If this is something their users actually wanted, they'd be paying xAI, not the other way around.
> There's a culture of indifference, an embrace of mediocrity. I don't think it's new, but I do think perhaps AI has given the lazy and prideless an even lower energy route to... I'm not sure. What is the goal?
I think pride in work has declined a lot (at least in the US) because so many large employers have shown that they aren't even willing to pretend to care about their employees. It's difficult to take pride in work done for an employee that you aren't proud of, or actively dislike.
[ty developer here]
We are happy with the attention that ty is starting to receive, but it's important to call out that both ty and pyrefly are still incomplete! (OP mentions this, but it's worth emphasizing again here.)
There are definitely examples cropping up that hit features that are not yet implemented. So when you encounter something where you think what we're doing is daft, please recognize that we might have just not gotten around to that yet. Python is a big language!
> For agency work building disposable marketing sites
Funny, because I did some freelancing work fixing disposable vibe-coded landing pages recently. And if there's one thing we can count on is that the biggest control-freaks will always have that one extra stupid requirement that completely befuddles the AI and pushes it into making an even bigger mess, and then I'll have to come fix it.
It doesn't matter how smart the AI becomes, the problems we face with software are rarely technical. The problem is always the people creating accidental complexity and pushing it to the next person as if it was "essential".
The biggest asset of a developer is saying "no" to people. Perhaps AIs will learn that, but with competing AIs I'm pretty sure we'll always get one or the other to say yes, just like we have with people.
I teach math at a large university (30,000 students) and have also gone “back to the earth”, to pen-and-paper, proctored and exams.
Students don’t seem to mind this reversion. The administration, however, doesn’t like this trend. They want all evaluation to be remote-friendly, so that the same course with the same evaluations can be given to students learning in person or enrolled online. Online enrollment is a huge cash cow, and fattening it up is a very high priority. In-person, pen-and-paper assessment threatens their revenue growth model. Anyways, if we have seven sections of Calculus I, and one of these sections is offered online/remote, then none of the seven are allowed any in person assessment. For “fairness”. Seriously.
Without a doubt the most impressive thing I've seen with CSS.
This immediately brought "A Single Div"[0] to mind, which stood as the coolest CSS demo I'd seen for... 11 years!
This one takes the cake. I'll be pouring over it. Thanks!
I don't believe the document does a great job in explaining what is otherwise a very simple idea (assuming I understood it well):
1. It creates a bitmap where each bit is a pixel in the image, if from frame 0 to frame 1 a given pixel changed, the corresponding bit is 1, otherwise it is 0.
2. All the 1s are added to the bloom filter, hashing their offsets. Now the bloom filter will be positive for all such indexes plus a percentage of false positive indexes.
3. We query the bloom filter to see all the indexes that are positive, and for all such pixels we store the raw pixel data of what changed. So we can reconstruct the next frame easily.
You can think at this like as storing the delta between two frames as: x,y,r,g,b of all the pixels that changed, but compressing a lot the x,y part at the cost of storing a bit more r,g,b than needed.
I have the feeling that since the pixels that changes from frame 0 to frame 1 are often similar (in their location) to what will change from frame 1 to frame 2, there is the possibility of further compressing that as well, by setting the right flags in the next frame and storing verbatim the only offsets that changed in addition to the previous or alike.
I guess I don't really get the attack. The idea seems to be that if you give your Claude an access token, despite what you tell it that it's for, Claude can be convinced to use it for anything that it's authorized for.
I think that's probably something anybody using these tools should always think. When you give a credential to an LLM, consider that it can do up to whatever that credential is allowed to do, especially if you auto-allow the LLM to make tool use calls!
But GitHub has fine-grained access tokens, so you can generate one scoped to just the repo that you're working with, and which can only access the resources it needs to. So if you use a credential like that, then the LLM can only be tricked so far. This attack wouldn't work in that case. The attack relies on the LLM having global access to your GitHub account, which is a dangerous credential to generate anyway, let alone give to Claude!
My wife quit Duolingo the week before this announcement after years of watching Duolingo prioritize attention manipulation over learning. She had a nearly 6-year streak and was on the paid version at the time, but realized that it wasn't actually helping her learn any more: she'd at some point begun maintaining a streak just for the sake of maintaining a streak.
