The lid angle sensor is also serialized to the motherboard: you cannot replace it, or the motherboard, without performing calibration, which can be performed by an apple authorized service provider, or alternatively, in Europe (and elsewhere where Apple offers parts for self-service repair), you can purchase the sensor from Apple, connect the machine to the internet after replacing it, to then perform the calibration, only if the sensor was purchased from Apple.
So the hardware is capable of performing the calibration, Apple just does not graciously grant you the right to install a recycled or third party sensor in your machine.
https://www.ifixit.com/Answers/View/759262/Torn+Lid+angle+se...
When founders put 996 in their job descriptions or Tweet about their 996 culture it’s a helpful signal to avoid that company.
The only time I’d actually consider crazy schedules was if I was the founder with a huge equity stake and a once in a lifetime opportunity that would benefit from a short period of 996.
For average employees? Absolutely not. If someone wants extraordinary hours they need to be providing extraordinary compensation. Pay me a couple million per year and I’ll do it for a while (though not appropriate for everyone). Pay me the same as the other job opportunities? Absolutely no way I’m going to 996.
In my experience, the 996 teams aren’t actually cranking out more work. They’re just working odd hours, doing a little work on the weekends to say they worked the weekend, and they spend a lot of time relaxing at the office because they’re always there.
I strongly believe LIDAR is the way to go and that Elon's vision-only move was extremely "short-sighted" (heheh). There are many reasons but that drives it home for me multiple times a week is that my Tesla's wipers will randomly sweep the windshield for absolutely no reason.
This is because the vision system thinks there is something obstructing its view when in reality it is usually bright sunlight -- and sometimes, absolutely nothing that I can see.
The wipers are, of course, the most harmless way this goes wrong. The more dangerous type is when it phantom-brakes at highway speeds with no warning on a clear road and a clear day. I've had multiple other scary incidents of different types (swerving back and forth at exits is a fun one), but phantom braking is the one that happens quasi-regularly. Twice when another car was right behind me.
As an engineer, this tells me volumes about what's going on in the computer vision system, and it's pretty scary. Basically, the system detects patterns that are inferred as its vision being obstructed, and so it is programmed to brush away some (non-existent) debris. Like, it thinks there could be a physical object where there is none. If this was an LLM you would call it a hallucination.
But if it's hallucinating crud on a windshield, it can also hallucinate objects on the road. And it could be doing it every so often! So maybe there are filters to disregard unlikely objects as irrelevant, which act as guardrails against random braking. And those filters are pretty damn good -- I mean, the technology is impressive -- but they can probabistically fail, resulting in things that we've already seen, such as phantom-braking, or worse, driving through actual things.
This raises so many questions: What other things is it hallucinating? And how many hardcoded guardrails are in place against these edge cases? And what else can it hallucinate against which there are no guardrails yet?
And why not just use LIDAR that can literally see around corners in 3D?
To be very clear on this point - this is not related to model training.
It’s important in the fair use assessment to understand that the training itself is fair use, but the pirating of the books is the issue at hand here, and is what Anthropic “whoopsied” into in acquiring the training data.
Buying used copies of books, scanning them, and training on it is fine.
Rainbows End was prescient in many ways.
You think it is the first half. You’ll realize later on that really you need a totally different structure to write the book. You’ll try to refactor the whole thing, but some obstinate subset of the original readers will insist on using the original index. In the interest of backwards compatibility, you’ll jam the new structure into the margins (so they can keep their precious index).
It’s text-debt.
If you're going to use SQLite as an application file format, you should:
1. Enable the secure_delete pragma <https://antonz.org/sqlite-secure-delete/> so that when your user deletes something, the data is actually erased. Otherwise, when a user shares one of your application's files with someone else, the recipient could recover information that the sender thought they had deleted.
2. Enable the options described at <https://www.sqlite.org/security.html#untrusted_sqlite_databa...> under "Untrusted SQLite Database Files" to make it safer to open files from untrusted sources. No one wants to get pwned when they open an email attachment.
3. Be aware that when it comes to handling security vulnerabilities, the SQLite developers consider this use case to be niche ("few real-world applications" open SQLite database files from untrusted sources, they say) and they seem to get annoyed that people run fuzzers against SQLite, even though application file formats should definitely be fuzzed. https://www.sqlite.org/cves.html
They fail to mention any of this on their marketing pages about how you should use SQLite as an application file format.
> a shared ledger implemented with traditional databases / protocols would be faster, easier, and more transparent.
Stablecoin is not a technology. It's an excuse. An excuse to do what banks do while not being regulated like a bank or using the infrastructure banks use. Similar to how Airbnb is not a technology but an excuse to do what hotels do without hotel's license.
