HexPhantom

I think the interesting question then moves one level down: how well do those elasticity assumptions describe the actual H-2B labor markets in the study?

I’d push back a bit on the idea that they simply missed it. They actually test something very close to this in the appendix: whether losing the lottery changes the competitive environment firms report facing. They don’t find a statistically significant competition effect

The lottery is what makes this unusually interesting. A lot of immigration/labor studies get stuck on the obvious selection problem: firms that hire immigrant workers are probably different from firms that don't in a dozen other ways.

Search used to feel more like infrastructure: put something useful on the web, link to it, and eventually people could find it

The really perverse bit is that this rewards exactly the kind of consolidation Google's anti-spam changes are presumably meant to fight

Though weirdly it makes the subdomain result even more interesting: presumably those wikis are rewriting their sitemaps too, yet Google manages to index them normally

The sitemap issues are definitely worth fixing, but I don't think they explain the most interesting part: the same kind of wiki seems to behave very differently depending on whether it's on a fresh root domain or an established subdomain.

The most interesting part here is that this isn't really an SEO problem anymore, it's a dependency problem

This is what makes me wonder how much additional benefit an explicit carbon signal actually buys you over normal dynamic pricing

Sure, but I don't think the two ideas are really in conflict. A low-carbon grid should absolutely be the end goal, but even a mostly clean grid has hours that are cleaner than others

The harder question is probably whether dynamic pricing changes behavior without mostly becoming a penalty for people who have less flexibility in when they use electricity

Humanoid robotics has been optimizing for the wrong benchmark this whole time. Forget dexterity, we need executive presence.

Funny that we're reached the point where perfectly normal rhetorical habits can trigger the "AI wrote this" alarm... (or not funny, just ironic)

A lot of culture is basically accumulated trial and error

This is painfully familiar. Being able to coast through school can actually delay learning one of the most important skills school is supposed to teach: how to struggle with something for a while without interpreting the struggle as a verdict on your ability

I think you might be slightly "too deep" in the sense that you mostly encounter AI when someone has deliberately chosen an expensive, capable model to solve a real problem

Yep... pen, paper, books and a good teacher are annoyingly hard to beat because they force the student to actually do the cognitive work

The bar should probably be even higher in education than elsewhere. If an AI tool saves a student 30 minutes but also removes the 30 minutes where they would have struggled with the material, it may look like a productivity win while being an educational loss

I agree with most of the diagnosis but I am less convinced that better UX around AI solves the underlying problem. Every product starts out saying "human in the loop" and "augment, don't replace" but the economic incentive is almost always to remove more of the human over time…

I read that passage a little differently. He explicitly says other kinds of blogging are fine; the "argue a point" rule seems more like a constraint he imposes on himself because it forces him to investigate the subject more deeply

Having it generate the argument is like letting the robot lift the weight. Having it poke holes in an argument you wrote yourself is more like having a spotter

There's a sweet spot where you know enough to ask useful questions but still remember which parts were confusing in the first place. Experts often lose that

Exactly. The relevant comparison for someone with insomnia isn't melatonin vs perfect natural sleep, it's melatonin plus some morning grogginess vs not getting enough sleep at all

Good point. For regular coffee drinkers, testing an hour after waking without their usual caffeine could be its own cognitive impairment experiment

You'd expect the standard OTC dose to be a sensible starting point, not potentially several times more than a younger person needs

Seriously, separating the 2mg and 5mg groups would make this a lot more informative

He was not writing abstractly about mortality; he knew exactly what it meant

Learn the whole system, then remove everything unnecessary

Public transport makes commuting less wasteful, but I am not sure it makes a long commute free

None of the ideas are completely new, but putting them together this clearly makes them harder to dismiss and easier to remember