How would the residents of this street know that Waymo had received all of these tickets, and that they were unpaid?

311 is the online request service for the city of San Francisco. The residents are complaining and nobody is responding, but that doesn't mean that Waymo is not paying their traffic fines.

Everyone here in the US is fully committed to doing all their work with AI, which means we're going to need people to pick components up and connect them together to make anything with real-world utility.

I think the author is optimistic about how good it was 20 years ago. Perhaps they are young. I got into writing code in the 80's, and for a living in the 90's, and we've been so bad for so long. A Calculator leaking 32GB is not necessarily bad - it could be a bug. The crime is that the damned Calculator requires 20MB of memory from a standing start. The crime isn't that some process periodically gets out of control and uses all the CPU and memory, that can be fixed, the crime is that my system has hundreds of background processes running after a fresh login. Nobody is able to grapple with that, so there is basically no chance that someone might think of a clever way to rearrange things to use half the resources. In fact, the best case is really that someone notices that three processes use shared structures, so they spin up a project to build a fourth process that the other three can be clients of.

When you come down to it, the real problem is that a lot of today's software isn't really necessary and doesn't really serve any particular end goal in terms of functionality, it's just aiming to track your usage and monetize things, so there's really not much point to doing a good job.

On the up side, this does mean that if we survive to see the very long term, we will be able to salvage a lot of performance out of simply wrangling away the lazy code. Speeds won't double every 9 months, but in 50 years our basic calculator programs might only need 4mb of memory. Unless someone adds skinning support.

I've been programming for 40 years, and awhile back needed to install VSCode to test out some Marlin changes. It's just like rainbow unicorn vomit. I know that different people have different preferences, but I simply cannot comprehend how anyone can get anything done with 10 separate panes of crap going on at the same time. Apple's Xcode system similarly lost me at some point, after a point, more is just more, showing me everything at once detracts from architecture and design considerations.

I'm not going to claim that Arduino produces a strong IDE. I just don't think that Winamp is a good model for an IDE.

Imagine a world where you have people to drive you around, people to make sure you are fed, people to make sure that other people don't bother you, people to make sure you have Internet when you need it and quiet when you don't, people to make sure that you don't have to wait in line at the DMV or the airport or Target.

Basically, these people have no idea what work/life balance is, because they never actually work, they just walk around imagining things all day and having other people do the work.

The recent news on Intel has not been the kind of news one associates with a successful company cruising from strength to strength. Neither of these "investments" seem likely to generate any sort of response to inroads by AMD and ARM.

Perl was one of my go-to languages for the longest time, but where I always got lost was when a system grew too big. I'd start with 20-30 lines, grow and grow, and then just lose containment. Being careful about using unit testing to help drive appropriate encapsulation helped grow things a bit bigger, but systems just tended to reach the point where all my time was spent fixing cascading issues caused by trying to add a feature or fix a bug.

For awhile, I was using Go for the bigger projects, and that worked well. I have found that gradually my Go usage is extending down towards simpler projects. Perl still has a material advantage when I just want to whack together a 30-line script, but if I know it'll be over 100 lines, I'll spend a couple minutes up front setting it up in Go. The main thing sticking point is a bit of clunkiness around spawning subprocesses and processing their output using regular expressions. It's not HARD in Go, it just involves a fair bit more setup compared to Perl, where there is basically no setup at all to do this because that's what Perl _does_.

You laugh, but ... I've spent the past few years trying to track down some physical deterioration which was really affecting my ability to get out and get even basic stuff done. Eventually we got to the sleep-doctor phase of things, and he reviewed my questionnaire and gave me the same old list of sleep-hygiene stuff, as if I just hadn't been listening for the past 50 years. "Oh? Just go to bed earlier? Why didn't I think of that!" And he prescribed a sleep study, just in case.

Turned up sleep apnea, now I'm on a CPAP. The difference between how I wake up now is like Dorothy walking out of the black-and-white house into Oz. Not just color, Technicolor (TM). I looked back at the sleep-doctor's quiz, and I think the problem is that I have been in such a deep hole for so long that I could no longer recognize good sleep from poor sleep. All I had gotten for many years was bad sleep or terrible sleep, so a "good" day was being able to think clearly by noon rather than 2pm.

I don't get much/any Republican email spam, but the shit I get from Democrats looks like it was written by an eighth grader who has gotten access to an espresso machine, and that is being generous. It's embarrassing. It's easy to foam at the mouth about "They're suppressing our emails!", but I think it's likely that if they documented the emails being "suppressed", most people would agree that the algorithm isn't making a mistake. The kinds of marketing incentives political operatives live under are basically identical to those that spammers live under, I wouldn't be surprised if they recruit spammers for their email campaigns.

