(Article is from 2023, so the title should be updated to say "32 years ago", or something)
The biggest loss in TUIs is the latest wave of asynchronous frameworks, which bring the joy of dropped keypresses to the terminal.
In any TUI released before the year 2000, if you press a key when the system wasn't ready, the key would just wait until the system was ready. Many TUIs today still do this, but increasingly frequently (with the modern "web-inspired" TUI frameworks), the system will be ready to take your keypress, and discard it because the async dialog box hasn't registered its event listener yet.
Other than that antipattern, TUIs are doing great these days. As for terminal IDEs, Neovim has never been more featureful, with LSPs and other plugins giving all the features this article discusses. I guess it isn't a mouse-driven TUI, so the author wouldn't be interested, but still.
> taking inspiration from command-line interfaces
IMO the take away from command-line interfaces is compact, precise and minimal design. In a transitional shell prompt like #~$, each character has its meaning. Merely copying these symbols to a watch face is the exact opposite spirit of command like interfaces.
Won't be available to anyone terminated for copyright stuff.
Woooow what a huge dick move.
That's the one massively imbalanced power dynamic that I hear people really fear losing their livelihood to for no good reason, and they're leaving it there to terrorize and ruin livelihoods for future generations.
I put up a video of a funeral service for my grandma and like 5 minutes later I was getting threatening legalese mail about my channel by cancelled forever because some record label has a recording of a thousand year old hymn and they don't give a shit about threatening people with no legal basis at all
We're doing something like this internally. Our monorepo context files were much too big, so we built a progressive tree of fragments to load up for different tasks.
I am struck by how much these kinds of context documents resemble normal developer documentation, but actually useful and task-oriented. What was the barrier to creating these documents before?
Three theories on why this is so different:
1) The feedback loop was too long. If you wrote some docs, you might never learn if they were any good. If you did, it might be years later. And if you changed them, doing an A/B test was impractical. Now, you can write up a context markdown, ask Claude to do something, and iterate in minutes.
2) The tools can help build them. Building good docs was always hard. Especially if you take the time to include examples, urls, etc. that make the documentation truly useful. These tools reduce this cost.
3) Many programmers are egotists. Documentation that helps other people doesn't generate internal motivation. But documentation that allows you to better harness a computer minion to your will is attractive.
Any other theories?
Not quite.
Ofcom in their reply make their point clear: "The [Online Safety] Act explicitly grants Ofcom the legal authority to regulate online safety for individuals in the United Kingdom [...]"
They are stating that companies operating in the UK and providing services to UK individuals, are required to conform with UK regulations in relation to those services, under UK law.
As an American business, you can choose to ignore that, but that has consequences if any of your board of directors ever sets foot in the UK.
The US does this, and US lawyers understand this. If I open an online poker and sports bookmaking site in the UK (where such sites are completely legal), and take business from all over the United States thereby breaking federal law, I can expect to be met at the plane door the next time I take a shopping trip to NYC. Arguing that my servers and my business are located in the UK is not going to impress the federal judge I'd appear in front of in the morning. Stating the US laws against my activities have a snowball's chance in hell of being enforced in the UK is surely going to risk me being charged with contempt.
The Online Safety Act is ridiculous on many levels, but in the same way that Google does certain things in relation to Tiananmen Square searches in China, and every tech company engages in regulatory alignment for the entire Middle East, the UK has asked that US companies do certain things in the jurisdiction of the UK. I'd argue, less harmful and egregious things in some respects.
Should the UK do this? No, probably not. I think it will just make VPN software vendors richer, and UK citizens - particularly children - barely any safer.
Are Ofcom claiming jurisdiction in the US? No, they're claiming jurisdiction in the UK. Which, I hasten to add they are legally required to do by the Online Safety Act, by the government they are an agency of. If they didn't, the government would literally be breaking its own law.
TIL that 4chan's lawyer is about as grown up, mature and able to engage in critical thinking about the law as the people who post on his client's site.
> Using UUIDv7 is generally discouraged for security when the primary key is exposed to end users in external-facing applications or APIs. The main issue is that UUIDv7 incorporates a 48-bit Unix timestamp as its most significant part, meaning the identifier itself leaks the record's creation time... Experts recommend using UUIDv7 only for internal keys and exposing a separate, truly random UUIDv4 as an external identifier.
So this basically defeats the entire performance improvement of UUIDv7. Because anything coming from the user will need to look up a UUIDv4, which means every new row needs to create an extra random UUIDv4 which gets inserted into a second B-tree index, which recreates the very performance problem UUIDv7 is supposedly solving.
In other words, you can only use UUIDv7 for rows that never need to be looked up by any data coming from the user. And maybe that exists sometimes for certain data in JOINs... but it seems like it might be more the exception than the rule, and you never know when an internal ID might need to become an external one in the future.
I like GOG a lot but it's wild to me that their GOG Galaxy client doesn't work on Linux! A lot of gamers who care about preservation and availability are spending money with Valve because Steam's DRM is mostly inoffensive and the Linux support is so good.
The addressable market segment of people who play PC games and also care about DRM-free accessibility would be larger if GOG's launcher ran on Linux and targeted Linux users. It seems like a logical overlap to me.
Valve is eating GOG's lunch in this segment but it could easily change. Sure it might be small but it's bigger than ever, still growing, and seems to fit GOG's mission.
I would definitely start repurchasing my Steam games DRM-free on GOG if only they provided a launcher with the tooling necessary to download & run them on my system.
As things stand now, and for all the good GOG does... it's not enough to be DRM-free but only distribute Windows installers. You've just outsourced the DRM scheme to Microsoft. If the software doesn't run on a DRM-free OS, the job is only halfway done.
And in the meantime, GOG's product is tragically subject to piracy, (I believe) partially enabled by their decision to _only_ package games for the OS upon which most piracy traditionally takes place! :( I hope this could be offset by packaging for a crowd with more ideological overlap.
His "$400B in next 12 months" claim treats OpenAI as paying construction costs upfront. But OpenAI is leasing capacity as operating expense - Oracle finances and builds the data centers [1]. This is like saying a tenant needs $5M cash because that's what the building cost to construct.
The Oracle deal structure: OpenAI pays ~$30B/year in rental fees starting fiscal 2027/2028 [2], ramping up over 5 years as capacity comes online. Not "$400B in 12 months."
The deals are structured as staged vendor financing: - NVIDIA "invests" $10B per gigawatt milestone, gets paid back through chip purchases [3] - AMD gives OpenAI warrants for 160M shares (~10% equity) that vest as chips deploy [4] - As one analyst noted: "Nvidia invests $100 billion in OpenAI, which then OpenAI turns back and gives it back to Nvidia" [3]
This is circular vendor financing where suppliers extend credit betting on OpenAI's growth. It's unusual and potentially fragile, but it's not "OpenAI needs $400B cash they don't have."
Zitron asks: "Does OpenAI have $400B in cash?"
The actual question: "Can OpenAI grow revenue from $13B to $60B+ to cover lease payments by 2028-2029?"
The first question is nonsensical given deal structure. The second is the actual bet everyone's making.
His core thesis - "OpenAI literally cannot afford these deals therefore fraud" - fails because he fundamentally misunderstands how the deals work. The real questions are about execution timelines and revenue growth projections, not about OpenAI needing hundreds of billions in cash right now.
There's probably a good critical piece to write about whether these vendor financing bets will pay off, but this isn't it.
[1] https://www.cnbc.com/2025/09/23/openai-first-data-center-in-...
[2] https://w.media/openai-to-rent-4-5-gw-of-data-center-power-f...
[3] https://www.cnbc.com/2025/09/22/nvidia-openai-data-center.ht...
[4] https://techcrunch.com/2025/10/06/amd-to-supply-6gw-of-compu...
The interview which I've watched recently with Rich Sutton left me with the impression that AGI is not just a matter of adding more 9s.
The interviewer had an idea that he took for granted: that to understand language you have to have a model of the world. LLMs seem to udnerstand language therefore they've trained a model of the world. Sutton rejected the premise immediately. He might be right in being skeptical here.
From the attorney's post:
> The infinite character of that power was most famously summed up by English lawyer Sir Ivor Jennings, who once said that “if Parliament enacts that smoking in the streets of Paris is an offence, then it is an offence”. This line is taught to every first-year English law student.