So it makes no sense to compare it to database, a technology.
Will this excuse work? Banking is a heavily regulated field so it's less likely than Airbnb, but it's ultimately up to lawmakers.
It sounds great, but every time I see this argument, I end up going down the rabbit hole of actually studying how stablecoins operate. And every time, I come to the same conclusion: they always rely on trust in an off-chain oracle or custodian. At that point, a shared ledger implemented with traditional databases / protocols would be faster, easier, and more transparent.
Bitcoin (and possibly a few others) is one of the few uses of blockchain that actually makes sense. The blockchain serves the currency, and the currency serves the blockchain. The blockchain exists to provide consensus without needing to trust any off-chain entity, but the blockchain relies on computing infrastructure that has real-world costs. The scarcity of Bitcoin (the currency) and arguably-fictitious reward for participation in mining is the incentive for people in the real world to contribute resources required for the blockchain to function.
Any real-world value given to Bitcoin is secondary and only a result of the fact that (1) mining infrastructure has a cost, and (2) people who understand the system have realized that, unlike fiat, stablecoins, or 1000 other crypto products, Bitcoin has no reliance on trusted, off-chain entities who could manipulate it.
You trust your stablecoin's issuer that they hold enough fiat in reserve to match the coin? You might as well trust your bank, but while you're at it, remind them that they don't have to take days to process a transaction - they could process transactions as fast as (actually faster than) a blockchain. But I imagine most banks would point to regulation as a reason for the delays, and they might be right.
So what are stablecoins really trying to do? Circumvent regulation? Implement something the banks just aren't willing to do themselves?
There are lots of crypto skeptics on HN (and we ourselves were disappointed with crypto's payments utility for much of the past decade), so it might be interesting to share what changed our mind over the past couple of years: we started to notice a lot of real-world businesses finding utility in stablecoins. For example, Bridge (a stablecoin orchestration platform that Stripe acquired) is used by SpaceX for managing money in long-tail markets. Another big customer, DolarApp, is providing banking services to customers in Latin America. We're currently adding stablecoin functionality to the Stripe dashboard, and the first user is an Argentinian bike importer that finds transacting with their suppliers to be challenging.
Importantly, none of these businesses are using crypto because it's crypto or for any speculative benefit. They're performing real-world financial activity, and they've found that crypto (via stablecoins) is easier/faster/better than the status quo ante.
There has been this trend recently of calling Wikipedia the last good thing on the internet.
And i agree its great, i spend an inordinate amount of my time on Wikimedia related things.
But i think there is a danger here with all these articles putting Wikipedia too much on a pedestal. It isn't perfect. It isn't perfectly neutral or perfectly reliable. It has flaws.
The true best part of Wikipedia is that its a work in progress and people are working to make it a little better everyday. We shouldn't lose sight of the fact we aren't there yet. We'll never be "there". But hopefully we'll continue to be a little bit closer every day. And that is what makes Wikipedia great.
It's the internet. When you talk to people online, it often descends into pettiness. When you talk to people in the real world, that rarely happens. But it's much easier to talk online, so people get the wrong impression.
You should talk to strangers. It's never gone wrong for me. Most people have a warmth and agreeableness that comes out when you are there with them, talking about stuff. There's also the interesting effect that people will give you their innermost secrets, knowing you won't tell anyone (I actually met a serial killer who did this, heh). For instance I was on a long haul flight earlier this year, and my neighbour told me everything about her divorce. Like a kind of therapy.
I also find when I have a real disagreement with someone, it's a lot easier when you're face-to-face. For instance, I have friends who are religious, in a real way, ie they actually think there's a god who created the earth and wants us to live a certain way. Being there in person keeps me from ridiculing them like I might on an internet forum, but it also keeps them from condemning me to hell.
So folks, practice talking to people. Much of what's wrong in the current world is actually loneliness, having no outlet for your expressions.
The 2022 drop for SWE is easy for me to explain, and it's not on these analysts' list of factors (though I'm not an economic quant, I don't know how you could really control for it): In 2017, a tax bill was passed that cut a particular tax incentive in 2022 in an effort to be counted as "revenue neutral" despite being otherwise a massive tax cut overall. The incentive in question was a writeoff for "Research and development". This means that in 2022, it got effectively much more expensive to hire anyone who falls under that category, including developers not directly necessary for the day-to-day function of a business (hell, one might argue they would have counted anyway) and scientists of most kinds. That this hit big firms, which have a higher relative amount of R&D efforts going at a given time, first makes a lot of sense.
