The software is certainly easier to build, but there's a lot of hardware involved here beyond the tractor. Claude is not necessarily going to make it easier to do soil sampling or measuring field conditions or yield outputs.
Farmers would be foolish to rely on an LLM because farming margins are too low to makeup for even a small quick mistake. Many farms will profit 1% on investment over 1-2 decades, although year to year yield can vary 30%.
Same, mostly. However today in particular Claude can't do front-end development with any competence. Never seen this before. I think the rumor is they are rolling out a new model and have to divide their infra across the new model vs the current model.
I don't know why you'd believe they've ever been capable of putting out quality software.
A bug in the software is a bug in the process, and the process is the job of leadership. They've never cared about software quality. They'll put out lots of books about it, lots of talks, lots of claims. But they won't actually put out quality software. It's not in their DNA, never was.
It's not their size nor their age that makes this hard for them. Plenty of larger, older companies put out better product ever day. It's just them. Someone in each size class is the best, and someone else is the worst. MS has been the worst the entire time.
I had to restore my son's desktop PC last night from a USB stick. It didn't even have drivers for graphics over 800x600 or the wifi card. I was flabbergasted. It's windows 11, 2026, a 6-month old PC. I genuinely don't know how someone could sell something this awful with a straight face.
Windows was only ever better than DOS, by the same vendor. It's been awful compared to any competitor it's ever had. Really. I don't see a non-gaslighting argument for Windows anywhere.
Faster than OS/2, sure. Now try to download a file in the background while doing work in the foreground. You would be lucky if your Windows 3.1 communications application could complete it without multiple retries.
The two operating systems were trying to solve different problems, and had different system requirements because of that. Windows 3.1 was fine for running multiple interactive applications since neither application would be doing real work in the background. When Windows 95 entered the picture, that changed and its system requirements weren't all that different from OS/2.
And that is just one example. Windows 3.1 didn't provide much in the way of memory protection. (From my recollections, it could detect a memory access violation. At that point it would blue screen.) One of OS/2's most noteworthy features was memory protection. All of a sudden you could use your computer for an entire day without losing work from crashing. Yeah, OS/2 would happily terminate an application (rather than the OS) when there was a memory access violation. On the other hand, it made it much easier for developers to detect and address such bugs.
On that last point: I have fond memories of bringing OS/2 boot disks to my high school programming classes after the upgrade to Windows 95 (and, when they started refusing to let me boot OS/2, they let me use the NT server). There was a world of difference between programming under OS/2 or NT verses Windows 95. No one bothered to try programming under Windows 3.1!
> No one bothered to try programming under Windows 3.1!
VB and Windows 3.11 paid for my first home. I wouldn’t enjoy programming in C in Windows though, and, IIRC, it was a while before Microsoft’s C tooling got a Windows version.
I should have said: no one in my programming class bothered to program under Windows 3.1. Clearly there were people out there writing software for Windows 3.1, and I have heard that VisualBASIC was a nice development platform for Windows 3.1.
I'm not sure what the C situation was like for Windows 3.1. I did have Borland C++ and fiddled around with the Windows IDE a bit, but never recalled making any progress.
3.1 was still nice compared to its main competitor, which was MacOS 7. Only the richest kids would be running things like Solaris, SCO, and other preemptive multitasking systems because memory demands were high and memory was very expensive.
Also, Windows 3 would run on 286 computers (as would OS/2), which made the barrier of entry very low. I started running it on a 286 with a Hercules adapter.
Yeah, I mean I'm looking at frameworks/thinkpads on one side and chromebooks on the other.
Not charging up to $440 (!) for a keyboard isn't a great act of engineering or generosity. This has been ridiculous for a very, very long time. Being less ridiculous isn't worth celebrating. The goal markers have moved so damned much.
Compare to a thinkpad keyboard FRU. They have fluid drains and still cost $99 for a top-end laptop. My daughter's chromebook keyboard replacement at school was $16.
> This has been ridiculous for a very, very long time. Being less ridiculous isn't worth celebrating.
So what I'm hearing is you don't want Apple to make their computers more repairable? Think of this like training a dog. My dog can open the cabinet in the kitchen on their own, pull out a specific requested item, close the door again and bring the item to me from anywhere in my house. Opening a door is just tugging on something, bringing something to me is just fetch, closing a door is just pushing with its nose. If I went into the training of this with the attitude of "oh wow, you pulled the door open" or "oh wow, you fetched the thing" and didn't reward my dog for doing those simple pieces because "any good dog can tug on a rope or fetch a ball", then my dog would never have gotten to the point of doing all of those things in a repeatable complex sequence that serves a useful purpose. Instead every part of it that my dog got right, they got all sorts of praise and rewards. And so once I started asking more, my dog eagerly tried to do those things because they knew if they did what I wanted, they could get the things they wanted.