The best documentation for Duolingo's decline is this article from a few years ago [0]. It's a piece by Duolingo's CPO (who was a former Zynga employee) where he discusses at length how Duolingo started using streaks and other gamification techniques to optimize their numbers. He has a lot to say about manipulating users into spending more time with them, but in the entire piece he barely even gives a token nod to the supposed mission of the company to help people learn. The date he cites for the beginning of their efforts to optimize numbers pretty closely correlates to my sense for when my wife began to complain about Duolingo feeling more and more manipulative and less and less useful.
This past month they finally jumped the shark and she decided to quit after 6+ years. The subsequent announcement that they'd be using AI to churn out even more lackluster content gave us a good laugh but was hardly surprising: they'd given up on prioritizing learning a long while ago.
[0] https://news.ycombinator.com/item?id=34977435
Finally a Lisp system wins the worse-is-better crown!
Good for Germany, but it is all too rare to see bad corporate behaviour punished like this. Steal £10k from a company, and you will probably go to prison for a long time. Start a company and steal billions from your customers and/or the tax payer, and you will probably get away with it. I believe Iceland was the only country to jail bankers after the 2008 banking disaster. We are still waiting for the British government to bring any individuals to account for wide scale corruption and profiteering during COVID.
My fav little-known trick is to test various syscalls fail with strace fault injection, like:
$ strace -e trace=clone -e fault=clone:error=EAGAIN
There are two kinds of engineers.
Those who can’t stop raving about how much of a superpower LLMs are for coding, how it’s made them 100x more productive, and is unlocking things they could’ve never done before.
And those who, like you, find it to be an extremely finicky process that requires extreme amount of coddling to get average results at best.
The only thing I don’t understand is why people from the former group aren’t all utterly dominating the market and obliterating their competitors with their revolutionary products and blazing fast iteration speed.
This is what happens when Google isn't sue-able by private entities.
In Germany, lieferando (subsidiary of takeaway.com) registers domains in the form of restaurantname-city.de, points them to their lieferando cloudflare account, and claims ownership for the google business entry where they set the phone number to their own call center.
Then they call the business owner and _force them_ to sign the contract with them, because effectively the owner knows they cannot be found anymore via google, and everyone that wants to order something will reach the call center hotline and leave a negative review after the hotline tells them wrong number, effectively destroying their business. And the people working for lieferando via Zeitarbeitsfirmen know this, and mention this in the call to pressure the restaurant owners to get their sales provision.
Crimeflare before it got taken down had around 130k domains that were pointing to the lieferando website using this kind of scheme, I helped provide the dataset for a couple of local business owners that were extorted this way and refused to abide by that scheme.
Guess what happened, nobody could be sued and the financial damages were too small to escalate it on the European court level. Sadly, class-action lawsuits don't work the same way as in the US, apparently.
Effectively Google does not abide by the laws and gets away with it due to their financial structures of their holding companies.
And they certainly know about this, they just don't give a single fvck.
Also, you know who did measure every angle to make sure it was correct? The engineers who put together the initial design. They sure as hell took their time getting every detail of the design right before it ever made it to the assembly line.
Who's filling that role in this brave new world?
> [Harper Reed] cautioned against being overly precious about the value of deeply understanding one’s code, which is no longer necessary to ensure that it works.
That just strikes me as an odd thing to say. I’m convinced that this is the dividing line between today’s software engineers and tomorrow’s AI engineers (in whatever form that takes - prompt, vibe, etc.) Reed’s statement feels very much like a justification of “if it compiles, ship it!”
> “It would be crazy if in an auto factory people were measuring to make sure every angle is correct,” he said, since machines now do the work. “It’s not as important as when it was group of ten people pounding out the metal.”
Except that the machines doing that work aren’t regularly hallucinating angles, spurious welding joints, etc.
To me it feels like part of the hype train, like crypto & VR.
I recently had the (dis)pleasure of fixing a bug in a codebase that was vibe coded.
It ends up being a collection of disorganized business problems converted into code, without any kind of structure.
Refinements are implemented as super-narrow patches, resulting in complex and unorganized code, whereas a human developer might take a step back to try and extract more common patterns.
And once you reach the limit of the context window you're essentially suck, as the LLM can no longer keep track of its patches.
English (or all spoken human language) is not precise enough to articulate what you want your code to do, and more importantly, a lot of time and experience precedes code that a senior developer writes.
