HN user
If you have an endless pattern of ..., -1, 1, -1, 1, -1, 1, ... and run box blur with a window of 2 or 4, you get ..., 0, 0, 0, 0, 0, 0, ... too.
Other than that, you're not wrong about theoretical Gaussian filters with infinite windows over infinite data, but this has little to do with the scenario in the article. That's about the information that leaks when you have a finite window with a discrete step and start at a well-defined boundary.
I think the US is markedly better for people in certain professions who want to become rich. This is obviously not true for the general population, but the amount of cash that's chasing profits in tech means that a non-trivial percentage of SF Bay Area techies have a shot at financial independence (let's say, $10M in the bank). This is certainly not true for IT workers in most of the EU.
But if you already have enough to never need to work again, you should be fine in almost any liberal and politically stable country, and there's something to be said about moving to a lower CoL place where you can afford a nicer home, etc.
They don't. I'm using Cloudflare and 90%+ of the traffic I'm getting are still broken scrapers, a lot of them coming through residential proxies. I don't know what they block, but they're not very good at that. Or, to be more fair: I think the scrapers have gotten really good at what they do because there's real money to be made.
Maybe I'm getting hung up on words, but my beef is with the parent saying they find real numbers "completely natural".
It's a reasonable assumption that the universe is computable. Most reals aren't, which essentially puts them out of reach - not just in physical terms, but conceptually. If so, I struggle to see the concept as particularly "natural".
We could argue that computable numbers are natural, and that the rest of reals is just some sort of a fever dream.
35 years ago, a good chunk of the current EU was under a Soviet-imposed totalitarian rule. Spain was a dictatorship until 1975. And it's been just 80 years since WWII.
It always boggles my mind that most Europeans are absolutely convinced that nothing like that could ever happen again. Meanwhile, many people in the US are convinced that the government will be coming for them any minute now.
I believe real numbers to be completely natural,
Most of real numbers are not even computable. Doesn't that give you a pause?
Because inexplicably, there's random pixel-level noise baked into the blue area. You can't see it unless you crank up contrast, but it makes the bitmap hard to compress losslessly. If you remove it using threshold blur, it doesn't change the appearance at all, but the size is down to 100 kB. Scale it down to a more reasonable size and you're down to 50 kB.
Modern web development never ceases to amaze me.
Meh. The room-temperature endurance of modern EEPROMs (e.g., ST M95256) is something like 4 million cycles. If you use a simple ring buffer (reset on overflow, otherwise just appending values), you only need to overwrite a cell once every 32k ticks, which gives you a theoretical run time of 250,000 years with every-minute updates or 4,100 years with every-second updates.
The history of journalism is written by journalists, often in a self-serving way. You'll be hard-pressed to pinpoint the purported golden age of impartial truth-seeking. Early newspapers in the US were often owned by a local railroad tycoon and published hit pieces about his opponents. From the 1960s, this morphed into a way to broadcast the ideological consensus of East Coast Ivy League graduates. Some of their ideas were good and some were bad, but every single day, this consensus influenced which stories made it to the front page and how they were framed.
Weirdly, I think this model was beneficial even in the presence of bias: when everyone read the same news, it helped with social cohesion and national identity, even if the stories themselves presented a particular viewpoint.
But now, everyone can get their own news with their own custom-tailored bias, so there's no special reason to sign up for the biases of Washington Post or The New York Times unless you want to signal something to your ingroup. I don't think this is as much Bezos' fault as it's just a consequence of the internet evolving into what it is right now: one giant, gelatinous cube of engagement bait.
But gcc is part of it's training data so of course it spit out an autocomplete of a working compiler /s
Why the sarcasm tag? It is almost certainly trained on several compiler codebases, plus probably dozens of small "toy" C compilers created as hobby / school projects.
It's an interesting benchmark not because the LLM did something novel, but because it evidently stayed focused and maintained consistency long enough for a project of this complexity.
They're not coming from anywhere. It's an LLM-written article, and given how non-specific it is, I imagine the prompt wasn't much more than "write an article about how OpenClaw is changing my life".
And the fact this post has 300+ comments, just like countless LLM-generated articles we get here pretty much daily... I guess proves the point in a way?
It's common for compilers to generate mildly unusual code because they translate high-level code into an abstract intermediate notation, run a variety optimization steps on that notation, and then emit machine-specific code to perform whatever the optimizations yielded. There's no constraint along the lines of "but select the most logical opcode for this task".
The claim that the code is inefficient is really not substantiated well in this blog post. Sometimes, long-winded assembly actually runs faster because of pipelining, register aliasing, and other quirks. Other times, a "weird" way of zeroing a register may actually take up less space in memory, etc.
Gun frames can be made out of plastic or aluminum, and there are fixtures for benchtop CNC machines that can be used to make them. This is not nearly as complicated as you make it sound. I think Cody Wilson was basically selling a turnkey solution for that, maybe still is.
There is a database. The insurance covers things that aren't in the database. Claims are exceptionally rare, so it's pretty cheap.