Train your companies the same way. Give them the positive PR and praise they're looking for when they do the things you want them to do. You'll get them to do what you want a lot faster if they have an actual incentive to do it.
I think we'll have to start configuring our client tools (e.g. browser, email client, etc) to render domain names with annotations for different character classes. E.g. our native character set is a standard color (blue/black) and then other character sets would have to stand out (purple background?).
i'm pretty sure Mox (email server with included webui written in Go) does that - at least the Umlauts in mails i get from Hetzner seem to always stand out.
it also defaults to not loading HTML in emails, which i love. really opened my eyes to how dumb it really is to just accept all kinds of dynamic content in unknown messages. (kinda same as how the modern web relies on remote code execution to work)
The website is pure slop with an infinitely low information to noise ratio. Judging by their documentation[0] it's just access control for web servers, so the better question would be how it compares to TLS client auth (which uses proper standards, doesn't require users to install extra software, and doesn't require any extra daemons server-side)
If a robot can do basic cleaning, laundry, and dishes, that's worth a lot to a lot of people. Dual-professional households have the money, and not having to do this housework could save some marriages.
I don't think it actually is worth a lot to people. I know dual-professional households who don't even use their dishwasher consistently, and multiple companies have gone bankrupt trying to bring automated laundry folding (which does exist in industry) to the consumer market.
Maid services are generally expected to handle "everything" for a pretty expansive definition of everything. They pick up scattered stuff and put in a sensible location, they arrange everything visible in an aesthetically pleasing way, they take out the trash, if there's some weird dirt that's hard to clean they creatively problem solve to find a way to get it off. I don't think there's a market for a service that can only handle basic cleaning.
(Will someone eventually invent a machine that can do all of that and more? Yes, probably, and they'll make billions when they do. But Tesla has offered no reason to believe this is on their horizon, and the focus on a humanoid form factor strongly suggests that they're optimizing for media appeal over practical capabilities.)
Maids are paid a VERY low wage in exchange for being able to take on an almost unlimited list of general tasks, from folding laundry to managing kids to mopping stairs. We are decades away from robots with that capability, and they are intended to replace people who are often not making even minimum wage? Please. Get real.
Robot vacuum with a mop, washing machine, tumble dryer and dishwasher reduce housework to like an hour per week, ie 30 min/person/week. This can be higher if you live in a big house, but if your marriage can’t tolerate 30 mins of house work a robot will not solve it.
> Dual-professional households have the money, and not having to do this housework could save some marriages.
Dual-professional households could hire a maid and pay for marriage counseling and still save money compared to a $20k robot plus whatever a subscription would run.
I can google "maid service seattle" and see dozens of entries. The first one in the yelp list is available to book and will clean a 1000 - 1500 sq ft, 2 bed, 2 bath house for well under $200. There's even a decent discount if you book is as a weekly or biweekly service.
That feels pretty affordable? I know it's a scale, but minimum wage here is $21/hr now.
I have enough time to take care of my own space, but for comparison Comcast internet is well over $120/month for crappy speeds. I think in comparison a little more than that for 1 deep cleaning a month is reasonable.
Doesn't matter if it was or wasn't, it was a failure that GM never followed up with. Why it was a failure is also irrelevant, because whether you feel it was a technical failure or killed by GM, GM never did anything with the project or knowledge. Effectively it was a curiosity.
If GM killed it to keep it from succeeding, then there is massive precedent to never reuse the tech. In fact, their NiMH battery patents were sold to Texaco/Chevron who held them close and never let anyone use them. From that point, they couldn't follow-up without dumping even more cash into it, effectively burying it. Until new lithium battery tech matured, there was no way to do it again.
Not only were electric cars available since the very beginning of cars, but they've always been available as niche options. There are tens of electric cars that postdate the EV1 and predate the Tesla. Do you even know their names?
We have stupidly cheap gas. An electric car has only ever been a curiosity for America. Even now, the primary driver of people buying electric cars is ideological, and a mild convenience of never having to go to a gas station.
Pre-lithium battery electric cars are a huge hassle, for very little gain, even outside the US. The history of cars is a global one, and no amount of conspiracy theory about GM can counter the fact that nobody else made electric cars either, even in places with drastically more expensive and unreliable gasoline.
They have always been a novelty, like hydrogen and LPG and compressed gas engines.
Hybrids were the closest anyone got to making older battery chemistries meaningful for car-style transportation, and even that was extremely limited.
Cheap gas, car culture and the incredibly long distances makes America a very different place from the urban centres of the Netherlands, China and Korea.
GM didn't sell EVs for years after releasing the EV1. They didn't get any market advantage from the EV1 because they left the market after, for a long time.
It is very widely known that GM held a 7 year head start on every other automaker on manufacturing the modern EV. Several other EVs were sold during it's time in low volume.