If you want to have this senior developer 'vibe' code, then you'll need to have a way to be more precise in your prompts, and be able to articulate all learnings from your past mistakes and experience.
And that is incredibly heavy. Remember, this is opposite from answering 'why did you write it like this'. This is an endless list of items that say 'don't do this, but this, in this highly specific context'.
Lottie for me is sadness.
I love the idea, it's really cool that you can generate the animations from what animators already use, but boy, the implementation of it is very disappointing.
The format is probably one of the worst choices they could do for a use case like this - it's JSON, for something that is usually a bunch of numbers and perfect fit for more compact binary format. This JSON can reference external files, so the animation is either
- a folder with bunch of files (sub pictures)\
- or those files are inlined in the JSON as base64
- or it's just a single file, which turns out to be a zipped folder of this amalgamation.
If you imagine loading this on the web, you have to load absolutely enormous SDK (which is not very actively maintained and isn't very well size optimized), and then loading the animation either means loading a bunch of files separately, or loading a single file but processing it through multiple different parsers in multiple passes (JSON, base64, png, lottie, zip). If you use the .lottie file, you have to include zip decompresser in the JS bundle (.lottie player, which is a different library, also uses 2MB wasm blob, not sure why).
It took me a while to squash the footprint of this craziness in our app and I'm glad we don't use it in a hot path, because this is just crazy - it's supposed to be this little cherry on top for special occasions, but it's by far the heaviest part of the codebase. I had to manually tweak tose animations, run them through some optimizers, fixup weird path and inlining issues, fixed issue with those exporters turning vectors to png, all sorts of stuff.
On top of that, the browser doesn't survive playing more than a few of them reliably at the same time (especially on lower end devices), because turns out (who would have guessed?) - animating stuff with JS and DOM is not quite performant.
I kinda want to try a weekend project to turn these into optimized svg sprites and try to play them with a CSS transision, see if this makes it more bearable.
"The average temperatures on the Victoria line have risen by almost seven degrees since 2013 – nearly a *30%* increase.
Conversely, the increase in the average annual temperatures across all Underground lines from 2013 to 2024 was merely *seven percent*, placing Victoria’s temperature rise vastly above that."
Using percentages to talk about changes in non-Kelvin temperatures is crazy.
The article kinda dances around this point, but I think the largest reason "old games never die" is simply the old games mentioned were the good ones of their generation.
Similar to the lindy effect[0] where shows that had been around a while were likely to stay around a while longer. The are the games good enough for people to host fan servers and make mods, and behind each good game there is a lot of forgotten stuff that didn't inspire anyone to preserve it.
[0] https://en.wikipedia.org/wiki/Lindy_effect#:~:text=The%20Lin...
A small thing, but I found the author's project-organization practices useful – creating individual .prompt files for system prompt, background information, and auxiliary instructions [1], and then running it through `llm`.
It reveals how good LLM use, like any other engineering tool, requires good engineering thinking – methodical, and oriented around thoughtful specifications that balance design constraints – for best results.
[1] https://github.com/SeanHeelan/o3_finds_cve-2025-37899
I've commented (probably too much) to argue with the harshest critics of this piece, but I am surprised to not have seen much this criticism which is my main one:
Supposing I've made peace with the main gist of this: Cut living expenses to a point where you can work ¼ or so of the time most of us spend working by living somewhere cheap and not being so materialistic.
The missing piece here is social connections. Family and friends. If I could take my in-laws and my 2 best friends and their families with me, I'd sign up to move to a rural place like this tomorrow. But it's impractical for nearly everyone in the whole country to make such a thing happen. This limits its appeal. This place is 90 minutes or so from the Montreal airport, which is actually not bad for rural places, but flights are not cheap, certainly not accessible on the budget described here, so for you to have contact with anyone outside this town, they're likely going to have to drop about $500 per person, per visit, and will be staying at the Super 8 since you probably don't have a guest room). So, implied but not acknowledged in this piece is the assumption that you are almost definitely going to only see your family and friends a few more times (maybe once a year each, if you're super lucky) for the rest of your life.
And unlike questions of money; food, entertainment, family and friends aren't fungible. You can start over and hope to make new friends out there, but you can't replace people. This is what would make this life untenable to me, and I'm not even all that extraverted.