Initially this seems like disrespect for another country's sovereignty. But really the crucial thing is:
> We explained to the UK that the Online Safety Act had a snowball’s chance in hell of being enforced in the United States
Ofcom has to go through the motions of telling 4chan they can't smoke in Paris because of the (very on-brand) nanny law.
>What takes the long amount of time and the way to think about it is that it’s a march of nines. Every single nine is a constant amount of work. Every single nine is the same amount of work. When you get a demo and something works 90% of the time, that’s just the first nine. Then you need the second nine, a third nine, a fourth nine, a fifth nine. While I was at Tesla for five years or so, we went through maybe three nines or two nines. I don’t know what it is, but multiple nines of iteration. There are still more nines to go.
I think this is an important way of understanding AI progress. Capability improvements often look exponential on a particular fixed benchmark, but the difficulty of the next step up is also often exponential, and so you get net linear improvement with a wider perspective.
Conversions to Nickelback, Poundz, Los Pesos, DJ Euro and Yen.
hi, i made this. thank you for posting.
unfortunately due to the government shutdown, the BLS inflation data for September 2025 is delayed from October 15 (as it normally is) until October 24[1], so please check back then to see if he is >109 Cent.
assuming future stability, the site will automatically update on the 15th of every month.
[1] https://www.bls.gov/bls/092025-cpi-reschedule-notice.htm
Sub-Headline from this article: "Plummeting resale values are threatening to derail the world’s transition to electric transportation."
Alternative take: "EVs now easy to afford for the 80% of Americans who don't have $50-90k to spend on an EV!"
This year I bought a 2022 EV with 16k miles. A luxury brand. The sticker price when new was $79,000. I paid $35k. It was an off-lease vehicle so if anyone took a bath, it was the bank. I would never in a million years spend 80 grand on a car but now I have a great EV.
Battery life is not a huge concern. Any more than timing belts/chains, transmissions, etc. can be dauntingly costly repairs for cars with 150k miles or more.
I also have a gas car which I love (spouse drives the electric for a much greater commute) so I'm no EV absolutist. But this whole premise is stupid. EV adoption has had 2 main blockers: 1. only rich people had justification to buy them until recently, and 2. Charging space for people who don't have their own private garage.
Now #1 is no longer a factor. This is a GOOD thing.
More odious nannying by silly civil servants. If Britain is to restore cultural leadership it needs to move policy away from this horrible trend of policing what people say and think, and focus its energy on better policing what people do.
> I've tried 30X redirects (which it follows)
301 response to a selection of very large files hosted by companies you don't like.
When their AWS instances start downloading 70000 windows ISOs in parallel, they might notice.
Hard to do with cloudflare but you can also tar pit them. Accept the request and send a response, one character at a time (make sure you uncork and flush buffers/etc), with a 30 second delay between characters.
700 requests/second with say 10Kb headers/response. Sure is a shame your server is so slow.
> You can just buy new, for the same or lower price
But, like the article says, new EVs are selling for about twice as much as a 2-year-old used vehicle of the same make and model. That's a very very far cry from "same or lower price".
> For Tesla owners in the U.S., their 2023 Model Ys are worth 42% less than what they paid two years ago
I want to suggest that there are recent reasons why Tesla, as a brand, has specifically gotten a bit less popular that are unrelated to the entire EV category. (It's Elon. He's the reason.)
That said, to the extent the result holds true for the entire category, I'd suspect it's because EVs are still fairly immature. It's like "resale value of desktop PCs falling rapidly" back in the 90s, when the field was advancing quickly enough that buying used was genuinely a bad idea.
Causality could be reversed here. In markets where technology advances quickly and prices drop, there is very little market for used goods, because why would you buy a 4-year-old whatever when you can get a new one that's twice as good for half the price? You see this in computers, smartphones, TVs, and solar panels (outside of the U.S, where prices are kept artificially high by tariffs). People almost never buy used because there's no reason to. You can just buy new, for the same or lower price, and get something way better.
Instead of threatening to derail the EV transition, lack of resale value might be evidence of the EV transition, particularly when coupled with quickly growing overall sales of EVs globally.
Main author of Anubis here. Have CloudFlare return a HTTP 200 response instead of a rejection at non-200. That makes the bots stop hammering until they get a 200 response.
I've done a fair amount of data-intensive fact checking for journalism articles and have had fact checking done on my own data-intensive reporting.
Couple things:
1. Fact checkers are not paid enough to do what they do. They're usually freelancers and they're usually financially struggling. The dynamics of that are difficult to say the least.
2. Editors change things last minute without informing the journalist whose name the piece is in. It's really not fun to receive threats of lawsuit from a powerful government agency because your editor added something that you never would have added. Once told an editor in-writing three times not to add something and he did it right before publishing.
It sucks being a journalist. Donate to your local investigative newsroom.
Really appreciate Matz stepping up to take on this difficult situation. As a Japanese developer, I’ve been worried about the direction things were going, so it’s reassuring to see this.
I think this is the right move. Thank you to Ruby Core and Matz for stepping up and providing stability to the language and community as a whole.
What is old is new again.
My employer is so conservative and slow that they are forerunning this Local Cloud Edge Our Basement thing by just not doing anything.
I cannot overstate the performance improvement of deploying onto bare metal. We typically see a doubling of performance, as well as extremely predictable baseline performance.
This is down to several things:
- Latency - having your own local network, rather than sharing some larger datacenter network fabric, gives around of order of magnitude reduced latency
- Caches – right-sizing a deployment for the underlying hardware, and so actually allowing a modern CPU to do its job, makes a huge difference
- Disk IO – Dedicated NVMe access is _fast_.
And with it comes a whole bunch of other benefits:
- Auto-scalers becomes less important, partly because you have 10x the hardware for the same price, partly because everything runs 2x the speed anyway, and partly because you have a fixed pool of hardware. This makes the whole system more stable and easier to reason about.
- No more sweating the S3 costs. Put a 15TB NVMe drive in each server and run your own MinIO/Garage cluster (alongside your other workloads). We're doing about 20GiB/s sustained on a 10 node cluster, 50k API calls per second (on S3 that is $20-$250 _per second_ on API calls!).
- You get the same bill every month.
- UPDATE: more benefits - cheap fast storage, run huge Postgresql instances at minimal cost, less engineering time spend working around hardware limitations and cloud vagaries.
And, if chose to invest in the above, it all costs 10x less than AWS.
Pitch: If you don't want to do this yourself, then we'll do it for you for half the price of AWS (and we'll be your DevOps team too):
Email: adam@ above domain
I think you can get much farther with dedicated servers. I run a couple of nodes on Hetzner. The performance you get from a dedicated machine even if it is a 3 year old machine that you can get on server auction is absolutely bonkers and cannot be compared to VMs. The thing is that most of the server hardware is focused towards high core count, low clock speed processors that optimize for I/O rather than compute. It is overprovisioned by all cloud providers. Even the I/O part of the disk is crazy. It uses all sorts of shenanigans to get a drive that sitting on a NAS and emulating a local disk. Most startups do not need the hyper virtualized, NAS based drive. You can go much farther and much more cost-effectively with dedicated server rentals from Hetzner. I would love to know if they are any north-american (particularly canadian) companies that can compete with price and the quality of service like Hetzner. I know of OVH but I would love to know others in the same space.
Making the obviously-abusive bot prohibitively expensive is one way to go, if you control the terminating server.
gzip bomb is good if the bot happens to be vulnerable, but even just slowing down their connection rate is often sufficient - waiting just 10 seconds before responding with your 404 is going to consume ~7,000 ports on their box, which should be enough to crash most linux processes (nginx + mod-http-echo is a really easy way to set this up)
These feeders are part of a stray cat control program in China, aiming to both feed & neuter stray cat populations there.
There have been some distasteful incidents of online groups organizing to try and harm/kill specific cats famous through this feeder program. China lacks animal welfare laws to protect these cats, it's not a crime. So people have taken to identifying these abusers and reporting them to their employer, university etc. Abusers have been fired and expelled over such cases. Governments overseas whose citizens participate in such online abuse groups need to be doing more. Membership in online animal abuse groups needs to be criminalized.
I have more than 100 books that I bought with actual money on Apple's iBooks (or whatever it was called back in 2010-2012). I no longer use an iPad and would like to be able to read them on my Kindle. Because of DRM, I can't. I'm all for supporting authors and the various editors, etc., but I feel like I've already done that in this case.
