mumblemumble

True.

That said in any non-trivial project, writing in the first place and rewriting aren't too different from each other in practice.

Assuming someone doesn't have the ability to maintain a consistent and always-accurate mental map of tens of thousands of lines of code, most of which they probably didn't write themself and some of which they might have never actually looked at before, they're going to very rapidly reach a point where most of the work involved in adding a new feature consists of reading existing code to understand how it behaves and how the new code needs to interact with it. So the expensive part of adding new code to a work in progress, and the expensive part of rewriting existing code, are more-or-less one and the same.

So, that code gets me thinking about premature optimization. In the "verschlimmbessern born of optimizing the wrong thing because you didn't use measurement to guide your efforts" sense.

The rule of thumb I hear is that, in a mature product, reading and maintaining code takes about 10 times as much effort as writing it in the first place. I've never tried to measure this myself, but it doesn't seem to be wildly off from what I see at work, so let's go with it. Let's also assume, for the sake of argument, that these AI tools double the productivity of people writing new code. Or, equivalently, they halve the time it takes to write it. (This is a lot more than what I see people typically claim, but I'm trying to steelman and also it makes the math easier.)

Anyway, this would imply that, speaking purely in terms of raw productivity (not, say, security or correctness) the AI coding assistant is a net win when the code that's written using it is less than 5% more difficult to maintain.

And I'm inclined to think that, if Copilot & friends are enabling more people to write more code that looks like what we see in this article (also not wildly off from what I see at work), then it's hard to see how they could possibly be making codebases where they are used less than 5% more expensive to maintain.

You spin up a fresh server from a cookie cutter image that lets you create a practically unlimited number of identical servers.

That's not even remotely like how rewriting code works. Rewriting code is more comparable to what spinning up a fresh server was like a quarter century ago. So, back in the days where they were important things because spinning up a new one was an unholy PITA and literally never went off without a hitch.

"Lower prices" doesn't necessarily mean they just suddenly and immediately drop. That's no surprise; dropping prices purely out of the goodness of your heart isn't terribly good business practice. Also, for a lot of retail, MSRP is MSRP, and that's a pretty big anchor point.

What I'd expect instead, based on my having taken exactly one class in economics as an undergraduate, is subtler effects that play out over time. Maybe the general growth in prices over time slows down a titch until a new equilibrium point is met. Maybe wages rise a little bit because retailers can afford to pay their employees more. Maybe life gets easier for smaller businesses that have less negotiation power than the multinational behemoths. Maybe some bank executive somewhere decides not to buy that third luxury car at the same time as ten thousand restaurant owners decide that, just today, they will treat themselves to an espresso drink from the coffee shop instead of making drip coffee at home. That kind of thing.

I think maybe that last example is most interesting to me, because it calls attention to how merchant/consumer is a false dichotomy and things are always a bit more subtle than how the news likes to make us think they are.

Apparent earlier draft of the paper: https://mpra.ub.uni-muenchen.de/100393/1/MPRA_paper_100393.p...

I haven't paid to read the linked article, but the abstracts do differ significantly. Perhaps the work in the above link is presumably not representative of the final published version, which is about 20 pages shorter. That could just be formatting differences, though. About 60% of the pages in the earlier version are figures and tables, many of which get, but do not fill, a dedicated page. The author has presumably had ~5 years to keep working on it in the intervening time. That could either be a good thing or a bad thing, depending on various factors including one's own opinion on how the scientific process should work.

Personally, I'm always pretty a priori skeptical of this style of economic "natural experiment". I'm rather fond of economist Ronald Coase's observation that, "if you torture the data long enough, it will confess." And 5+ years is a long time to spend p-hacking one set of data. But it's hard to know what all is in a 30-page paper based on just a 150-word abstract. So I would encourage people to be very cautious about getting too caught up in dissecting and discussing this without taking some time to digest the whole thing first. Arguably Hacker News is just not the right venue in the first place because it's set up to reward opinions that people can assemble in a matter of minutes, and digesting and understanding the merits of a 30-page economics paper probably deserves more like a week's worth of time.

My understanding is that excise taxation is generally agreed upon as the preferred method for governments to manipulate the market, both because it incurs fewer deadweight losses and because it tends to actually work better.

