‘… a decaying empire experiencing the stagnation of all of its core products that's using OpenAI's billions in inference spend to cover up the collapse of Azure's growth.’
I’d like to put that in the headline but to follow HN guidelines put the actual ‘haters’ headline instead. I’m not aligned with that terminology. I do think these are fascinating and scary numbers.
> Planetary roller screws are the gold standard for high-performance joints such as knees, ankles, and hips.
It's hard to understand how these are used for joints. I think of a screw as something that rotates many times. Are these used for things that rotate only a few degrees, as a knee might?
Yes, this is just the power for the joint, not the hinge.
How to do this in a tight space is a tough mechanical engineering problem. Tesla's Optimus uses a 4-bar linkage as the hinge, and some kind of cylindrical linear
actuator as the power drive. Can't tell much about the actuator from the patent for the hinge.
Boston Dynamics used to use hydraulic pistons in their legs, but that did not scale down well from their Big Dog mule-sized machine. They finally went electric, and their machines became far less clunky. Motor power/weight ratios have improved a lot since the early BD days.
Electrical linear motors would be a nice solution. They're rarely used, because they tend to have to be custom for each application. But we might see more of that as humanoid robots approach volume production. The technology has reached 15:1 power/weight ratio.[1] With cooling.
these translate rotary motion into linear motion. if you hold the screw fixed (with bearings), and let the nut float, then turning the screw moves the nut back and forth along the screw
I really like Railway, and have deployed many sites with them, but got worried by their recent funding round. At some point those investment bills are going to come due.
> If "just build it natively" were actually easier, faster, or cheaper at scale, everyone would do just that
Value prop of product quality aside, isn't the AI claim that it helps you be more productive? I would expect that OpenAI would run multiple frontends and that they'd use Codex to do it.
Ie are they using their own AI (I would assume it's semi-vibe-coded) to just get out a new product or using AI to create a new product using the productivity gains to let them produce higher quality?
On a side note, the company I work for (RemObjects, not speaking on their behalf) has a value ethos specifically about using the native UI layers, and encouraging our customers to do the same. (We make dev tools, a compiler supporting six languages (C#, Java, Go, etc) plus IDEs.)
Our IDE does this: common code / logic, then a native macOS layer and a WPF layer. Yes, it takes a little more work (less than you'd think!) but we think it is the right way to do it.
And what I hope is that AI will let people do the same -- lower the cost and effort to do things like this. If Electron was used because it was a cheap way to get cross-platform apps out, AI should now be the same layer, the same intermediate 'get stuff done' layer, but done better. And I don't think this prevents doing things faster because AI can work in parallel. Instead of one agent to update the frontend, you have two to update both frontends, you know?
We're building an AI agent, btw. Initially targeting Delphi, which is a third party's product we try to support and provide modern solutions for. We'll be adding support for our own toolchains too.
What I fear is that people will apply AI at the wrong level. That they'll produce the same things, but faster: not the same things, but better (and faster.)
It's about consistency - you want to build an app that looks and functions the same on all platforms as much as possible. Regardless of if you are hand-coding or vibe-coding 3 entirely separate software stacks, getting everything consistent is going to be a challenge and subtle inconsistencies will sneak in.
It comes back to fundamental programming guidelines like DRY (Don't Repeat Yourself) - if you have three separate implementations in different languages for everything, changes will be come harder and you will move slower. These golden guidelines still stand in a vibe-code world.
> This shift coincided with a decrease in adjective TTR below a defined threshold, occurring approximately ten years before Pratchett’s formal diagnosis.
The diagnosis was announced in 2007, meaning the shift occurred in 1997. 1997 was after Jingo and before Carpe Jugulum and The Last Continent, and 2007 was after Making Money and before Unseen Academicals.
The Last Continent is the first identified in the paper as below the cutoff for adjectives which they use to identify the start of the decline.