Fun fact: this is one of the few situations in the US where a prosecutor could claim that this is criminal speech (though I hope and trust they would not, and if it did it would get thrown out by any court respecting the First Amendment).
Not a civil issue, like libel or fraud, but the sort of talk that can get a policeman to come and drag you off to jail. If you've ever wondered why DRM is so roundly hated by engineers of a certain age, it's because not only it dumb makework that they are required to implement, not only is it extremely irritating to discover it interfering with your own computer, but if you do effectively point out how dumb, irritating, and eminently circumventable it is, they made it against the law to even tell anyone.
https://www.eff.org/press/releases/licensing-scheme-fair-use...
My coworker recently showed me this plugin [1] that fades out all Rust code that is unrelated to the variable under the cursor. Think of it as a more powerful version of the "click to highlight all appearances" you can do in most IDEs but it actually does information flow analysis on the code.
[1]: https://github.com/willcrichton/flowistry
This is tangential, but I worked for a food delivery startup (a conscientious one) for a couple of years and food delivery is a terribly extractive business that kills restaurants. Either order from the restaurant directly or just go there yourself, Doordash et. al. will kill your favorite restaurant with your help. The numbers don't add up in the kitchen's favor.
For books only available through Amazon my workflow used to be buying it, downloading it with their desktop app, importing into Calibre, converting to epub and stripping DRM, then pushing it onto my Kobo.
They broke that a while ago by making their DRM even worse, so now I just pirate those books.
Hell hath no fury like an engineer angered! This was such a good read and epitomizes hacking:
"Was it worth it? To read one book? No. To prove a point? Absolutely. To learn about SVG rendering, perceptual hashing, and font metrics? Probably yes."
You forgot mcp-everything!
Yes, it's a mess, and there will be a lot of churn, you're not wrong, but there are foundational concepts underneath it all that you can learn and then it's easy to fit insert-new-feature into your mental model. (Or you can just ignore the new features, and roll your own tools. Some people here do that with a lot of success.)
The foundational mental model to get the hang of is really just:
* An LLM
* ...called in a loop
* ...maintaining a history of stuff it's done in the session (the "context")
* ...with access to tool calls to do things. Like, read files, write files, call bash, etc.
Some people call this "the agentic loop." Call it what you want, you can write it in 100 lines of Python. I encourage every programmer I talk to who is remotely curious about LLMs to try that. It is a lightbulb moment.
Once you've written your own basic agent, if a new tool comes along, you can easily demystify it by thinking about how you'd implement it yourself. For example, Claude Skills are really just:
1) Skills are just a bunch of files with instructions for the LLM in them.
2) Search for the available "skills" on startup and put all the short descriptions into the context so the LLM knows about them.
3) Also tell the LLM how to "use" a skill. Claude just uses the `bash` tool for that.
4) When Claude wants to use a skill, it uses the "call bash" tool to read in the skill files, then does the thing described in them.
and that's more or less it, glossing over a lot of things that are important but not foundational like ensuring granular tool permissions, etc.
I fear the conceptual churn we're going to endure in the coming years will rival frontend dev.
Across ChatGPT and Claude we now have tools, functions, skills, agents, subagents, commands, and apps, and there's a metastasizing complex of vibe frameworks feeding on this mess.
That's short sighted. A burrito is a perfect candidate for ballistic trajectory. They can easily absorb the high g-force associated with traditional mortar-style launch system, even up to exotic "space gun" capable of intercontinental delivery.
best case scenario is walkable neighborhoods with lots of little tasty restaurants at affordable prices around the corner from everybody.
We've got a long way to go on actually building out our own country in a desirable way.
I might be in the minority here but I've consistently found Gemini to be better than ChatGPT, Claude and Deepseek (I get access to all of the pro models through work)
Maybe it's just the kind of work I'm doing, a lot of web development with html/scss, and Google has crawled the internet so they have more data to work with.
I reckon different models are better at different kinds of work, but Gemini is pretty excellent at UI/UX web development, in my experience
Very excited to see what 3.0 is like
We're still moving thousands of pounds of vehicle around a public highway to carry a 1lb burrito, obviously lightweight aerial drones are the future for food and grocery delivery.
I don't think you need to "justify layoffs". If a company feels an employee is not being productive, they should be allowed to let them go. Same way if an employee feels he is getting underpaid or wants to work somewhere else, they should be allowed to leave.
I've actually always liked working for companies in which the objective was straight forward. None of this "we're a family" stuff. You should be kind, and all the places I was at were kind. But layoffs are a reality and reducing headcount at times is part of that. You need a way to get rid of dead wood, otherwise you would be too afraid to grow and hire when you need to.
> "EA is not a struggling company," the statement reads. "With annual revenues reaching $7.5 billion and $1 billion in profit each year, EA is one of the largest video game developers and publishers in the world."
Seems to be a common theme in 2025: Actually-healthy companies cosplaying as struggling companies, as an excuse to justify layoffs and other activities that transfer wealth and power from employees to management and shareholders. Like, does anyone think any of these BigTech (and MediumTech) companies who are all doing layoffs are really "struggling" and "vulnerable"? It's always just an unbelievable excuse.
Precisely why I built https://freetofile.com (it’s a simple static site with React for internationalization that automatically renders in Spanish, Chinese, Haitian Creole, or English depending on browser settings). It’s shocking and depressing how many low income people don’t know they don’t need to spend $100-200 to file their taxes.
I want to blanket my area (well the whole country really but baby steps…) in signs with the URL during tax season. I really do loathe the entire industry at this point due to their gross practices around free filing. Some offer “free” online filing but deceptively upsell until they squeeze some money out of the customer. So I want to make any little push back I can against these companies.
Big tech companies (the money behind the Open Source Initiative) have done a few things.
1. They co-opted the free software movement and made it more business friendly.
2. They convinced people that Open Source is pure and software that isn’t Open Source is unclean.
3. They convinced a bunch of developers that their definition of Open Source that was specifically crafted to protect business interests is canon.
4. They convinced a well meaning subset of those developers to police the other devs and pressure them to release their software under big tech approved licenses.
Why does the US have a tax prep industry in the first place?
In every other country in the world, taxes are handled by their respective financial authorities.
Why must every service and thing in the US must be a private profit making thing?
First sign of a profession having a backbone in months.
Although the silent treatment the generals dished out at recent meeting wasn’t bad either
The progressive introduction of automated type checking in Elixir should serve as a reference on how to improve a programming language gracefully without breaking changes.
So many examples of programming languages have huge breaking changes between versions that end up creating a split in the ecosystem that takes years to resolve.
Thankfully José has been very clear about Elixir being done since at least 2018. The language is stable and the language/core foundation is not changing anymore.
https://www.youtube.com/watch?v=suOzNeMJXl0
Truly outsanding work and stewardship.
> I would say that specifically with Secure Boot, Microsoft actually promoted user choice: A Windows Logo compliant PC needs to have Microsoft's root of trust installed by default. Microsoft could have stopped there, but they didn't.
This was not the case with the initial rollout of Secure Boot, it was combined with locked BIOS to lock PCs so that they could only boot Windows 8 on some devices. This was the case on Windows RT ARM machines from that era.
All that has to be done today for machines to be locked down again is to flip a bit or blow an e-fuse. It's already the case on phones and tablets.
There is also a real potential for abusing TPMs or cryptographic co-processors to enforce remote attestation.
I say this as someone who agrees with your first paragraph and uses Secure Boot + TPMs on all of my machines.
For all the hate that Google (rightly) gets for some of their work in other domains, I appreciate that they continue to put major resources behind using AI to try and save lives in medicine and autonomous driving.
Easy to take for granted, but their peer companies are not doing this type of long term investment.
I am a IT solutions provider for the public and small business. I think the changes to Windows 11 is gearing up to work with organizations to create a surveillance state.
So I have to decided to promote Linux over Windows for computers I build for customers. If you have any suggestions on how I can make this promotion, better let me know.