So, what if we get rid of all these complicated rent control and rent freezes and affordable housing schemes, and instead just implement a rent tax, to be paid by the landlord, and make it progressive? I don't know exactly how it should scale; you wouldn't want it to be just by rent because that would have a regressive impact on families who have kids, because they need more space and more space naturally costs more. Maybe price per square foot?

At least in my city this would probably also reduce real estate prices in general, because a huge source of demand for houses is actually real estate speculators who buy up houses and then put them on the market as rental units. I gather, based on one conversation with an acquaintance who had been a realtor but was looking to pivot into this line of business, is that a lot of what's fueling that is, in effect, not-exactly-loopholes in US and local housing, lending and tax laws - many of which are ostensibly aimed at making housing more affordable - that allow people with sufficient resources to financially engineer together a speculative source of income while externalizing all the risk onto everyone but themselves.

And even then, I no longer believe that "removes as much friction as possible compared to local functions" is a desirable goal in the first place.

Over the course of my short 20-year career, I've encountered far too many applications that accumulated terrible performance or reliability problems that were directly attributable to their authors feeling a bit too comfortable about getting chatty with the slow, unreliable remote procedure calls.

By contrast, the most successful microservices shop I ever worked at was very conscientious about doing nothing to hide the messiness of RPC. (Ease the implementation, yes, but not hide the messiness.) And that seemed to foster a culture of being much more thoughtful about interaction patterns and data flows and how they might contribute to or detract from the overall performance and stability of the system.

The last mile might not be too bad, but Google Maps says the total trip is 2 hours one way.

I haven't. Brookfield isn't really accessible if you live in the city and don't have a car.

If we renamed regulatory terms of art every time members of the public misinterpreted them, we would be changing terminology so often that we'd end up with everyone being confused, not just some people.

It was for manufacturing. It's not as great as one would hope for actually playing. The city has an ordnance that says that you need a special "public place of amusement" license if you have more than 3 machines and are open to the public, and it can be rather expensive. So places to play are generally limited to a relatively small number of barcade-type venues that tend to be loud and crowded.

I prefer the pinball scene in Milwaukee. Seemingly every neighborhood has at least a few relaxed, cozy bars with a decent number of machines in the back room, and there are also a few decent venues that are open to minors.

I don't know Portland as well, but based on what little time I've spent there it seemed like the situation was more similar to what it's like in Milwaukee. I've stumbled across some great old classic machines when I was there.

The organizer of a pinball tournament league I used to participate in expressed regret about buying his own machine. The up-front price of buying one isn't terrible, no. What he disliked was that he felt committed to that one single machine, and ended up playing it so much that he was thoroughly bored of it within a year.

OK, so this is definitely something to follow up on. But also, that other recall for the warning lights being too small that the article mentions really caught my eye. Seems like Tesla failed the brown M&M test[1] on that one.

1: https://www.snopes.com/fact-check/brown-out/

Read about different note-taking methods, try them, and decide which one you like best.

The best tool for that kind of thing is developing good note-taking skills.

When I'm doing language study, for example, the backbone of my memory reinforcement strategy is a plain old pen-and-paper notebook. Anki is only used for specific, targeted reinforcement.

Most my time with Anki these days is learning Chinese characters. That's arguably an ideal use case for brute-forcing with SRS, and that is indeed a very popular way to learn them. But I prefer to start with good note-taking there, too. I keep a notebook where I write down new characters and make some notes about their composition, etymology, and the nature of any relationship they might have with other characters that have a similar appearance or show up as components in this new one. IME even the simple act of physically jotting down a handwritten note to not confuse 买 with 卖 or 找 with 我 is worth some large number of flashcard repetitions all by itself.

I also create cards for hanzi in Anki, but typically only after I've already encountered it a few times in my reading and it's starting to feel familiar. Leech cards are a huge waste of time when tackling a large subject, so I don't really like to add a card to my deck until I'm reasonably confident that I won't be hitting the "again" button on it more than once or twice, if ever.

I rent individual movies for $3-4 apiece, and that turns out to be a big win over Netflix because I typically only watch 1-2 movies per month, and the rental service has a bigger library.

But I think that I'm a very unusual consumer in that respect. If I liked to watch TV series more then I'd be bingeing those, and that's where I think Netflix really attracts consumers, not movies. I don't think there's an economical way to do a la carte viewing of TV series on a regular basis.

Take a look at a video of a pin setting machine in action - https://youtu.be/Iod6uwUGM2E?si=BKNipS2E-geJEJEk, for example.

