I have a similar dilemma with my car. I drive a 25-year old Lexus with a bizarre electrical glitch. The ABS sometimes goes off as you come to a stop, for no reason at all. It only ever happens below ~10mph, and only when decelerating gradually. Never happens under heavy braking. It's not a safety hazard, and honestly you get used to it. Yet, anyone who test drives this car will run for the hills because it feels spooky.
It's still a terrific car. Comfortable, well made, fast enough for all practical situations. Unusally low mileage for its age. An engine that's sought after in the tuner community. But, it's unsellable. I'm stuck with it, whether I like it or not.
The good news is that I like it. The funny news is that I took a new job that will move me to the Bay, and whatever my new employer is paying to move my car out there is definitely more than my car is worth.
> Manufacturers should fit all-weather tires by default (not all-seasons) - they are decent in both summer and snow (3PMSF).
Won't happen. Tires affect fuel economy in EPA testing. Your commuter car will always come equipped with the hardest all-season or summer tires the manufacturer can source.
>Your commuter car will always come equipped with the hardest all-season or summer tires the manufacturer can source
And instead of taking a step back and realizing that there are competing tradeoffs here and that a compromise needs to be made people will just screech harder about "the side of the tradeoff I care about is not being pushed hard enough by the .gov".
I grew up in the Chicago suburbs, but my parents and my whole extended family are from an area in eastern Michigan, about 2 hours north of Detroit.
Especially back then, the local economy was heavily dependent on Dow Chemical, who has a massive facility in the area. My grandfather worked there. He was one of the first people to ever handle Saran wrap. He also recalls hilarious stories, like a time when someone pranked the foreman by dumping some stuff in the soap dispenser that turned into a sticky, snotty goo when exposed to water. He also worked in an area that used lots of iodine for a couple years. His whole body was sunflower yellow when he came home from work, but he says he never got a cold!
Some stories are more harrowing. There were air raid sirens to warn people when Dow was venting something into the atmosphere. If you heard the sirens, you went inside ASAP.
And then there was the dioxin plant (aka agent orange). He says men in their 40s who worked in the dioxin plant looked like they were in their 80s. Many hard-working people died young there.
Luckily, he was never in the Dioxin plant on a daily basis - he was a diesel mechanic and a welder. Had he been in the dioxin plant, he probably wouldn't be alive today to share those stories. I fear that, as this generation leaves us, so will the cautionary tales.
> I fear that, as this generation leaves us, so will the cautionary tales.
Cautionary tales that are not being learned where the similarly dreadful silicosis is concerned; there is an incredible increase in the number of cases of silicosis among kitchen fitters working on granite worktops.
(I fear we may also find this issue among people working near 3D printers using glass and carbon-fibre-filled filaments.)
I (in Australia) was looking at weatherboards on the weekend for some work i’m doing and came across the James Hardie ‘Linea’ weatherboards - they’re fibro-cement boards. A couple of things struck me:
1) I’ve noticed the James Hardie ‘brand’ being used openly a lot in the past 5 years. I guess the stigma from their evil handling of asbestosis claims is gone.
2) In the installation manual it had an extensive list of things you have to do to not get silicosis.
No thanks Satan, regular pine weatherboards for me!
We spent some time this summer in the Saginaw and Midland area. The "mark" that the Dow family made is very visible, insofar as charities, buildings, etc. I would imagine that's the tip of a very scary industrial contamination iceberg.
Yep. Midland is basically Dow, the City. My dad went to Dow High School. One of the main tourist attractions is Dow Gardens (worth visiting if you didn't, by the way). Or you can go watch a baseball game at Dow Diamond. It's pretty, too - one of the best collections of Prairie architecture in the world... much of it designed by Alden B. Dow, son of the company's namesake, and student of Frank Lloyd Wright. Just don't go in the Tittabawassee river.
We were only in that area for a couple of days. It seems to be a nice area. Midland has a bit of a theme park / artificial feel to it, though. Saginaw seems a little more "authentic".
We were impressed by Dow Gardens. I'm also particularly interested Alden B. Dow's architecture. I'm hoping to attend one of the various tours the studio offers next year.
I happened to see an article[0] about a Dow house coming up for sale last year. That ignited my interest in that area of the state. (I'm not a credible enough potential buyer to actually see this property in the flesh, sadly.)
