A lot of writing (maybe most) is almost the same. Code is a means of translating a process into semantics a computer understands. Most non-fiction writing is a means of translating information or an idea into semantics that allow other people to understand that information or idea.
I don’t think either is inherently bad because it’s AI, but it can definitely be bad if the AI is less good at encoding those ideas into their respective formats.
There’s a difference between “this concept has value” and “a company can capture that value”.
I do see value in this, but like you I think it’s too trivial to implement to capture the value unless they can get some kind of lead on a model that can consume these artifacts more effectively. It feels like something Anthropic will have in Claude Code in a month.
I think it’s a) volume of scrapers, and b) desire for _all_ content instead of particular content, and c) the scrapers are new and don’t have the decades of patches Googlebot et al do.
5 years ago there were few people with an active interest in scraping ForgeJo instances and personal blogs. Now there are a bajillion companies and individuals getting data to train a model or throw in RAG or whatever.
Having a better scraper means more data, which means a better model (handwavily) so it’s a competitive advantage. And writing a good, well-behaved distributed scraper is non-trivial.
It reflects even worse on Google for vacuuming up and keeping the data.
They can’t really refuse to hand over the data, but they could purge and stop collecting identifying data on Americans. As is, they are tacitly complicit by collecting data they know will be used against protesters.
Their business model is ads, which doesn’t inherently require storing information the government would want. Some of that data probably isn’t useful in that space, others might reduce efficacy by a tiny margin but it wouldn’t shut them down.
Eg I can’t think of a reason why they’d need to store your exact location. Do people target ads down to a precise GPS location, or even a street? I can’t imagine they need things more granular than a ZIP code.
That also doesn’t absolve them of supporting autocracy. The difference between the morally upright and the morally bankrupt is what they do when doing the right thing will cost them, not what they do when the right thing is free.
This makes sense to me. Part of me wonders if this system wouldn't work better in reverse, a blocklist instead of a banlist. Blocklists can spread via URL, in the same way that DNS or email blocklists work. Subscribe to the blocklists of people you trust.
I _think_ this removes the motivation for low-quality PRs. Get on a major blocklist and the GitHub account is basically dead. People could make new GitHub accounts, but then you never get an "impressive" GitHub account.
I was talking to a coworker that really likes AI tooling and it came up that they feel stronger reading unfamiliar code than writing code.
I wonder how much it comes down to that divide. I also wonder how true that is, or if they’re just more trusting that the function does what its name implies the way they think it should.
I suspect you, like me, feel more comfortable with code we’ve written than having to review totally foreign code. The rate limit is in the high level design, not in how fast I can throw code at a file.
It might be a difference in cognition, or maybe we just have a greater need to know precisely how something works instead of accepting a hand wavey “it appears to work, which is good enough”.
There are absolutely things wrong with that, because React was designed to solve problems that don't exist in a TUI.
React fixes issues with the DOM being too slow to fully re-render the entire webpage every time a piece of state changes. That doesn't apply in a TUI, you can re-render TUIs faster than the monitor can refresh. There's no need to selectively re-render parts of the UI, you can just re-render the entire thing every time something changes without even stressing out the CPU.
It brings in a bunch of complexity that doesn't solve any real issues beyond the devs being more familiar with React than a TUI library.
It’s fine in the sense that it works, it’s just a really bad look for a company building a tool that’s supposed to write good code because it balloons the resources consumed up to an absurd level.
300MB of RAM for a CLI app that reads files and makes HTTP calls is crazy. A new emacs GUI instance is like 70MB and that’s for an entire text editor with a GUI.
I would still call that small, maybe medium. emacs is huge as far as CLI tools go, awk is large because it implements its own language (apparently capable of writing Doom in). `top` probably has a similar number of interaction points, something like `lftp` might have more between local and remote state.
The complex and magic parts are around finding contextual things to include, and I'd be curious how many are that vs "forgot to call clear() in the TUI framework before redirecting to another page".
I don’t even think that’s rational, but it may be what’s propping them up.