For customer service, my explanation is that companies literally do not care about customer service. Automated phone trees, outsourced call centers whose reps have no real power to help a customer, and poorly-made websites have been frustrating people for decades, but businesses never seem to try to compete on doing better at it. It's a cheap win with investors who want to hear about AI initiatives to lay off yet even more of this department, because it doesn't matter if the quality of service declines, there are no market or regulatory forces that are punishing this well enough to ever expect firms to stop breaking it, let alone fix it
These claims wouldn't matter if the topic weren't so deadly serious. Tech leaders everywhere are buying into the FOMO, convinced their competitors are getting massive gains they're missing out on. This drives them to rebrand as AI-First companies, justify layoffs with newfound productivity narratives, and lowball developer salaries under the assumption that AI has fundamentally changed the value equation.
This is my biggest problem right now. The types of problems I'm trying to solve at work require careful planning and execution, and AI has not been helpful for it in the slightest. My manager told me that the time to deliver my latest project was cut to 20% of the original estimate because we are "an AI-first company". The mass hysteria among SVPs and PMs is absolutely insane right now, I've never seen anything like it.
At this rate it’s better to start a company and get aquihired vs applying and getting hired.
Seems like OpenAI speed ran through the Facebook phase and are out of ideas
What it actually looks like: https://fonts.google.com/specimen/B612
Anecdote here, but when I was in grad school, I was talking to a PhD student i respected a lot. Whenever he read a paper, he would try to write the code out and get it working. I would take a couple of months but he could whip it up in a few days. He explained to me that it was just practice and the more you practice the better you become. He not only coded things quickly, he started analyzing papers quicker too and became really good at synthesizing ideas, knowing what worked and didn't, and built up a phenomenal intuition.
These days, I'm fairly senior and don't touch code much anymore but I find it really really instructive to get my hands dirty and struggle through new code and ideas. I think the "just tweak the prompts bro" people are missing out on learning.
Utilization is not a lie, it is a measurement of a well-defined quantity, but people make assumptions to extrapolate capacity models from it, and that is where reality diverges from expectations.
Hyperthreading (SMT) and Turbo (clock scaling) are only a part of the variables causing non-linearity, there are a number of other resources that are shared across cores and "run out" as load increases, like memory bandwidth, interconnect capacity, processor caches. Some bottlenecks might come even from the software, like spinlocks, which have non-linear impact on utilization.
Furthermore, most CPU utilization metrics average over very long windows, from several seconds to a minute, but what really matters for the performance of a latency-sensitive server happens in the time-scale of tens to hundreds of milliseconds, and a multi-second average will not distinguish a bursty behavior from a smooth one. The latter has likely much more capacity to scale up.
Unfortunately, the suggested approach is not that accurate either, because it hinges on two inherently unstable concepts
> Benchmark how much work your server can do before having errors or unacceptable latency.
The measurement of this is extremely noisy, as you want to detect the point where the server starts becoming unstable. Even if you look at a very simple queueing theory model, the derivatives close to saturation explode, so any nondeterministic noise is extremely amplified.
> Report how much work your server is currently doing.
There is rarely a stable definition of "work". Is it RPS? Request cost can vary even throughout the day. Is it instructions? Same, the typical IPC can vary.
Ultimately, the confidence intervals you get from the load testing approach might be as large as what you can get from building an empirical model from utilization measurement, as long as you measure your utilization correctly.
By the way, a pet peeve of mine right now is that reporters covering court cases (and we have so many of public interest lately) never seem to simply paste the link to the online PDF decision/ruling for us all to read, right in the story. (and another user here kindly did that for us below: https://storage.courtlistener.com/recap/gov.uscourts.dcd.223... )
It seems such a simple step (they must have been using the ruling PDF to write the story) yet why is it always such a hassle for them to feel that they should link the original content? I would rather be able to see the probably dozens of pages ruling with the full details rather than hear it secondhand from a reporter at this point. It feels like they want to be the gatekeepers of information, and poor ones at that.
I think it should be adopted as standard journalistic practice in fact -- reporting on court rulings must come with the PDF.
Aside from that, it will be interesting to see on what grounds the judge decided that this particular data sharing remedy was the solution. Can anyone now simply claim they're a competitor and get access to Google's tons of data?
I am not too familiar with antitrust precedent, but to what extent does the judge rule on how specific the data sharing need to be (what types of data, for what time span, how anonymized, etc. etc.) or appoint a special master? Why is that up to the judge versus the FTC or whoever to propose?
This is an astonishing victory for Google, they must be very happy about it.
They get basically everything they want (keeping it all in the tent), plus a negotiating position on search deals where they can refuse something because they can't do it now.