Some of my family lived in the Traverse City area and we still have property there. We spend time there every summer and, coming from western Ohio, we always take a westerly route to get up there. I'd never spent any time on the eastern side of the peninsula before. We're definitely going back for another couple days sometime.
Hah, I can see that perception for sure. Downtown Midland is very prim and proper, and doesn't really match the culture of the rest of the Tri Cities. Bay City has an adorable little downtown if you're ever in the area again. It's a must-visit if you like antique shops, and much more authentic.
In college, some times a kid would go pass out in his car rather than drive, and people would Saran Wrap the car so they couldn’t open the doors to get out.
I spent months doing research for a blog post about One Laptop Per Child last year, and came to a related, but more broad conclusion: it's extremely easy to reach misleading conclusions when studying novel educational methods. No strong conclusion comes without qualifiers related to culture and economics. Moreover, a shocking amount of harm has been done by people trying to apply an educational method outside of the socioeconomic context where its efficacy was proven.
There's a dilemma here, because in order to find ways to improve education, we have to try stuff, right? But how do we remedy the situation when those experiments fail? That's less related to the Montessori thing, but it's interesting to think about.
In OLPC's case, the remedy was a retroactive, panicked attempt at teaching teachers how to use the laptops, an effort that largely failed.
You're exactly right, though. OLPC failed mostly because it didn't think to teach the teachers how to use the laptops as classroom tools (not that they would have succeeded otherwise). Countries that had the infrastructure to do the onboarding themselves were relatively well-set up to teach their kids anyway.
If this is interesting to you, I highly recommend Morgan Ames' The Charisma Machine.
i think the constructivist idealism got in the way here. i believe the expectation was that they wouldn't need to train the teachers because the students should figure out the laptops on their own.
seems they missed that figuring out the laptop and integrating it into the curriculum are two different things.
i read your post btw, one thing i am wondering about is that you wrote that countries didn't improve electricity in schools because OLPC claimed that this wasn't necessary.
my own speculation is that they simply didn't enough research and didn't expect that the situation would be so bad. it is also my understanding that the hand crank was dropped early because the laptop could not handle the physical stress of cranking, it would break apart. but then a separate hand crank charger was eventually produced after all: https://wiki.laptop.org/go/Peripherals/Hand_Crank but if i did the math right then it would still take an hour to charge the battery with that.
since there was no hand crank the need for electricity was already well known before any deployment, and part of the deployment efforts included improving the electricity infrastructure.
> i think the constructivist idealism got in the way here.
Oh yeah, they had their heads miles up their own asses. That's absolutely a major part of the story. Idealism led to hubris.
The hand crank is a big part of the story, though its role is more complicated than you might expect. It wasn't a silver bullet. Some developing nations, like Paraguay, had decent electrical infrastructure, but their OLPC deployments still went poorly due to lack of training and lack of maintenance/repair programs.
Also, it takes very basic physics to prove that the hand crank could never have been the silver bullet in the first place. My math in the essay agrees that no child-operated hand crank was ever going to be sensible.
DNT doesn't solve all problems, though. Not only is DNT being deprecated, it also lacks the proper customisability the law actually prescribes for data processing.
There's no value you can give DNT that says "you can do your own on-site tracking and telemetry and I accept sharing my data with Sendgrid for your newsletter, but I do not want third-party trackers".
As a practical example: there are news sites that will not play videos if you hit "deny all" because their video host does some viewership analytics. I'm fine with that, but not the 750 other advertisers the news site tries to have me track.
Of course, "deny all" should be an option, "accept all or deny all" isn't control.
For the longest time we had https://en.wikipedia.org/wiki/P3P as a basis to build on, but that officially died the day Edge became Chromium-based.
> you can do your own on-site tracking and telemetry and I accept sharing my data with Sendgrid for your newsletter, but I do not want third-party trackers
I'm sorry, but does a user who would want this actually exist? This seems like a hypothetical dreamed up by the marketing team to avoid having to accept that a large group of users hate all their tracking shit.
I do not want my data sent to data brokers or used for advertising. I have less of an issue if my data is used to improve a service I use and only for this, as long as I value/trust the service. The problem is that many websites really want to sell your data to third parties and/or use if for advertising, that often it feels safer to just refuse any consent.