Last earnings call Musk said Optimus wasn’t doing “meaningful work” at Tesla and as far as I’m aware they haven’t done meaningful work anywhere. I think they’re behind the curve there. Figure AI recently finished an apparently successful feasibility trial of their humanoid robots with BMW and Boston Dynamics has a deal with Hyundai for their Atlas humanoid robots.
I’m not even convinced humanoid robots are going to pan out in general. They only really make sense in a scenario where you’re back porting robotics to factories built for humans. That has value but feels temporary; factories designed to be robotic feel like the future, and there’s no need for them to do the job the same way a human would.
Our world is adapted for humans, so humanoid robots will fit in most places. They might not be the best choice, but the universality has a good chance of making it worth it through economies of scale.
Building a custom robot that can stock shelves at a supermarket won't be worth it for a long time, but programming an existing humanoid platform might work. Find a couple hundred tasks like this (including household use), and that platform now has huge economies of scale.
Now, when you're starting a small factory, using the existing humanoids might make more sense than getting custom tooling, at least for some tasks. You'll often see factories where some tasks that could, in theory, be automated are left to humans because they're relatively small tasks and not worth automating with a custom machine. Humanoids could fill that gap.
> Building a custom robot that can stock shelves at a supermarket won't be worth it for a long time, but programming an existing humanoid platform might work.
This feels inverted to me, but perhaps I’m reading it wrong. A lot of the core challenges are shared, but the humanoid has to solve a bunch of additional challenges. Eg balancing is difficult with moving loads of various weights. Humanoids have to deal with that, while something more forklift-like practically opts out of that issue by just being designed with a high mass and low center of gravity.
I don’t see a universe where a humanoid is ever cheaper, but I could maybe see it generalizing well enough for usage to make it worth it. I’d still be a bit surprised, because operating costs would surely be higher (way more servos or hydraulics to fail, higher power usage hauling around unnecessary parts and weight).
This seems doubly true for factories where opex is so much more meaningful than capex. It’s worth spending $4M on custom tooling rather than $2M on generic tooling if it drops your opex by $500k/year on a factory with a 20 year lifespan.
>I’m not even convinced humanoid robots are going to pan out in general.
I want one personally, so it can rake the leaves, mow the lawn, tend the garden, do the laundry and dishes, replace the roof, etc., when I'm old. But they should also be used to pick up litter along the highway, paint over graffiti, etc..
I absolutely do too, I’m just not convinced a single humanoid robot is going to do the job cheaper and better than a dozen purpose-built robots (which you might own, or might rent from Home Depot or whatever when the need arises).
Eg lawn mowing robots already exist, and have for a decade or so. Garden tending also exists, though I think only commercial prototypes at the current moment. Roofing feels very possible, but I only roofed once so ymmv.
Is the future going to be buying a humanoid robot with a thousand servos for $100,000, or texting a number to have a self-driving car drop off a bladed roomba made from bargain bin brushless motors and plastic to mow your lawn for $0.50?
I feel like the humanoid form is getting in the way for that, and that a "Spot" like design with a hand on top is better suited for that. Also i think laundry and dishes are already 95% automated since about 50 years.
I _think_ these are meant to replace humans working alongside the industrial robots rather than the big industrial robots themselves. I don’t work in manufacturing though, and the press releases are too buzzword-y for me to grasp the actual tasks they’re going to do.
I would guess the long term strategy is to do this for economies of scale and then push into new markets opened up by the lower price point. I would guess these are horribly expensive right now, given something like Spot is way simpler and still like $40k
this is something that also never made sense to me - it felt like star wars got it right - for repairs and remedial tasks a trash can (rs-d2) or all the little service droids are more appropriate, but c3p0 or other nurse and protocol droids makes sense to look more humanistic since they serve functions to facilitate human activitiy - but there is no way those functions are numerous enough to be priofitable.
I don’t think either is inherently bad because it’s AI, but it can definitely be bad if the AI is less good at encoding those ideas into their respective formats.
reply