I think there's almost a 1:1 skill transfer from mentoring people to skilled LLM use. If you can supervise and train someone skilled & confident but with big blind spots, you can use an LLM well. Patience, clear instructions, expecting things won't be perfect on the first try, build skill over time.

I knew it intellectually, but getting an e-bike showed me _viscerally_ how wildly battery technology has advanced.

My (pretty heavy!) pedal-assist bike and I (pretty heavy!) have tooled around a few hours up hills that were otherwise daunting and this battery that I can easily hold is still ~100%.

It bugged the heck out of me when a developer started calling a chunk of downtown Toronto near UofT + the hospitals "the discovery district" and even more when it started getting more more widely used. Only now dawning on me that it's maybe smart long-term messaging to support advocacy?

I do feel like that sometimes! But you’re right, it’s a very dramatic difference

I'd even go a little further and propose that "optimizing for the weirdness of LLMs" is a special case of "understanding the idiosyncrasies of your stakeholders and collaborators and customizing communications for them", although I'm less sure about that.

I think we underestimate how poor the ambient level of collaboration skills is; getting those skills up to work-with-humans level is a much bigger jump than optimizing for LLMs.

Further I think LLM usage can help _build_ basic skills like "making a clear request" and "giving actionable feedback"

Trying to explain St Augustine to the pope, the former head of the Augustinian order, who wrote his doctoral thesis on Augustine, on his way back from celebrating mass at the Basilica of St Augustine in Annaba, Algeria, overlooking the site where Augustine lived is peak Adult Catholic Convert.

I think it's really an underrated moral issue (let along a political economy one) that we've sort of let scams and scam adjacent activities (sports betting, car dealerships, most retail interactions with skilled trades, crypto, etc) pervade the economy from top to bottom.

If we open 150 grocery stores, grocers will actually be paying *us* money to take their food! If we open 250, they’ll be paying us as much as we used to pay!

Each grocery store will by itself save Canadians $1.2B/yr in grocery costs while costing only $6M/yr, generating a free money machine we can scale up.

It’s very simple, everyone. By setting up 50 public-run grocery stores (fine, whatever), one for every 800,000 Canadians, at a cost of $7M each to set up (??) and $6M/yr to run (???) this will cut $60B (???) a year off of Canadian’s grocery bills. Elementary arithmetic.

“Innovation” is something organizations should only be allowed to try after having mastered “best practices”

Eg there’s real reasons for that meaningful difference of why there’s a bunch of off the shelf finance apps for us but not for large enterprises and governments; even the “enterprise solutions” mostly depend on bespoke, long, expensive and somewhat brittle integration work!

There is a bunch of stuff horrifically wrong with governmental procurement especially around IT, and there are people inside trying to fix this (partly by, correctly, trying to build internal capability) but we can’t fix things by trying to pay 350k public servants with stuff we YOLOed together

I’m guessing you’ve since had other conversations about this topic - do you agree that (a) the code per se is only part of the problem and (b) the vetting, integration, and maintenance of the donated system would cost (in $ and in time) would be a significant fraction of the full procurement?

Oh hey, you’re here - I’ve been an on-again off-again subscriber, because you genuinely have a lot of great insights, are a talented and compelling writer, but then sometimes write stuff like that which makes me worry about your other stuff I’ve read.

Hey Matt Gurney, I’ve got a great new personal finance/banking app - it’s awesome, just finishing coding it up just for you. Dare you to reject it!

This is why progressives win in Toronto but never Really Win. Instead of saying "Bradford's support proves we should have moved on platform doors sooner," you leave the issue out there for him to own, and in six months you'll complain that a transit safety and efficiency issue is 'right-coded.'

(Sorry for the reply churn, trying to turn this into something more coherent and a couple notches less intemperate)

Maybe if we started to care about these other public employees too, we could work towards making a better and safer city for them, as well as for us.

All such employees work for the common good, putting their own health in harms way - in this case literally to protect children - and they do it without the high salaries, unassailable job security, deference, and constant stream of accountability nightmares of our police departments.

And I want this for firefighters, ER nurses, public health workers, city-paid construction workers or cleaning staff or anyone killed on the job or dying on the job.

I want a 100-car funereal motorcade of city vehicles for Peter Clark's procession the way we do for cops. I want the roads closed, the City flag flown at half-mast, and it to be covered on the radio the way we do when a police officer dies.

I also feel like consensus has been "mission-creep-ed" towards 'unanimity'. Consensus is a fine, even admirable stretch goal to strive *towards* in a collective decision-making process. Unanimity however - even consensus - is an absolutely insane _requirement_ for a decision making process.

How does that play out - more railing against the feds for lack of pipeline capacity, or more chafing at a boom/bust economy with too many not seeing the boom? Does the provincial government start spreading money around and talk about how AB will help even if Canada won’t?

So $175 for ~18 months is interesting from an Alberta politics point of view, yes? Windfall profits and AB gov’t flush with cash, but huge consumer pain; too short a timeframe and too
much uncertainty to justify investment/jobs other than some debottlenecking/optimisation? Does that sound right?

"You are responsible for what you put into the world and its impact, regardless of how you came to it" actually crosses a few streams of The Discourse today.

There’s very real problems in using effectiveness measures as comparisons *between* organizations - it’s often much cheaper to deliver services to already-well serviced communities, for instance! - but longitudinally *within* an org can be vital, even just for the clarity it imparts

Looking forward to fighting the "irreducible complexity" talking points again, 20 years later and in a completely different context.

Ours the task eternal, I guess