Yes, it's quite common for users to want this. I think a lot of people don't realize functionality like "remember I want dark mode every time I visit" or "keep me logged in when I reopen my browser tomorrow" constitutes first-party tracking and requires consent under EU law.
Sorry, but no. Those functionalities fall under "functional cookies" and as such do not require consent. Also, there is no tracking needed for the dark mode at all. And "logging in" does not mean "tracking"
Strictly necessary cookies, session tokens and such, are exempted. But there’s no general exemption for cookies that provide functionality a user might like. If your site will function without remembering who I am when I come back tomorrow, you have to disclose that you’re going to try to remember me and give me a chance to say I don’t want you to. Doesn’t matter how benign your plans for that information are - the whole point is that the user is in control and they get to make that decision.
At my first job I took phone calls for an insurance carrier and agents definitely didn't like finding out that all the unhandled exception screens the rater had simply disappeared into the abyss.
You download a specific tool which only has the purpose of collecting your local error reports and sending them to Microsoft". Later on that tool became just a button in your control panel that submitted all your local errors and told you if those errors had an already developed solution.
That's how they did all their error telemetry until like late XP era, and it worked just fine.
All the people insisting that they need* this telemetry is also horse shit. Companies are demonstrably not producing better and more bug fixed software, and demonstrably are not using that data to make serious improvements, but demonstrably ARE using that data to choose where to focus dark pattern and other sales funnel based efforts.
If Unity and Unreal and GPU drivers can ask me "Do you want to send this error report" with a default no, nobody else has any excuse.
Even now, a significant amount of companies use the system of "Please upload your error log and the output of this command to this forum" as their bug report solution and it works just fine if that company actually intends to fix bugs.
The solution is not to turn your software into spyware. Stop being entitled. You don't have a right for me to QA your software for you, that's your job. Even with all this telemetry, companies only fix the most common and most obvious bugs anyway, so the perfect telemetry is utterly useless. Those bugs would have surfaced anyway.
Developers in the 80s did not need telemetry to get bug reports and fix things and release patches. Learn some history of your profession people.
Has throwing a hundred thousand bugs onto your sprint backlog actually helped anyone develop better software? No. Meanwhile it has exposed all your customers and users to predatory bullshit from your marketing and sales departments, and enabled your worst product managers to optimize hostility and extraction.
It's already seen as a valid opt-out signal against this sort of thing in Germany. LinkedIn got in trouble and lost a court case for not respecting the DNT header if memory serves me right.
I tried it as a little preview window for writing my blog, which is (in my opinion) very basic HTML and CSS. Whole page rendered wrong, though I admit I didn't bother to find out why. Give it a shot, but keep your expectations low.
If you have a basic site that doesn't work you can open an issue on the repo. If you have some relatively simple site, its useful for the team to know what features that people are using are broken.
> I'm curious how much trained in bias damages in-context performance.
I think there's an example right in front of our faces: look at how terribly SOTA LLMs perform on underrepresented languages and frameworks. I have an old side project written in pre-SvelteKit Svelte. I needed to do a dumb little update, so I told Claude to do it. It wrote its code in React, despite all the surrounding code being Svelte. There's a tangible bias towards things with larger sample sizes in the training corpus. It stands to reason those biases could appear in more subtle ways, too.
> What was the barrier to creating these documents before?
In a proprietary system, there is pressure against creating quality technical documentation because it can be used to train your replacement. Writing docs solely for your own benefit, or your colleagues' benefit, is also dubious because you already know the things you wrote. Although returning to a thing you made months/years ago can be painful, it's not the day-to-day common case in enterprise software development.
AI assistants flip the incentives. Now, your docs are helping to steer your personal digital goblin in the right direction. The docs are tangibly augmenting your own abilities.
I'd like to know this. I really like Zed, but their paid AI thing is terrible at inline suggestions, at least for my use cases. At least Copilot's C was coherent, even if it was wrong all the time.
It's still a terrific car. Comfortable, well made, fast enough for all practical situations. Unusally low mileage for its age. An engine that's sought after in the tuner community. But, it's unsellable. I'm stuck with it, whether I like it or not.
The good news is that I like it. The funny news is that I took a new job that will move me to the Bay, and whatever my new employer is paying to move my car out there is definitely more than my car is worth.
reply