Or to fit in a narrower window:
Chip | Process | CPU | GPU | Neural | Memory | Unified | Geekbench6
| | Cores | | Engine | Bandwidth | Memory | Single / Multi
-----|---------|-----------|------|---------|-------------|---------|----------------------
M1 | 5 nm G1 | 8: 4P+4E | 7–8 | 16-core | 68.25 GB/s | 16 GB | 2346 / 8346
M2 | 5 nm G2 | 8: 4P+4E | 8–10 | 16-core | 100 GB/s | 24 GB | 2586 / 9672
M3 | 3 nm G1 | 8: 4P+4E | 8–10 | 16-core | 100 GB/s | 24 GB | 2965 / 11565
M4 | 3 nm G2 | 10: 4P+6E | 8–10 | 16-core | 120 GB/s | 32 GB | 3822 / 15031
M5 | 3 nm G3 | 10: 4P+6E | 10 | 16-core | 153 GB/s | ≤32 GB | 4133 / 15437 (9 core)It’s incredibly annoying to read. So many super short sentences with the “not just X. Also Y” format. Little hooks like “The attack vector?”
“Not fancy security tools. Not expensive antivirus software. Just asking my coding assistant…”
I actually feel like AI articles are becoming easier to spot. Maybe we’re all just collectively noticing the patterns.
Let's see if I can turn this into an ASCII table and have it survive HN's reformatting.
+------+------------------+--------------+----------+----------------+-------------------+-------------------+---------------------------+
| Chip | Process | CPU Cores | GPU | Neural Engine | Memory Bandwidth | Unified Memory | Geekbench6 (Single/Multi) |
+------+------------------+--------------+----------+----------------+-------------------+-------------------+---------------------------+
| M1 | 5 nm | 8 (4P+4E) | 7–8 | 16-core Neural | 68.25 GB/s | 16 GB | ~2346 / 8346 |
| M2 | 5 nm (G2) | 8 (4P+4E) | 8–10 | 16-core Neural | 100 GB/s | 24 GB | ~2586 / 9672 |
| M3 | 3 nm (first-gen) | 8 (4P+4E) | 8–10 | 16-core Neural | 100 GB/s | 24 GB | ~2965 / 11565 |
| M4 | 3 nm (second-gen)| 10 (4P+6E) | 8–10 | 16-core Neural | 120 GB/s | 32 GB | ~3822 / 15031 |
| M5 | 3 nm (third-gen) | 10 (4P+6E) | 10 | 16-core Neural | 153 GB/s | up to 32 GB | ~4133 / 15437 (9-core) |
+------+------------------+--------------+----------+----------------+-------------------+-------------------+---------------------------+This article is so interesting, but I can’t shake the feeling it was written by AI. The writing style has that feel for me.
Maybe that shouldn’t bother me? Like, maybe the author would never have had time to write this otherwise, and I would never have learned about his experience.
But I can't help wishing he'd just written about it himself. Maybe that's unreasonable--I shouldn't expect people to do extra work for free. But if this happened to me, I would want to write about it myself...
Base models only:
- M1 | 5 nm | 8 (4P+4E) | GPU 7–8 | 16-core Neural | Memory Bandwidth: 68.25 GB/s | Unified Memory: 16 GB | Geekbench6 ~2346 / 8346
- M2 | 5 nm (G2) | 8 (4P+4E) | GPU 8–10 | 16-core Neural | Memory Bandwidth: 100 GB/s | Unified Memory: 24 GB | Geekbench6 ~2586 / 9672
- M3 | 3 nm (first-gen) | 8 (4P+4E) | GPU 8–10 | 16-core Neural | Memory Bandwidth: 100 GB/s | Unified Memory: 24 GB | Geekbench6 ~2965 / 11565
- M4 | 3 nm (second-gen) | 10 (4P+6E) | GPU 8–10 | 16-core Neural | Memory Bandwidth: 120 GB/s | Unified Memory: 32 GB | Geekbench6 ~3822 / 15031
- M5 | 3 nm (third-gen) | 10 (4P+6E) | GPU 10 | 16-core Neural | Memory Bandwidth: 153 GB/s | Unified Memory: up to 32 GB | Geekbench6 ~4133 / 15,437 (9-core sample)
Pretty cute pelican on a slightly dodgy bicycle: https://tools.simonwillison.net/svg-render#%3Csvg%20viewBox%...
"transforming real estate with blockchain" is the only red flag needed
To any Linux users, I recently bought a fully loaded M4 MacBook pro to replace my aging Lenovo and strongly regret it. I thought I would use it for playing with LLMs, but local dev on a Mac is not fun and I still don't have it fully set up. I'll probably replace it with a framework at some point in the near future.
Edit: okay, that garnered more attention than I expected, I guess I owe a qualification.
1. Everything is just slightly different. I had to split all my dot files into common/Linux/Mac specific sections. Don't expect to be able to clone and build any random C++ project unless someone in the project is specifically targeting Mac.
2. Not everything is supported natively on arm64. I had an idea and wanted to spin up a project using DynamoRIO, but wasn't supported. Others have mentioned the docker quirks.
3. The window manager. I'm not a fan of all the animations and needing to gester between screens (and yes, I've been down the hotkeys rabbit hole). To install a 3rd party window manager you need to disable some security setting because appearantly they work by injecting into the display manager and calling private APIs.
So my person takeaway was that I took the openness of the Linux ecosystem for granted (I've always had a local checkout of the kernel so I can grep an error message if needed). Losing that for me felt like wearing a straightjacket. Ironically I have a MBP at work, but spend my day ssh'd into a Linux box. It's a great machine for running a web browser and terminal emulator.
> Tahoe however makes my M1 Air feel sluggish doing the exact same tasks ive been last couple of years.
I have a work provided M2 Pro with 32GB of RAM. After the Tahoe upgrade it feels like one of the sluggish PCs at the house. It is the only one that I can see the mouse teleporting sometimes when I move it fast. This is after disabling transparency in Accessibility settings mind you, it was even worse before.
The M5 MacBook Pro still gets the Broadcom WiFi chip but the M5 iPad Pros get the N1 and C1X (Sweet).
All in all, apple is doing some incredible things with hardware.
Software teams at apple really need to get their act together. The M1 itself is so powerful that nobody really needs to upgrade that for most things most people do on their computers. Tahoe however makes my M1 Air feel sluggish doing the exact same tasks ive been last couple of years. I really hope this is not intentional from Apple to make me upgrade. That would be a big let down.
I was super interested in the Vision Pro when it was first released. Then I found out they went with an app model and the device could only display a single MacOS window. There went my dream of surrounding myself with a bunch of vim windows and terminals.
If they'd focused on maximizing the device's usefulness instead of its revenue stream, maybe things would have worked out better.
I once worked for the yellow pages in Switzerland. Our paid clients had a dashboard which reported how many users visited their business entry.
We at engineering decided to filter out bots. Figures fell dramatically by more than 50%.
In less that a day business mandated us to remove the filter.
Bots are real people after all
The modern Apple feels like their hardware teams way outperforming the software teams.
They tried to call this "universal" until people pointed out it is the opposite of universal. This program is a wild distortion of what UBI is meant to be.
Everyone who would _like_ to be an artist, but can't afford to be one, is disqualified. Meanwhile, the acquaintance of mine who sold his house in London at a large profit and retired to a cottage in Westmeath to live off his gains and noodle around on the guitar a bit is a recipient of funds from this program.
Tellingly there's very little information about how to _become_ an artist with this program.
Edit/addendum: Worth noting they've produced some _very_ dubious numbers to claim this program is a net gain economically. https://www.rte.ie/culture/2025/0923/1534768-basic-income-fo...
""" A key component of the total benefits came from psychological wellbeing, which contributed almost €80 million. In addition, the report estimates that audience engagement with the arts generated €16.9 million in social value, based on public willingness-to-pay for cultural experiences. """
And, as much as I like psychological wellbeing (who doesn't!) - saying that it's worth €80 million when you didn't actually get €80 million doesn't help things when it comes time to pay for the program. I'm unsurprised that giving people money improved their psychological well being.
I'd be more excited to see basic income for Deliveroo riders and people working in chippers.
Seems organised crime is taking advantage of less-policed (or lawless) regions in Cambodia and Burma.
The spectre of a similar scam has resulted in lower Chinese tourism to Thailand. Basically Chinese (among other nationalities) tourists were led from the airport by fake immigration staff, put into cars and driven to Burma and forced to work 18 hour days in scam call centers.
A chilling but extremely worthwhile read.
https://www.reuters.com/graphics/SOUTHEASTASIA-SCAMS/mypmxwd...
The takeaway here isn’t that serverless doesn’t work, it’s that the authors didn’t understand what they were building on. Putting a latency-critical API on a stateless edge runtime was a rookie mistake, and the pain they describe was entirely predictable.