Quite why the judge is so concerned about the rise of AI factoring in here is beyond me. It's fundamentally an anticompetitive decision.
> Amazon argued that the class was too large to be manageable
Sorry, we've wronged too many people to be held accountable! What a wild argument.
The whole LLM era is horrible. All the innovation is coming "top-down" from very well funded companies - many of them tech incumbents, so you know the monetization is going to be awful. Since the models are expensive to run it's all subscription priced and has to run in the cloud where the user has no control. The hype is insane, and so usage is being pushed by C-suite folks who have no idea whether it's actually benefiting someone "on the ground" and decisions around which AI to use are often being made on the basis of existing vendor relationships. Basically it's the culmination of all the worst tech trends of the last 10 years.
The issue is that American media/discourse paints a very distorted view of what life under authoritarian rule is like. The truth is in many countries, unless you’re some kind of minority, politically active, or in legal trouble, day-to-day life is mostly similar to life in the west. But people don’t want to hear that, because we want to feel better than them. Like we wouldn’t tolerate that kind of life.
Of course the most frustrating part about that is as the US and other western countries start sliding into authoritarianism, people deny it because they don’t feel like it’s authoritarian.
Edit: To clarify, I don’t think life is exactly the same - just that the consequences of authoritarianism are much more insidious than they’re portrayed.
The author gets close to what I think the root problem is, but doesn't call it out.
The truth is that in python, async was too little, too late. By the time it was introduced, most people who actually needed to do lots of io concurrently had their own workarounds (forking, etc) and people who didn't actually need it had found out how to get by without it (multiprocessing etc).
Meanwhile, go showed us what good green threads can look like. Then java did it too. Meanwhile, js had better async support the whole time. But all it did was show us that async code just plain sucks compared to green thread code that can just block, instead of having to do the async dances.
So, why engage with it when you already had good solutions?
> Then Pharaoh also called for the wise men and the sorcerers, and they also, the magicians of Egypt, did the same with their secret arts. For each one threw down his staff and they turned into serpents. But Aaron’s staff swallowed up their staffs.
- Exodus 7:1-12 (NIV)
Many moons ago I had a girlfriend who worked on an nationally broadcast afternoon show where they often had guest chefs demonstrating dishes, so I would come home from my thankless PhD work to eat Michelin-starred food from a lunchbox. Overall not so bad.
The compute moat is getting absolutely insane. We're basically at the point where you need a small country's GDP just to stay in the game for one more generation of models.
What gets me is that this isn't even a software moat anymore - it's literally just whoever can get their hands on enough GPUs and power infrastructure. TSMC and the power companies are the real kingmakers here. You can have all the talent in the world but if you can't get 100k H100s and a dedicated power plant, you're out.
Wonder how much of this $13B is just prepaying for compute vs actual opex. If it's mostly compute, we're watching something weird happen - like the privatization of Manhattan Project-scale infrastructure. Except instead of enriching uranium we're computing gradient descents lol
The wildest part is we might look back at this as cheap. GPT-4 training was what, $100M? GPT-5/Opus-4 class probably $1B+? At this rate GPT-7 will need its own sovereign wealth fund
There is no such thing as "the best engineers." Some engineers are definitely better than others, but once you pass the bar of "really smart, great work ethic," the tech tree diverges pretty dramatically.
Some engineers (like Notch) are amazing at quickly putting out vast quantities of mediocre code, prototyping ideas, maintaining a clear product vision, and bringing something into reality quickly. Other engineers (like John Carmack) are great at generating well-founded opinions and finding clever solutions to difficult issues. Some engineers (like Bill Atkinson) worked mostly remotely and developed amazing technology, while other engineers (like Joel Spolsky) insisted on in-office and built a best-in-class mentorship organization.
While hiring people with exceptional talent is a step-change when it comes to any organization's ability to accomplish its goals, there is no one metric for "best." Much better to identify the specific skills for which you need exceptional talent, and to create a hiring funnel that identifies people who excel in that dimension.
I 100% agree. I've ran into the same issues, and I would never use Next.js for anything, and I will encourage every team at work to use something else.
In general Next.js has so many layers of abstraction that 99.9999% of projects don't need. And the ones that do are probably better off building a bespoke solution from lower level parts.
Next.js is easily the worst technology I've ever used.
> the stance of the 'Open Source' crowd
The original stance of the open source crowd was more along the lines of the GPL -> GPLv3 -> AGPL, which expressly prevents this kind of thing.
The proliferation of "give everything away for free" MIT/BSD/Apache licenses seems to me to have been an intentional campaign by corporate interests to undermine free software ideals