Different people understand "theft of identity" in different ways. If someone is impersonating you on the internet, or steals your credit card info and makes purchases on your behalf, that probably qualifies.
As for the nature of the scam, there are different levels of this. Most likely, the mark is the buyer / the escrow agency.
First, you're getting upset at a random person on the internet for expressing their political views. Second, your objection almost certainly has nothing to do with this attack. It targeted some specific subset of users of Notepad++, not the maintainer.
That's the most honest assessment you can expect from any small-scale developer. What do you expect them to say or do? Their adversary is presumably a national intelligence agency of a superpower.
The odds may be better if you operate the way OpenSSH does: move slow, security first, architect everything to be very difficult to attack. But if you're building a text editor, it's not your mindset, and probably never will be.
Is this materially different than giving all files on your system 777 permissions?
Yes, because I can't read or modify your files over the internet just because you chmod'ed them to 777. But with Clawdbot, I can!
It might be an interesting LLM benchmark: how many can they list without breaking the rules (repetition or non-animals). Although I bet that big bucks would be then thrown at pointlessly optimizing for that benchmark, so...
That's not what I'm saying. I mean citations that aren't citations: a "source" that doesn't discuss the topic at all or makes a different claim.
FWIW, this is a fairly common problem on Wikipedia in political articles, predating AI. I encourage you to give it a try and verify some citations. A lot of them turn out to be more or less bogus.
I'm not saying that AI isn't making it worse, but bad-faith editing is commonplace when it comes to hot-button topics.
Silver has plenty of industrial uses.
About one third of this demand (photographic film and paper) more or less evaporated in the 2000s. You don't see that on the price chart, so I don't think you can seriously argue that the price is dictated by industrial uses.
It's more complex than that. It's called SynthID-text and biases the probabilities of token generation in a way that can be recovered down the line.
No. No one is looking for em-dashes, except for some bozos on the internet. The "default voice" of all mainstream LLMs can be easily detected by looking at the statistical distribution of word / token sequences. AI detector tools work and have very low false negatives. They have some small percentage of false positives because a small percentage of humans pick up the same writing habits, but that's not relevant here.
The "humanizer" filters will typically just use an LLM prompted to rewrite the text in another voice (which can be as simple as "you're a person in <profession X> from <region Y> who prefers to write tersely"), or specifically flag the problematic word sequences and ask an LLM to rephrase.
They most certainly don't improve the "correctness" and don't verify references, though.
Honestly, with all my love for the HN community, I think we have a couple of topics that just get upvoted without reading because they signal that you're in the ingroup. Few years back, another reliably upvoted thing was anything with "Bayesian" in the name. In the past couple of years, "busy beavers" would also get upvotes even though they have no practical use, their mathematical significance is dubious, and few people understand them in the first place.
Also, as a website owner, these residential proxies are a real pain. Tons and tons of abusive traffic, including people trying to exploit vulnerabilities and patently broken crawlers that send insane numbers of requests, and no real way to block it.
It's just nasty stuff. Intent matters, and if you're selling a service that's used only by the bad guys, you're a bad guy too. This is not some dual-use, maybe-we-should-accept-the-risks deal that you have with Tor.
I have a CNC mill made in 2006. It's still perfectly fine. It should still be fine in 2036. The most significant threat to its existence is the compatibility of OS drivers and software support in CAM tools. That and USB ports getting replaced by something else, which was a problem for earlier-generation machines that used RS-232.
Part of me wants to be wary. The useful life of industrial machinery such as CNC mills is much longer than the lifespan of websites, so locally-installed software you own is usually a better choice.
But another part of me realizes that everyone is using Fusion360, despite the fact they have a history of taking away features to force people to migrate to paid tiers. So it probably doesn't matter.
I think you might be underestimating the level of control that an average parent, especially a working parent, has over a teenage kid. Short of taking away devices, it's tough, especially if they're going through a phase of doing precisely the opposite of what you recommend / demand.
I'm not saying that parents don't have any responsibility, but it's about practicalities. If a teenager can easily buy smokes or alcohol, many will, no matter what the parents say. If you make the goods harder to buy, usage drops. So, shops / software vendors do have some responsibility for societal outcomes.
In a libertarian utopia, anything goes, but kids are... weird in that they often try to push the boundaries of their autonomy without always knowing the risk, and it's in our collective best interest not to let them go too far.
economic failure to build cheaper purpose-built hardware for things like sending audio over a radio.
You're literally just wasting sand. We've perfected the process to the point where it's inexpensive to produce tiny and cheap chips that pack more power than a 386 computer. It makes little difference if it's 1,000 transistors or 1,000,000. It gets more complicated on the cutting edge, but this ain't it. These chips are probably 90 nm or 40 nm, a technology that's two decades old, and it's basically the off-ramp for older-generation chip fabs that can no longer crank out cutting-edge CPUs or GPUs.
Building specialized hardware for stuff like that costs a lot more than writing software that uses just the portions you need. It requires deeper expertise, testing is more expensive and slower, etc.