My own feeling is that many of his strongest works were before 2000, though he had several excellent ones after (the City Watch and first two Moist von Lipwig; I know the ongoing Tiffany Aching series are good, but in terms of writing I found them not as intricate as his earlier books.) I found Snuff harder to read, and Raising Steam, sadly, very difficult. I could tell the genius was there, but my memory of the writing was that it used much longer sentences, had less intricate plotting, and far fewer puns and wordplay. It was this book that made me really feel a sense of grief for what was happening to him, and it was this one where I first felt there was an observable threshold that was crossed.
I have sometimes wondered if it would be respectful if another author was brought into assist in editing or rewriting his last two novels. I know his unpublished works were destroyed, and any writing assistance is not his own voice. Yet I feel, in a sense, seeing books with such clear decline could in itself have let his legacy down. I don't know what his own view was or would be. While I admire Sanderson' continuance of the Wheel of Time, I would not wish such a drastic change in tone for some similar effort for the last of Pratchett's works. Yet I deeply wish that his last books were, somehow, different, more representative of him that I feel they were, in that his illness (in a sense, of course!) let him down. They cause me sadness.
GNU Terry Pratchett. (My own site sends this too.)
Personally I just ignore them(Raising Steam and Snuff). There are too many inconsistencies with how Vimes and Lipwig are characterized in those books, that I can't see them as the same characters. I haven't started the Shepherd's Crown for the same reason.
I don't think that any fan of his is under any illusion that those books are up to his standard, but he has so many good books that his legacy will be safe.
my two favourite works of his are "night watch" and "going postal". it's honestly super impressive that they were written so long after he started to decline and were still so good.
I'm currently in the middle of a complete chronological reread of the discworld books, just finished "thud" and "wintersmith" back to back and while they were definitely weak in places it's amazing how much of his genius still shone through. feeling a little apprehensive about the later books though, I remember some of them being really bad, especially "snuff" :(
Night Watch is definitely as good as the peak 90s City Watch books. Thud had bits of standout quality too. Some of the weaknesses of later books might be less pure weaknesses and more people distinctly remembering the same characters encountering similar characters with a similar message before.
> My own feeling is that many of his strongest works were before 2000
Interestingly he hired Rob Wilkins in the year 2000. So, did Rob's presence affect the books? Or did Terry hire Rob because he subconsciously knew he needed to compensate for a decline that started to become apparent, by offloading some tasks?
While I have the natural questions like 'why not Python?':
> 'Thadus' — a digital learning tool that teaches coding to beginners.
> ...purposely built to run offline in areas with patchy internet connection...
> ...split into three courses aimed at giving users a basic understanding of coding concepts and how they relate to real world industries.
does sound intriguing and well targeted. The article is _very_ light though on what this actually means.
https://www.thaduscodelabs.com/Course-Covers/ (looks like their website) has material on general pgramming concepts it covers, but again no reason it has its own environment or language.
their website says
"This course ensures students never approach Python “cold.” They develop genuine computational literacy before advancing into real programming and higher-level algorithm design."
Which isn't very convincing, yeah.
Also its very expensive at $20 a month, for something that is supposed to help with equitable access for poorer countries....
Its probably just a throw-away high school project. By that i mean the creator probably doesn't seriously believe in it. It probably is a high effort project though.
I grew up in Sri Lanka and now live in the Cambridge, UK. I used to get great signal, at least 3G often 4G in most rural areas of the country when in was in Sri Lanka, whereas now I lose signal completely all the time around the town, and when I step into most grocery stores.
UK signal is so bad. Was getting 5G in remote regions of Ha Giang in northern Vietnam, lucky if I get reliable 4G in Brighton. Brighton has 5G spots that can work great, but suffer with overload, and much of the residential areas still have overloaded 4G. London is even worse but at least they can blame the buildings.
I’d like to put that in the headline but to follow HN guidelines put the actual ‘haters’ headline instead. I’m not aligned with that terminology. I do think these are fascinating and scary numbers.
reply