Basically, it's not just the force to move a single 1.5 kg pin. It's the force needed to move ten of those pins, plus the machinery for lifting and setting them into place, all of which needs to be overbuilt because it's being repeatedly hit by flying pins and possibly even the occasional ricocheting 7kg ball, and needs to stand up to that kind of treatment day in, day out for years or decades without completely falling apart. And then, because people like their games to keep moving, the drive mechanism for moving that whole huge assembly around needs to apply enough force to be able to raise and lower it quite quickly.

I'm no expert in any of this, but just taking one quick look at it I'm not at all surprised to find out that these machines are invariably hazardous. And probably most the ones in operation are even scarier than the one in that video I linked above, because it's a relatively recent model and, at least at the place I have been bowling, all the equipment looks like it could easily be older than I am.

If you want a system that only needs to be able to apply enough force to lift a single 1.5kg pin, well, it sounds like that's why the string setting machine was invented.

Zip disks were Iomega.

That's presumably what the last 4 or 5 words of the post were alluding to.

I think, though, that your argument depends very directly on hardware implementation concerns, which is exactly the kind of thing that big O notation is meant to abstract away.

IEEE 1789 lays a lot of that out. In a nutshell, LEDs are harsher.

Even flicker that isn't perceptible to humans is still associated with increases in all sorts of things including eyestrain, headaches, anxiety, migraines, and seizures.

150Hz is well below the human perceptual threshold but still within the range that can cause problems.

That's kind of exactly what I'm getting at. Officially licensed Unix is all but dead. It was outcompeted by freer (choose your own capitalization, either works) POSIX operating systems.

  > Remember when UNIX was given away free?
  > How many of you are using UNIX now?  It is licensed.

Fast forward 30 years... or 20. Ten, for that matter.

I guess it's a little hard to find from their home page because the top search widget thingy is for longer-term reservations, but if you scroll down and choose the "coworking" tab, there's a tile for coworking day passes with a link you can click for the reservation page.

Seems unlikely they'd go for that under any circumstances. I know they claim billions in assets, but I'm guessing it's all stuff whose resale value has tanked since 2020 and WeWork's actual liquidation value is $0 give or take.

Regus has been selling that for ages and ages. They have more locations in more cities in more countries, too. If they're not as well-known, I'm guessing it's because they're just a boring old real estate company and not a SoftBank-fueled moon rocket.

I have learned to expect buses to be delayed by up to 45 minutes during commuting times. Trains slow down at peak hours, too, because more people getting on and off means more time stopped in stations, but in my experience it typically only adds about 5 minutes to a 20-minute trip.

I am strongly coming to suspect that a lot of the talk about buses being all you need is fueled more by a sort of NIMBYism than by a fully-informed position. Buses are great and good transit systems need more of them than they do trains, but an ideal urban transit system has both.

Buses are only a good compromise as long as car usage remains low.

For example, in Chicago car ownership has increased as wealthy people move back into the city, and quality of bus service has declined because buses are also stuck in all the congestion that the cars generate. The city has started creating dedicated bus lanes, but they aren't actually improving the situation much because painting the right lane red doesn't actually prevent it from being jammed full of cars waiting to make right turns and double-parked delivery and rideshare vehicles.

That isn't a whole lot more than the cost per kilometer to construct urban roads with a comparable passenger throughput, though. And construction cost alone isn't exactly a fair price comparison because users of urban roads tend to need to supply their own vehicles, at considerable cost to themselves, and more money also has to be spent to provide space to store those vehicles when not in use.

And you could argue that roads are actually even more expensive to invest in because they tend to create a feedback cycle that promotes even more spending on road construction. Because as more people use roads, you need to widen them to increase their carrying capacity, and because road-oriented infrastructure encourages sprawl that pushes people toward choosing driving, and spending more time on the roads, which increases the need for more roads to be built.

A nice case study here is Chicago versus Detroit. Proportionally speaking, they had similar and approximately contemporaneous population growth and loss trends over the 20th century. But that trend didn't cause the same financial catastrophe for Chicago that it did for Detroit, and one argument for why this is is that, since Chicago retained a denser, more public transit oriented urban plan, its infrastructure costs relative to population stayed lower, which in turn meant that it was able to weather a lot more population decline without its economy shrinking too much to be able to affordably maintain its infrastructure, including transportation infrastructure.