Replying to
@Strandjunker@mstdn.social being kind and listening to the civil servants goes a long way, in my short experience as an elected official
Remote
David Huggins-Daines
@dhd6@jasette.facil.services
mastodon 4.7.3Étudiant en géomatique, consultant en linguistique, développeur de logiciels libres. Piéton, cycliste et skieur. Ex-conseiller municipal et heureux de l'être. Posts randomly in French and/or English. Fan of cheap, durable, yet supposedly obsolete technology in all domains, with the notable exception of 3-pin ski bindings, those are bad and you should not use them.
91 Followers
405 Following
36 Posts
Joined August 14, 2025
Site web (professionnel):
BlueSky:
GitHub:
Site Web (politique):
Nouveau billet de blogue! Créer une carte de zonage interactive, première partie (avec plein de #QGIS et un peu de #uMap) https://ecolingui.ca/fr/blog/carte-zonage/
Replying to
@pinhman@humanwords.party how much did Toronto build since the 70s, though? What is the vacancy rate? How many people are delaying household formation and thus don't even count in the vacancy rate? How long are the waiting lists for public housing? These are the statistics you should care about. Condon was wrong in 2022 and he's still wrong in 2026: https://doodles.mountainmath.ca/posts/2022-01-31-no-shortage-in-housing-bs/
RE: @radiocanada@mastodon.quebec
Rapatrier l'expertise, ça serait la moindre des choses, mais s'il faudrait que les développeurs acceptent de travailler obligatoirement au bureau à des salaires de fonctionnaire... (moi, à part le présentiel, ça m'irait très bien, par contre)
RE: @radiocanada@mastodon.quebec
Je ne savais pas qu'elle était une ancienne de l'IEDM.
Elle dit bien sûr la vérité - c'est assuré que la cible ne sera pas atteinte avec elle comme ministre de l'Environnement!
Bienvenue à la vie après la politique à l'honorable député de Laurier-Sainte-Marie. Crois-moi Steven, c'est beaucoup mieux de ce côté! Tu pourras dire ce que tu penses vraiment en tout temps, par exemple.
Replying to
@heliomass@cosocial.ca I'm old enough to remember, vaguely, Lord Buckethead!
Replying to
@jfmezei@cosocial.ca @EdwinG@mstdn.moimeme.ca don't forget to validate your ticket, on the single validator that's conveniently located on the opposite platform
Replying to
@EdwinG@mstdn.moimeme.ca @jfmezei@cosocial.ca not sure if it's relevant but there is also a special bus-only fare which covers zones A, B and C (not D!) as long as you only take buses.
Replying to
@jonny@neuromatch.social @rysiek@mstdn.social big "the self-driving cars can't help but hit pedestrians so let's just put up fences to prevent pedestrians from crossing the street anywhere" energy
$MSFT "AI chief" needs to $STFU
Je viens de découvrir le site «Better Images of AI» et c'est vraiment chouette. J'aime celle-ci en particulier, «The Sleep of Autonomy Produces Monsters » (Daniela Zampieri / https://betterimagesofai.org / https://creativecommons.org/licenses/by/4.0/)
I did not think it was possible for the reply thread to be even better than the link in the post but stranger things have happened
J'aurais aimé qu'il mette «AI» en «guillemets» ou "guillemets" ou „guillemets” ou bien 「guillemets」 mais à part ça je trouve ceci très juste. L'utiliser de façon responsable ce n'est pas seulement des modèles locaux ou ouverts, mais très spécifiquement des petits modèles «plates» qui aident les professionnel·le·s à faire des choses utiles pour la collectivité https://danielvanstrien.xyz/ai-patterns-for-glam/discovery/boring-ai.html
"Then, they suggested, roughly following the 2008 credit crunch, we've had some 15 years of bullshit in tech."