Hi HN. I run a marketing agency and fell down this rabbit hole after a client's analytics made no sense (50k visitors, 47 sales). I ended up building a simple script to track user behavior and analyzed 200+ small e-commerce sites. The average was 73% bot traffic that standard analytics counts as real.
The bots are getting creepily good at mimicking engagement. I wrote up my findings, including some of the bizarre patterns I saw and the off-the-record conversations I had with ad tech insiders. It seems like a massive, open secret that nobody wants to talk about because the whole system is propped up by it.
I'm curious if other developers, founders, or marketers here have seen similar discrepancies in their own data.
People oppose everything.
* Lattice overhead powerlines? Eyesore (should use the new T style ones), house values, wind noise, hums, WiFi interference, cancer, access roads, hazard to planes, birds
* T-frame pylons: boring (https://www.theguardian.com/commentisfree/2015/apr/13/electr...), eyesore (we prefer the lattice ones), most of the above too
* Underground: damaging to the environment, end stations are eyesores/light polluters, more construction traffic, should be HVDC not AC, house values
* Solar farms: waste of good land (golf courses are fine) noise somehow, construction, eyesore (but a 400 acre field of stinky bright yellow rapeseed is OK), house values
* Onshore Wind farms: all the birds all the time, access, eyesore, noise, dangerous, should be offshore, house value, waste of land, I heard on Facebook the CO2 takes 500 years to pay back
* Offshore wind farms: eyesores, radar hazard, all the birds, house values somehow, navigation hazard, seabed disruption
* Build an access road: destroying the countryside, dust if not surfaced, drainage, house values
* Don't build an access road: destroying roads, HGVs on local roads, house values
* Nuclear: literally all the reasons plus scary
Some of them are fair on their own, but it really adds up to a tendentious bunch of wankers at every turn who think the house they bought for 100k in 1991 and is now worth 900k is the corner of the universe.
> As a foreign influence
I'm sure these people would never take foreign cash: https://www.bbc.co.uk/news/articles/c93k584nvgeo https://www.bbc.co.uk/news/articles/clyk1j92195o
Why can't they just partner with postmarketOS here?
Why do we have to have /e/OS instead of a better supported LineageOS, because /e/ is a 1:1 copy anyways?
Why do we have to have a Librephone project now instead of partnering with say, Fairphone and the Pine64 people?
Open source loses this war because proprietary devices are streamlined. The only thing that comes close to this is GrapheneOS, LineageOS, and postmarketOS.
LineageOS has huge problems since the mandatory eBPF requirements of late Android versions, which postmarketOS and its upstreamed kernel drivers could fix. GrapheneOS has huge problems because of Pixel devices, which LineageOS could help with.
We need a unification of this ecosystem because each on their own is hardly surviving on their own against the megacorporations.
Wow. $15 billion is an astronomical, nearly unfathomable amount of money in Cambodia. This must have been a huge operation. It's almost 1/3 of the country's entire GDP!
Well done DOJ. Hopefully the victims get their money back.
I find LLM generated code ends up pushing review/maintenance burden onto others. It "looks" right at first glance, and passes superficial tests, so it's easy to get merged. But then as you build on top of it, you realize the foundations are hastily put together, so a lot of it needs to be rewritten. Fine for throwaway or exploratory work, but heaven help you if you're working in a project where people use LLMs to "fix" bugs generated by previous LLM generated code.
So yes it does increase "velocity" for the person A who can get away with using it. But then the decrease in velocity for person B trying to build on top of that code is never properly tracked. It's like a game of hot potato, if you want to game the metrics you better be the one working on greenfield code (although I suppose maintenance work has never been looked at favorably in performance review; but now the cycle of code rot is accelerated)
Ultimately, I don't think the most important challenge is in binary firmware blobs, but the software which people depend upon to run their lives. What does it matter if you can run a completely free software stack on your phone, if your bank software (or your required government ID, as is looking depressingly likely) requires you to run a Big Tech approved phone OS? Perhaps the FSF can't do much about that, but that is where I feel they could truly make the biggest difference for freedom for the average user.
“None of the safe programming languages existed for the first 10 years of SQLite's existence. SQLite could be recoded in Go or Rust, but doing so would probably introduce far more bugs than would be fixed, and it may also result in slower code.”
Modern languages might do more than C to prevent programmers from writing buggy code, but if you already have bug-free code due to massive time, attention, and testing, and the rate of change is low (or zero), it doesn’t really matter what the language is. SQLIte could be assembly language for all it would matter.
For a real world example of the challenges of harnessing LLMs, look at Apple. Over a year ago they had a big product launch focused on "Apple Intelligence" that was supposed to make heavy use of LLMs for agentic workflows. But all we've really gotten since then are a couple of minor tools for making emojis, summarizing notifications, and proof reading. And they even had to roll back the notification summaries for a while for being wildly "out of control". [1] And in this year's iPhone launch the AI marketing was toned down significantly.
I think Apple execs genuinely underestimated how difficult it would be to get LLMs to perform up to Apple's typical standards of polish and control.
[1] https://www.bbc.com/news/articles/cge93de21n0o
Hi HN, submitting from a burner since I'm an applicant this current medical residency admissions cycle. I thought it was interesting to show the real world implications of using LLMs to extract information from PDFs. For context, thalamus is a company that handles the "backend" for residency programs and all the applications they receive (including handling who to invite for interviews, etc). One of the more important factors in deciding applicant competitiveness is their medical school performance (their grades), but that information is buried in PDFs sent by schools (often not standardized). So this year, they decided to pilot a tool that would extract that info (using "GPT-5o-mini": https://www.thalamusgme.com/blogs/methodology-for-creation-a...). Some programs have noticed there is a discrepancy between extracted vs reported grades (often in the direction of hallucinating "fails") and brought it to the attention of thalamus. Unfortunately, it doesn't look like the main company is discontinuing usage of the tool.
Regardless, given that there have been a number of posts looking into usage of LLMs for numerical extraction, I thought this story useful would be a cautionary tale.
EDIT: I put "GPT-5o-mini" in quotes since that was in their methodology...yes, I know the model doesn't exist
I think that if we didn't do TLS, every ISP would be injecting ads into websites these days. Making it difficult for middle-of-the-road interlopers is a good thing. ISPs don't want the customer service burden of proxy configurations and custom certs (god knows your IT department hates the support aspect of this tampering), so TLS keeps us free of excessive advertising. (Of course, they like do tampering with DNS which is why we have to do DNS-over-HTTPS. If you make it easy to tamper with your traffic, your ISP has a good business case to tamper with your traffic. Sad but true.)
I'm not as convinced as the author is that nation states can easily tamper with certificates these days. I am not sure how much CT checking we do before each page load, but either nation states are compelling the issue of certs that aren't in the CT database, or they are and you can just get a list of who the nation states are spying on. Seems like less of a problem than it was a decade ago.
The author seems to miss the one guarantee that certificates provide; "the same people that controlled this site on $ISSUANCE_DATE control the site right now". That can be a useful guarantee.
THANK YOU. People look at me like I’m insane when I tell them that their overly-complicated pipeline could be easily handled by a couple of beefy servers. Or at best, they’ll argue that “this way, they don’t have to manage infrastructure.” Except you do - you absolutely do. It’s just been partially abstracted away, and some parts like OS maintenance are handled (not that that was ever the difficult part of managing servers), but you absolutely need to configure and monitor your specific XaaS you’re renting.
Just to be honest for a bit here... we also should be asking what kind of scale?
Quite a while ago, before containers were a thing at all, I did systems for some very large porn companies. They were doing streaming video at scale before most, and the only other people working on video at that scale were Youtube.
The general setup for the largest players in that space was haproxy in front of nginx in front of several PHP servers in front of a MySQL database that had one primary r/w with one read only replica. Storage (at that time) was usually done with glusterfs. This was scalable enough at the time for hundreds of thousands of concurrent users, though the video quality was quite a bit lower than what people expect today.
Today at AWS, it is easily possible for people to spend a multiple of the cost of that hardware setup every month for far less compute power and storage.
I'm absolutely loving KDE since I returned to desktop Linux after a long absence.
What really shocks me is how few of the big distros make KDE a default or "first class" DE choice. If I was a novice user coming from Windows, I'd much prefer KDE, which if you stick to the GUI is very navigable and similar in some ways.
Some new information has just emerged in the Dutch press.