I review my spam box periodically. VERY little in there is coherent. The periodic ham bits generally contain low-content commercial email from companies I do have a relationship with, so they aren't spam, but I completely understand how they got there. Quite infrequently I'll rescue an email from a known person, and they are almost always trivial responses like "Me too!" or somesuch, again, completely understandable how they get mis-classified.

Part of the reason that Linux has troubles competing with the commercial operating systems for mind share is that once things are to a prosumer level of fit and finish, people start spending effort on minor respins to build semi-custom distros to target specific market segments. This is great for prosumer users who enjoy tinkering around to get a perfect fit, but it doesn't stack up into a strong offering for casual users, because often each respin is backed by one or six people who don't have time to properly nail down all of the exposed edges and corners.

It's a little like how PC vendors all tend to have 8 separate lineups with significant overlap. More choice doesn't help, because you have no idea if you want a unit from the X39-extreme line, or from the B2-19Gr4 line, or perhaps the Alpherio line would be a better fit? The reality is that it's just a stew of acronyms, most of which aren't super relevant, and the noise makes it mostly impossible to tell which will serve you better.

Low velocity in the marketplace is not great, because it inhibits price discovery. But we are well past the point where traders at investment banks are actively skimming profits off the market. It is easy to assume that those profits are just coming from other high-frequency traders - but, it reality the aggregate of the HFT profits are coming from the rest of the market, the buy-and-hold investors who's money is sitting in retirement funds. High-frequency trading is like a kind of inflation which is slowly eating away at your assets.

Unfortunately, the problem is that if the main exchanges don't provide what the trading desks want, they'll just shift to dark pools where the public has no visibility at all.

There have been past projects to do this. They didn't. Intending to improve things is not the problem, improving things without destroying the entire system is the problem.

Imagine that you're working in a regular old IT shop EXCEPT that your bosses are required to always shop projects out to third-party contractors with no skin in the game, and you effectively have somewhere north of 535 "bosses" who bungi in periodically to make you the background of a press conference. Imagine anytime you take a risk on something and it doesn't go perfectly, half of them go on Fox News to scream about you. Then imagine that you actually put in your time and get the shit done in spite of all that, and some 20yo college dropout swings by and gives you the boot.

Yeah, it is SUPER WEIRD that nothing gets done.

Some studies suggest that we're extinguishing 2% of insect biomass per year. We're extincting insects all the time because it's convenient to raze a forest to grow cattle. Even doing the research in preparation for intentionally extinction a mosquito species would probably tell us all sorts of horrors we're storing up for the future ... and we probably won't change anything.

I did hunt-and-peck for about a decade of my career. Then I realized that I was often sitting there staring at my keyboard, not at the output on my screen, so I'd have to then go back and find my cursor and probably fix some errors (again looking at the keyboard). So I sat down and taught myself to touch type. And then I could touch type. And then it wasn't a problem any longer.

Something that isn't clear in this anecdote is that for those ten years, I wasn't employed, I earned my keep doing software development independently, and I started that while I was still in college. I didn't ask someone's permission on whether to write code without touch typing, I just went and did it, and since I was paying the bills, nobody cared. Can someone make a living as a software developer without knowing how to touch-type? Sure! But if you aren't going to bother to learn a skill that is obviously useful in your field, then you'll need to have compensating skills elsewhere, which is to say that someone who can't be bothered to learn to touch-type most likely doesn't have those compensating skills to fall back on.

Like, seriously, this is a trivial thing litearally anyone can do to make themselves more employable. Who even cares about quantizing how useful it would be? For sure some people can't learn to touch type, like literally, because of neural wiring or other issues. But, unfortunately, NOBODY CARES, that's a you problem. If you're in that position, you'll need to invest time in compensating elsewhere for the fact that you can't equip the touch-typing skill. It will probably involve a lot more work than it would take the average person to learn to touch type.

People honestly have no idea how much harm it does to their career when the most basic project turns into a subquest. People gain stature not by being ABLE to do things, but by having a faster cycle time. Remove barriers rather than preserving them. Quit asking for the parameters of your accomodations and just learn your shit!

Like will they add another 32GB or so to my computer to account for the memory they are devoting to this new function which I don't want? It's embarrassing. "Speed, Security, Stability, Simplicity" indeed.