(This person sounds an awful lot like a former colleague from 1999 who I'm pretty sure still works for Red Hat 🤣)
Replying to
@simon@gotosocial.grnwds.uk @edmond@vael.town meta helpfully provides this invocation to find all of their subnets (there are lots of them): https://developers.facebook.com/docs/sharing/webmasters/web-crawlers#identify-5
Replying to
@mttaggart interesting... Deleted my previous reply since I assume their calculation of CO2e takes the electricity mix into account, but it's still surprising given that the French grid is 67% nuclear and 25% renewable. Did they train the models somewhere else?
Replying to
@mttaggart they don't seem to have done a similar estimate for OLMO3 unfortunately
Replying to
Replying to
@gatesvp@mstdn.ca @emilymbender@dair-community.social as the other reply says, it's all about informed consent and participation (in other words the exact opposite of what we currently have with LLMs), but also one does not necessarily need enormous data sets to do useful things! Rumors of the demise of "obsolete" technologies like WFSTs have been greatly exaggerated 🙂
Replying to
@simon@gotosocial.grnwds.uk @edmond@vael.town if you don't mind playing whack-a-mole you can rate-limit them down to something silly, since they seem to operate from a previously downloaded list of URLs and don't take a hint once you start denying them access...
Si tu cherches le vrai trafic 🔥😈🤘infernal à Québec, c'est pas sur les ponts... https://ici.radio-canada.ca/nouvelle/2256647/etude-circulation--mtmd-autoroute-capitale
Replying to
@infinimatt@hci.social gonna chime in and say that #notallxml here as some applications are legitimately excellent, like SVG and XSLT ... and then, there's all the rest of them (Maven POMs for instance 🤮)
Whoa, keyman-ibus really, really doesn't work on Debian 13 under Wayland, and merely doesn't work under Xorg. I've been teetering around the brink of this rabbit hole all week and will probably just write my own input method in Emacs Lisp for what I need, but is there anyone out there who has any experience with this stuff who could explain how to debug it? It's a bit complicated to explain the problem unfortunately...
Replying to
@pluralistic @tante @simonzerafa As always, yes and no. A bug zapper is designed to zap bugs, it is a simple mechanism that does that one thing, and does it well. An LLM is designed to read text and generate more text.
That we have decided that the best way to do NLP is to use massively overparameterized word predictors that we have trained using RL to respond to prompts, rather than just, like, doing NLP, is just crazy from an engineering standpoint.
Rube Goldberg is spinning in his grave!
Replying to
@pluralistic @tante @simonzerafa indeed, I guess the question is whether the scale of the *ahem* waste, fraud and abuse *ahem* of resources that LLMs seem to imply, even in benign use cases like yours, is out of line with historical precedent or not.
Am I an old man yelling at a cloud?
No, it's the children who are wrong!
Replying to
@jorijn@toot.community @forgejo@floss.social I suppose you could have a persistently running VM and spawn containers inside that 😀
Replying to
@EdwinG@mstdn.moimeme.ca @mhoye@cosocial.ca @mayintoronto@beige.party not sure about the smelling like French fries part, that's hearsay (and it's been some time since I've ridden an MR-63), but the peanut oil is true even for the latest trains (which don't brake very much at all)
Étant donné que ça roulera sans problème sur ma collection de petits GPUs à 100$, j'ai hâte de l'essayer sur d'autres sortes de données semi-structurées (oui bien sûr, des grilles de zonage historiques obtenues par l'accès à l'information 😉 ): https://danielvanstrien.xyz/posts/2026/structured-records-from-cards/
Replying to
@EdwinG@mstdn.moimeme.ca @mhoye@cosocial.ca @mayintoronto@beige.party Wikipedia suggests the odor is more like popcorn 🍿
Replying to
@pinhman@humanwords.party the 1970s were *checks notes" half a century ago
Replying to
@tante @pluralistic @simonzerafa But ALSO: using a multi-billion-parameter synthetic text extruding machine to find spelling and syntax errors is a blatant example of "doing everything the least efficient way possible" and that's why we are living on an overheating planet buried under toxic e-waste.
If I think about it harder I could probably come up with a more clever metaphor than killing a mosquito with a flamethrower, but you get the idea.