According to a just published article by the NRC newspaper [0], the owner wanted to use Nexperia's funds to finance another business (WingSkySemi) which was in financial trouble. He would have done this by making Nexperia place large orders for wafers it did not need. The article mentions Nexperia only requiring $70-80 million in wafers, while Nexperia would have ordered wafers from WingSkySemi with a value of $200 million.
In order to achieve this, the owner is said to have put strawmen without financial experience in place and fire European directors. When other directors raised the alarm about this, they were fired according to the article. This issue was raised to the Dutch government, which then intervened.
Have a look at the original article through google translate, it provides a lot of interesting and important details to this story.
[0] https://www.nrc.nl/nieuws/2025/10/14/chinese-topman-gebruikt...
> Chief Pentagon spokesman Sean Parnell said (...) "This has caused reporters to have a full blown meltdown, crying victim online."
Interesting use of language... seems like the mask is coming off everywhere now, not just where I live (Hungary).
I've been intentionally skipping on a lot of our local political reporting, so I was really quite surprised to see recently how lowbrow the language used by politicians, specifically those in power, has gotten these days. Especially how flagrant they are about it too.
This is a very meta, and to many I'm sure trivial, thing to take issue with, yes, but if those in authority are this unashamedly drunk on power, and look down on those they rule over so openly, I'd really question how fit they are to represent people's collective best interest.
> if you’re working on novel code, LLMs are absolutely horrible
This is spot on. Current state-of-the-art models are, in my experience, very good at writing boilerplate code or very simple architecture especially in projects or frameworks where there are extremely well-known opinionated patterns (MVC especially).
What they are genuinely impressive at is parsing through large amounts of information to find something (eg: in a codebase, or in stack traces, or in logs). But this hype machine of 'agents creating entire codebases' is surely just smoke and mirrors - at least for now.
Marketing is being done really well in 2025, with brands injecting themselves into conversations on Reddit, LinkedIn, and every other public forum. [1]
CEOs, AI "thought leaders," and VCs are advertising LLMs as magic, and tools like v0 and Lovable as the next big thing. Every response from leaders is some variation of https://www.youtube.com/watch?v=w61d-NBqafM
On the ground, we know that creating CLAUDE.md or cursorrules basically does nothing. It’s up to the LLM to follow instructions, and it does so based on RNG as far as I can tell. I have very simple, basic rules set up that are never followed. This leads me to believe everyone posting on that thread on Cursor is an amateur.
Beyond this, if you’re working on novel code, LLMs are absolutely horrible at doing anything. A lot of assumptions are made, non-existent libraries are used, and agents are just great at using tokens to generate no tangible result whatsoever.
I’m at a stage where I use LLMs the same way I would use speech-to-text (code) - telling the LLM exactly what I want, what files it should consider, and it adds _some_ value by thinking of edge cases I might’ve missed, best practices I’m unaware of, and writing better grammar than I do.
Edit:
[1] To add to this, any time you use search or Perplexity or what have you, the results come from all this marketing garbage being pumped into the internet by marketing teams.
Some of the stuff that was extracted from the unencrypted traffic in the link:
- T-Mobile backhaul: Users' SMS, voice call contents and internet traffic content in plain text.
- AT&T Mexico cellular backhaul: Raw user internet traffic
- TelMex VOIP on satellite backhaul: Plaintext voice calls
- U.S. military: SIP traffic exposing ship names
- Mexico government and military: Unencrypted intra-government traffic
- Walmart Mexico: Unencrypted corporate emails, plaintext credentials to inventory management systems, inventory records transferred and updated using FTP
This is insane!
While it is important to work on futuristic threats such as Quantum cryptanalysis, backdoors in standardized cryptographic protocols, etc. - the unfortunate reality is that the vast majority of real-world attacks happen because basic protection is not enabled. Good reminder not take our eyes off the basics.
It's a great thing they are not backing down. Given how many institutions have complied in advance, we need as many exemplars of better behaviour as possible.
The headline and article try to bias and frame the story to make people question: "Is OpenAI snitching on me?"
In reality, Uber records and conflicting statements incriminated him. He seems to be the one who provided the ChatGPT record to try to prove that the fire was unintentional.[1]
> He was visibly anxious during that interview, according to the complaint. His efforts to call 911 and his question to ChatGPT about a cigarette lighting a fire indicated that he wanted to create a more innocent explanation for the fire's start and to show he tried to assist with suppression, the complaint said.
[1] https://apnews.com/article/california-wildfires-palisades-lo...
Having worked in a job shop, a factory that did gears down to quantity one, I became quite aware of the differences between IT, my previous job, and actual physical production.
The machine tools were all made 50+ years ago. Changing anything was a dangerous thing to do, because you might cause jobs that have known and reliable setups that are done a few times a year in quantity, to fail, erasing the profits for the job, and possibly losing customers.
The rush to fill brand new high energy intensive data centers with hardware that has commercially useful lifetimes measured in months (instead of decades for machine tools) seems quite short sighted to me.
I work for one of the largest packaging companies in the world. Customers across the board in the US are cutting back on how much packaging they need due to presumably lower sales volume. Make of that information what you will.
I've seen a few posters ask already, so I figured I'd answer what the PS2 analog button's function was.
The button switches between two modes of the analog joysticks, either to behave with their normal functionality, or to simply be a digital input (so just round all movement to either up/down/left/right). For PS2 games, you typically wouldn't want to do this. Instead, the functionality exists because the PS2 was backwards compatible with PS1 titles. The original PS1 controller didn't have analog sticks at all, just the D-Pad for navigation. After a few years (and the success of Nintendo's N64 analog controller) Sony released a revised version of the controller that included two joysticks, which their controllers still mimic to this day. However, those PS1 games released prior to the analog controller wouldn't always behave correctly if you tried to use an analog input scheme, so Sony added a mode to allow the Joysticks to function the same as the D-Pad, in case players preferred it.
Other fun fact, the analog controller was not the same as their more famous Dualshock controller. There was a short-lived PS1 Dual Analog controller which just added the joysticks. It only lasted a few months before Sony replaced it with one that supported rumble functionality (also after being inspired by the N64), this was the Dualshock.
Everything is unless your app is a React todolist or leatcode questions.
This shouldn’t be a surprise to anyone who has been using Python and has tried uv.
Python dependency management and environments have been a pain for 15 years. Poetry was nice but slow and sometimes difficult.
Uv is lightning fast and damn easy to use. It’s so functional and simple.
> the repo is too far off the data distribution
ah, this explains why these models have been useless to me this whole time. everything i do is just too far off the data distribution!
Everybody is a fan of free access and capital markets, until a foreign entity purchases something of importance.
It’s a continuation of recent trends and closing markets.
Nobody in their sane mind would allow a company like ASML or the likes to be purchased by competitors.
But the irony is that when a non-European entity were to do something like this, e.g. nationalize their oil or mining etc. industry or a firm, the whole hell would brake loose.
Interesting exchange on the use of AI coding tools:
curious how much did you write the code by hand of it?
Karpathy: Good question, it's basically entirely hand-written (with tab autocomplete). I tried to use claude/codex agents a few times but they just didn't work well enough at all and net unhelpful, possibly the repo is too far off the data distribution.
Anecdotally, our company's next couple quarters are projected to be a bloodbath. Spending is down everywhere, nearly all of our customers are pushing for huge cuts to their contracts, and in turn literally any costs we can jettison to keep jobs is being pushed through. We're hearing the same from our customers.
AI has been the only new investment our company has made (half hearted at that). I definitely get the sense that everyone pretending things are fine to investors, meanwhile they are playing musical chairs.
Back in my economics classes at college, a professor pointed out that a stock market can go up for two reasons: On one hand, the economy is legitimately growing and shares are becoming more valuable. But on the other hand, people and corporations could be cutting spending en masse so there's extra cash to flood the stock markets and drive up prices regardless of future earnings.
My 4xe died in my driveway on Saturday after the update. Let me explain, from the perspective of a 4xe owner, how bad the response has been from Jeep/Stellantis:
- As of Monday 8am ET, zero legitimate communication from any Jeep-related accounts on any social media platform, or any other form of acknowledgement from the company (unless I've missed something?)
- I only found out about the issue after finally searching a few Jeep groups on Facebook (of all places) to see if anyone else was experiencing the weird failure mode I was after the update.
