done', meanwhile we don't have infinite time to solve such problems. I'm only using claude within the limits of a pro subscription, and achieve that with tough claude_md instructions - not to go digging rabbit holes without consulting me. Sometimes it analyses and fixes autonomously and efficiently, but you have to keep alert - sometimes I interrupt and say no there's a simpler way, and draft better algorithms / structures myself. Also I use scala whose sophisticated (non-ai) tooling constrains mistakes and its mcp/lsp makes searching and refactoring across a large codebase much more efficient than claude's normal grok by subagents. Combine tools carefully, not brute force.

Evidently a big unclarified issue is the energy cost of training these things. But we don't need so much more training - for my purposes they are already good enough. The frequent new releases are about scary headlines to pump the IPOs. If this race could slow down, we could just learn to use what we've got more efficiently. In the general public discussion, I'd also appreciate clarification about how much of 'AI' energy-use is going into creating images and videos, rather than text and code, my hunch is it’s much worse for videos most of which are about trivial stuff. Also loads of datacenter energy is wasted transmitting talking-head videos around the world - that's really inefficient. So well designed code, part-aided by ai, might help find more efficient ways to run needed global dialogue.

  • source
  • [–] 4 points 4 months ago

    Indeed - why so low in most of Poland, compared to neighbouring regions with similar climate/crops, also I wonder if tibetan plateau is really so great for Fungi? Or are these artefacts from the machine-learning algorithm?

  • source
  • parent
  • context
  • [–] 1 point 4 months ago

    Good to see where people live, not just colouring whole countries. Although maybe the really empty areas (desert, tundra, jungle) should be paler or dots. Within each family there is much diversity, and each language combines multiple roots. Maybe terms like indo-european are too broad, obscuring the minority mix-ins?

  • source
  • [–] 3 points 9 months ago (2 children)

    Ashes are great for shade just when we need it, as their leaf-season is relatively short. I wonder whether the lack of tree-neighbours helps it avoid the beetles that bring fungi to other ashes. Also how are it’s roots getting water ?

  • source
  • [–] 8 points 9 months ago* (last edited 9 months ago) (3 children)

    This article feels to me really out of date. Scala3 was launched nearly five years ago, The tooling and lib-support was indeed dodgy back then but works very smoothly now. Scala3 also broke Scala2 macros, and some people whose business-model was selling support for clever libraries built on those macros made a lot of fuss (bad publicity). Meanwhile Scala3 has new more robust macros which work fine.

    I develop in scala an interactive climate-scenario model web-app . It's running the model in your browser (500 years x 250 countries x many gases, sectors, feedbacks etc. - so it's complex)... The scala code compiles to js (or wasm) -which is what runs this web app - but the same code also compiles with scala-native to run fast batch- calculations or tests. It also compiles to the jvm app like my older java code, but I rarely use this now.

    Scala3 code looks more like python than java - minimal brackets, and much nicer to read and higher level than rust.
    As for tools I just use Zed editor with Metals for LS, Mill for build, and other libs from the lihaoyi ecosystem, no web 'frameworks'. Scala is both robust and flexible. In general - if the code compiles, typically it runs correctly first time, if not the very-intelligent compiler identifies precisely what to fix where (very different from so-called 'AI'). So instead of reams of junk 'tests', it's usually just enough to check whether my climate system plots look and behave as expected - higher level thinking.

    As for Kotlin it was effectively a russian-led (at the time) fork of Scala, staying closer to Java - so less flexible, but they did much more systematic marketing - and I suspect some of that deliberately pushed blog posts knocking Scala.
    What Scala lacks is promotion, so those following fashions of this hype-driven world won't find it.
    For those who use it, it's a great language, to do complex stuff that scales robustly.

  • source
  • [–] 3 points 9 months ago

    Good, makes sense, as Scala is 'made in Europe' (mainly swiss and polish teams), and makes very robust software. Only it’s under-hyped. Here you can try my interactive climate-system web-model written in latest Scala, which compiles three ways - to the web-app you see, to native code for fast calculations, and to a jvm desktop app (with 25 years history, originally java).

  • source
  • [–] 4 points 10 months ago (1 child)

    Yes. Persistent contrails - i.e. aviation-induced cirrus clouds which spread in supersaturated air layers - are indeed bad for the climate, but often their effect is ignored as hard to quantify, while the simpler small effect of short-lived contrails is conveniently cited instead.
    Also, while all high clouds have a warming effect by reflecting infra-red radiation back to earth, there can also be a cooling effect due to reflecting solar radiation, which is greater when the angle of the sun is low. So the net effect is warming in the middle of the day and at night, but cooling in morning and evening.

  • source
  • parent
  • context
  • [–] 2 points 11 months ago

    SF6 emissions are bad news, but thanks for reporting it. Wondering whether Solvay HQ in belgium had any role in covering this up ?
    I recall using SF6 30 years ago to study ocean gas fluxes, but only microlitres, as we knew about it’s crazy high GWP even then.

  • source
  • [–] 4 points 11 months ago

    Maybe Euroclear is not the only reason - rather a convenient excuse for De Wever to waste time - as his seat of power is Antwerp with a huge chemical industry partly fueled by Russian gas, and his party is right-wing nationalist and maybe more sympathetic to the Putin-Trump vision than they dare to admit (as not in line with sentiment in the country as a whole).

  • source
  • [–] 2 points 11 months ago

    Does anybody know how much energy it costs to dig such tunnels through rock ?
    Anyway, I like the ferries, you get to see the sea, rocks, waves.
    So boring in those tunnels, in some other one they had to put changing colour patterns to stop drivers falling asleep.

  • source
  • [–] 5 points 11 months ago

    Some comments below the article are good. I like especially this one:

    One thing that seems to me to be shaking out of LLMs: if an LLM can do it well enough to pass muster, it probably didn’t need doing in the first place. Coming at it from a programming perspective: if an LLM can generate the required code, it shows that you could have better abstractions that describe your behavior with less code. If an LLM can generate a document, it’s a sign that the document isn’t necessary in the first place- it’s a bit of leftover make work that should be subsumed into a better process. And so on.

  • source
  • [–] 17 points 11 months ago (3 children)

    The money to run EV comes from the voters (paying per sms etc), as well as tv advertising. So in this situation it seems likely that many individuals will also choose to boycott it, choosing not to watch and not to vote. That might lead to a financial penalty for EBU, but at the same time this will distort even further the 'results' of what what may be considered one of the largest (albeit very biased) global experiments in democracy, especially for teenagers. Maybe an alternative can emerge ?

  • source
  • view more: next ›