- The only remotely-official info was from a 'JeepCares' account (which is ran by Jeep) on some random off-roading forum? We were seriously all living off of screenshots from this forum, and the advice coming from the JeepCares accounts was contradictory: they claimed that the Uconnect update was separate from the telematics update, and that there was no way to stop the telematics update if the vehicle received it. Later they gave advice to defer the Uconnect update, making it sound like they were coupled.
- Due to the lack of info from Jeep, people were coming up with all kinds of "if you reboot Uconnect while the Jeep's in ACC mode, it clears the check engine light". This probably did clear the CEL but didn't fix the fault.
- There is no way to tell if you received the bad update.
- There is no way to tell if you received the 'fix' either.
- Dealerships have literally no idea what is going on.
- You're basically at risk of your Jeep going limp (power loss, unable to safely make it to the shoulder) and being stranded on the highway, even as I write this.
Antitrust action is badly needed in this area. It is ridiculous that I need permission from my device manufacturer to install software on hardware I own. There is no viable alternative than to live in Apple and Google’s ecosystems. This duopoly cannot be allowed to keep this much control of the mobile platforms.
We need to stop calling it "sideloading", we should call it freely installing software. The term "sideloading" makes it sound shady and hacky when in reality it is what we have been able to do on our computers since forever. These are not phones, they are computers shaped like phones, computer which we fully bought with our money, and I we shall install what we want on our own computers.
Installing any app I want outside the Play Store was the primary reason I decided to go with Android, despite most of the people I know using iPhones. If I can't do this anymore, I may as well switch and be able to use iMessage and FaceTime with them.
I always felt that I'm spending too much time in front of a computer, but it was at least somewhat meaningful because I had opportunities to create: write code, blog, and so on.
When smartphones came out, I made a decision early on that I'm just not going to use them in a way that makes my internet footprint follow me everywhere I go. I set them up using a throwaway email account, turned off almost all notifications, and added just family and real-world friends. I think this served me well for nearly two decades. I really only use my phone for maps, photos, and maybe 2-5 messages a day. I honestly never found myself in a situation where I thought to myself, "gosh, I wish I could read my e-mail right now".
But in the past five years, there's been this mounting pressure from app vendors to make sure I can no longer enjoy that. Every other time a friend sends me a web link, I get a popup that detects I'm on mobile and demands I install an app. And they increasingly can't be dismissed, so if I want to view that URL, I need to mail it to myself and open it on a desktop.
If you work for a place that does that, I just hope you stub your toe every morning.
Not why it can’t be done so much as why it isn’t done. Share buybacks allow companies to reward executives directly as their compensation is tied to stock price. If we started not doing that, the priorities might shift, but those executives like things the way they are.
Before Tim Cook Apple had never done a buyback - Jobs was always thinking Apple could do better with the money in R&D than paying off shareholders. Wall Street did not approve of this position, but Jobs wasn’t one to listen to anybody, so it did not matter. Most CEOs are not going to take such a strong position when they, the stockholders, and every other executive can be guaranteed a financial reward through a buyback.
> The first way is to not have recommendation media (think Instagram, TikTok, and all the rest). I'm pro deleting these accounts completely, because it's really easy to re-download the apps on a whim, or *visit them in-browser.*
Tiktok having a borderline unusable web app has done wonders for me. I'll end up on it because someone sent me a link, I can watch that ONE video, a single time, before normally I get a spot-the-boat style captcha or an "install the app" modal. Even trying to get past that point, it feels like the site is somehow falling apart at the seams as you navigate around. I know the concept is "well people will install the app then" but that's also annoyingly frictionful.
They unintentionally made the most literal social media experience: some one sends me media, I watch it once, I leave before the site crumbles to pieces like an ancient tomb that was only held together by a load-bearing dog video.
> In the 20th century, U.S. companies put their excess profits into corporate research labs. Basic research in the U.S. was done in at Dupont, Bell Labs, IBM, AT&T, Xerox, Kodak, GE, et al. This changed in 1982, when the Securities and Exchange Commission ruled that it was legal for companies to buy their own stock (reducing the number of shares available to the public and inflating their stock price.) Very quickly Basic Science in corporate research all but disappeared. Companies focused on Applied Research to maximize shareholder value. In its place, Theory and Basic research is now done in research universities.
I'm not seeing how you get from share buybacks to a shift in priorities in corporate research. If there's a fundamental reason why it can't be done now how it was before the 80's it's not that.
Some people spend the vast majority of their time on their own machine. The gains of convenience can be worth it. And they know enough of the classic tools that it's sufficient in the rare cases when working on another server.
Not everybody is a sysadmin manually logging into lots of independent, heterogeneous servers throughout the day.
These may be objectively superior (I haven't tested), but I have come to realize (like so many others) that if you ever change your OS installation, set up VMs, or SSH anywhere, preferring these is just an uphill battle that never ends. I don't want to have to set these up in every new environment I operate in, or even use a mix of these on my personal computer and the traditional ones elsewhere.
Learn the classic tools, learn them well, and your life will be much easier.
I am baffled that people in this thread write something along "well, depends what you call win" - the goal of Free Software is quite clear. The goal is freedom, computing freedom, freedom of the software user. It is very easy to notice that in 2025 users have less freedom even if they run some Libre Linux distro on their Thinkpads than they had running Win98, because of everything that happens OUTSIDE PC software ecosystem (phones, SaaS etc), and even inside PC world things sometimes are not obvious.
Free Software is losing, simple as that. Even with Kubernetes, as the goal was never to provide free labor and free software infra to companies.
Some more context from a dutch news source[0]:
The ministry of economic affairs intervened out of a fear that crucial technological skills and capacities will leave the Netherlands and Europe. The ministry stated in a press release[1] that there was a threat of a "knowledge leak" (w/e that means exactly) and a possible threat to the European economy.
After this intervention the Dutch government can now stop or reverse decisions within the company. That's only allowed if those decisions are harmful to the interests of the company, or for the future of the company as a Dutch or European business, or to the retaining of this crucial value chain for europe.
The company can appeal this decision in court.
For context, the law that allows this all to happen was passed in 1952 and has never before been used. As much as I think our government is currently ran by a bunch of nincompoops, I am inclined to believe that something quite significant was about to happen for this law to get invoked. What exactly that was can for now only be speculated about.
I can recommend you run google translate (or equivalent) on the press release. It's as close as you can get to the source of this news for now. I can only imagine the government is going to be having plenty of debates on the topic coming up, seeing as this is a very rare use of a very heavy-handed tool.
[0] https://nos.nl/artikel/2586270-kabinet-grijpt-hard-in-bij-ch...
[1] https://www.rijksoverheid.nl/actueel/nieuws/2025/10/12/wet-b...
A couple of months ago, I saw a tweet from @awilkinson: “I just found out how much we pay for DocuSign and my jaw dropped. What's the best alternative?”
Me being naive, I thought “how hard could would it actually be to build a free e-sign tool?”
Turns out not that hard.
In about a weekend, I built a UETA and ESIGN compliant tool. And it was free. And it cost me less than $50. Unlimited free e-sign. https://useinkless.com/
The US is choosing to make itself increasingly redundant. Increasing speed whilst in Reverse gear whilst talking as if they're highway cruising in 5th.
It's nice to know that the rest of the world is still pushing forwards, up a steep hill, in first gear. I hope it can continue, and the hill shrinks.
> "Winning" doesn't have to mean owning every transistor
It absolutely does.
Corporations are pushing remote attestation now. They can detect if we "tampered" with our devices now. They discriminate against us for it. Installed your own open source software? All services denied. Can't even log into your own bank account.
We're marginalized. Second class citizens. There is no choice, it's either corporate owned computers or nothing. What good is free software if we can't run it?
I think this post overstates the "loss" of free software. Yes, closed firmware and locked hardware are real gaps...but that doesn't erase the fact that open software has completely reshaped the modern stack. From Linux and K8s to Postgres and Python, it is the infra of the internet. "Winning" doesn't have to mean owning every transistor; it means setting the norms and powering most of what's built.
I tend to see this kind of absolutist, binary tone a lot from people deeply involved in FOSS... and sometimes I think maybe that mindset is necessary to push the movement forward, but it also feels detached from how much open software has already changed reality.
I've been working on a 3D voxel-based game engine for like 10 years in my spare time. The most recent big job has been to port the world gen and editor to the GPU, which has had some pretty cute knock-on effects. The most interesting is you can hot-reload the world gen shaders and out pop your changes on the screen, like a voxel version of shadertoy.
https://github.com/scallyw4g/bonsai
I also wrote a metaprogramming language which generates a lot of the editor UI for the engine. It's a bespoke C parser that supports a small subset of C++, which is exposed to the user through a 'scripting-like' language you embed directly in your source files. I wrote it as a replacement for C++ templates and in my completely unbiased opinion it is WAY better.
https://github.com/scallyw4g/poof
Currently a one-man side project: https://laboratory.love
Last year, PlasticList found plastic chemicals in 86% of tested foods—including 100% of baby foods they tested. Around the same time, the EU lowered its “safe” BPA limit by 20,000×, while the FDA still allows levels roughly 100× higher than Europe’s new standard.
That seemed solvable.
Laboratory.love lets you crowdfund independent lab testing of the specific products you actually buy. Think Consumer Reports × Kickstarter, but focused on detecting endocrine disruptors in your yogurt, your kid’s snacks, or whatever you’re curious about.
Find a product (or suggest one), contribute to its testing fund, and get full lab results when testing completes. If a product doesn’t reach its goal within 365 days, you’re automatically refunded. All results are published publicly.
We use the same ISO 17025-accredited methodology as PlasticList.org, testing three separate production lots per product and detecting down to parts-per-billion. The entire protocol is open.
Since last month’s “What are you working on?” post:
- 4 more products have been fully funded (now 10 total!)
- That’s 30 individual samples (we do triplicate testing on different batches) and 60 total chemical panels (two separate tests for each sample, BPA/BPS/BPF and phthalates)
- 6 results published, 4 in progress
The goal is simple: make supply chains transparent enough that cleaner ones win. When consumers have real data, markets shift.
Browse funded tests, propose your own, or just follow along: https://laboratory.love
Perhaps not surprising, working breeds – many of which are known to have been artificially selected for high toy or predatory motivation – were overrepresented in the sample.
Hope the PM in charge with the scammy copy designed to trick people into turning this on is happy with the boost in free users falling for it.
My dad turned it on not knowing what it meant and it completely messed up his workflows and now I have to figure out how to safely disable it and move his files back.
I will remember Microsoft causing this problem for him every time I think of or get asked if someone should use a Microsoft product or service.
I used to use Windows Backup with One Drive years ago but it just really pissed me off, especially with how My Documents is handled.
There was that time I discovered several GB of screenshots had been automatically saved to My Pictures from some setting they snuck into the printscreen screen grab tool and then that of course those were automatically uploaded to the cloud. After disabling the option it would sometimes reenable itself.
And game devs throwing random shit into My Documents was also fun. Ubisoft were terrible for this, after playing a game I'd notice a bunch of cache files they dumped into My Docs being uploaded. I mean putting save games and config files inside my docs is annoying enough, random cache files is just taking the piss.
Also windows backup would mess up my desktop between systems on occasion which was also very fun!
I disabled most of the shit but it was still annoying on occasion. Then a year or two ago I solved the problem by just using Linux for 90% of things, Mint at first but now Fedora, and grudgingly booting back into Windows for the other 10% of my needs.
This is chilling (from https://www.jlwranglerforums.com/forum/threads/2024-4xe-loss...):
> On my drive home I abruptly had absolutely no acceleration, the gear indicator on the dash started flashing, the power mode indicator disappeared, an alert said shift into park and press the brake + start button, and the check engine light and red wrench lights came on. I was still able to steer and brake with power steering and brakes for maybe 30 seconds before those went out too. After putting it into park and pressing the brake and start button it started back up and I could drive it normally for a little bit, but it happened two more times on my 1.5 mi drive home.
If that happened on the highway I could easily see people being killed.
I won't make a claim to the accuracy of the numbers, but I can offer an example of how long COVID can be undercounted. My daughter was a competitive long distance runner. Months after recovering from the acute symptoms of COVID, her performance numbers were still about 10% down and no amount of training would allow her to reach her previous level of performance. After many visits to doctors and specialists, she was eventually diagnosed with long COVID due to lung damage. She's still very fast by most measures, but in track, the difference between being in the 99th percentile and the 89th percentile is the difference between a top 3 finish and a bottom 3 finish. This basically ended her track career. In our case, if we didn't have hard data tracking her before and after performance levels, we may not have ever noticed a difference. How many people who aren't competitive athletes are walking around with 10% degraded pulmonary function and just didn't notice?
The Plan 9 operating system.
It's the closest thing to a Unix successor we ever got, taking the "everything is a file" philosophy to another level and allowing to easily share those files over the network to build distributed systems. Accessing any remote resources is easy and robust on Plan9, meanwhile on other systems we need to install specialized software with bad interoperability for each individual use case.
Plan9 also had some innovative UI features, such as mouse chording to edit text, nested window managers, the Plumber to run user-configurable commands on known text patterns system-wide, etc.
Its distributed nature should have meant it's perfect for today's world with mobile, desktop, cloud, and IoT devices all connected to each other. Instead, we're stuck with operating systems that were never designed for that.
There are still active forks of Plan9 such as 9front, but the original from Bell Labs is dead. The reasons it died are likely:
- Legal challenges (Plan9 license, pointless lawsuits, etc.) meant it wssn't adopted by major players in the industry.
- Plan9 was a distributed OS during a time when having a local computer became popular and affordable, while using a terminal to access a centrally managed computer fell out of fashion (though the latter sort of came back in a worse fashion with cloud computing).
- Bad marketing and posing itself as merely a research OS meant they couldn't capitalize on the .com boom.
- AT&T lost its near endless source of telephone revenue. Bell Labs was sold multiple times over the coming years, a lot of the Unix/Plan9 guys went to other companies like Google.
Overall it feels like UV is the best thing to happen to python packaging in two decades, by circumventing the endless non productive discussions on peps and instead just building something that works and is fast. In Rust naturally.
I find the word "engineering" used in this context extremely annoying. There is no "engineering" here. Engineering is about applying knowledge, laws of physics, and rules learned over many years to predictably design and build things. This is throwing stuff at the wall to see if it sticks.
Adobe Flash / Shockwave. After all these decades, I've yet to see a tool that makes it as easy to make games or multimedia as Flash did. One of many reminders recently (many others in politics) that humanity doesn't just inevitably or linearly move forward in any domain, or even 2 steps forward 1 step back. Some things are just lost to time - maybe rediscovered in a century, maybe never.
- Photon, the graphical interface for QNX. Oriented more towards real time (widgets included gauges) but good enough to support two different web browsers. No delays. This was a real time operating system.
- MacOS 8. Not the Linux thing, but Copeland. This was a modernized version of the original MacOS, continuing the tradition of no command line. Not having a command line forces everyone to get their act together about how to install and configure things. Probably would have eased the tradition to mobile. A version was actually shipped to developers, but it had to be covered up to justify the bailout of Next by Apple to get Steve Jobs.
- Transaction processing operating systems. The first one was IBM's Customer Information Control System. A transaction processor is a kind of OS where everything is like a CGI program - load program, do something, exit program. Unix and Linux are, underneath, terminal oriented time sharing systems.
- IBM MicroChannel. Early minicomputer and microcomputer designers thought "bus", where peripherals can talk to memory and peripherals look like memory to the CPU. Mainframes, though, had "channels", simple processors which connected peripherals to the CPU. Channels could run simple channel programs, and managed device access to memory. IBM tried to introduce that with the PS2, but they made it proprietary and that failed in the marketplace. Today, everything has something like channels, but they're not a unified interface concept that simplifies the OS.
- CPUs that really hypervise properly. That is, virtual execution environments look just like real ones. IBM did that in VM, and it worked well because channels are a good abstraction for both a real machine and a VM. Storing into device registers to make things happen is not. x86 has added several layers below the "real machine" layer, and they're all hacks.
- The Motorola 680x0 series. Should have been the foundation of the microcomputer era, but it took way too long to get the MMU out the door. The original 68000 came out in 1978, but then Motorola fell behind.
- Modula. Modula 2 and 3 were reasonably good languages. Oberon was a flop. DEC was into Modula, but Modula went down with DEC.
- XHTML. Have you ever read the parsing rules for HTML 5, where the semantics for bad HTML were formalized? Browsers should just punt at the first error, display an error message, and render the rest of the page in Times Roman. Would it kill people to have to close their tags properly?
- Word Lens. Look at the world through your phone, and text is translated, standalone, on the device. No Internet connection required. Killed by Google in favor of hosted Google Translate.