Because, we have pretty convincing historical precedent that 'just following orders' does not work as a defense when your government does something indefensible.
To be fair, Coq has ProofGeneral and Agda has its emacs mode. Once you go outside these established channels, oftentimes using the tool becomes incredibly difficult. I guess for interactive theorem proving in general you may need some sort of editor at some point.
Yeah, I'm not a fan of the encouragement to use vscode; that said it was pretty easy for me to get neovim set up with Lean tooling, and that's what I use generally.
Otherland by Tad Williams, where the powerful oligarchs of corporations are kept bodily alive bybmachinery, their mind have been transferred to a cirtual reality and they keep on going by making said virtual reality the next hypercapitalist venture.
And yet we are discussing this in the context of a reporter having been fired from Ars Technica for publishing an article which included inaccurate LLM-generated summaries in 2026. How come?
Given that the ones that surfaced on the frontpage were pretty interesting, vibe coded or not, I’d say the voting mechanism is working as a good filter.
Interesting? I'd say they were interesting if you find looking at vibe-coded stuff interesting. If you're instead into learning from projects based on the author's unique insight, experience and research, they're utterly boring...
I find that I just don't learn anything new from Show HN vibe-coded side projects, and I can often replicate them in a couple of hundred of dollars, so why bother looking at them? Also why bother sharing one in the first place, since it doesn't really show any personal prowess, and doesn't bring value to the community due to it being easy to replicate?
> Interesting? I'd say they were interesting if you find looking at vibe-coded stuff interesting.
There's a lot of ways things can be of interest. The problem being solved, how it's being solved, the UI, UX, etc.
THAT it is vibe coded may or may not be interesting to some, but finding it un-interesting because it's vibe coded is no better than finding that it is.
This assumes that pre-LLM projects were based on the author's unique insight, experience and research, and not just boilerplated framework code, copying the design trends of the week.
I'd challenge the lack of personal prowess argument. Piecing together technology in novel ways to solve highly targeted problems is a skill, even if you're not hand-crafting CSS and SQL.
I liken it to those who tune cars, who buy cars made in a factory, install parts made by someone else, using tools that are all standardized. In the middle somewhere is the human making decisions to create a final result, which is where the talent exists.
I agree that some (many) pre-LLM Show HN projects were worthless as well. But at least they were fewer, which meant that interesting projects were harder to miss.
> Piecing together technology in novel ways to solve highly targeted problems is a skill
The LLM outputs this out of the box? Where's the skill?
I don't believe the comparison to car tuners benefits your thesis here. The spectrum of people I know who tune their cars varies from utter idiots to professional engineers. You cannot state as a fact that anyone who does it has insight or even natural talent. The bar is so low that anyone who has enough money can do it (just like coding with LLMs). In fact one can say that most people are incompetent, and by tuning their cars to varying degrees they endanger themselves and others, enlarge their running/maintenance costs, lower their car's resale value, and harm the environment.
Yes, I find looking at vibe coded stuff interesting when they solve a worthy problem.
No amount of denial will roll back the technology that millions can use now, that makes it realistic to produce in a day software that would take at least months five years ago.
The primary purpose of something like the F-35 program is not producing a bunch of jets that we can use to win wars. Similar to how NASA's purpose is not to make large rockets that send things to orbit for cheap.
It is to investigate new technologies (i.e. how do we control a thousand drones) and preserve domain knowledge in a large number of engineers spanning multiple generations. If all these engineers go work at $BIG_TECH optimizing ad revenue for watching short videos, we'll have to rediscover basics the next time.
When we have to fight the next serious war, we are not going to primarily use F-35 jets built twenty years ago, it's going to be something built on a similar platform in larger numbers to specifically address challenges of that era. If it can not be made cheap enough, whatever contractors involved are going to be nationalized. All major wars between comparable powers were fought with technology hot off the assembly lines, not billion dollar prototype models developed twenty years ago to bomb caves in deserts.
If you look at it from this angle, all the idiosyncrasies make sense. There's of course the inefficiency of defense contractors skimming off profits at multiple layers, but if you find a solution to that while preserving productivity, you'd win the economics nobel tomorrow.
> When we have to fight the next serious war, we are not going to primarily use F-35 jets, it's going to be something built on a similar platform in larger numbers to specifically address challenges of that era. If it can not be made cheap enough, whatever contractors involved are going to be nationalized.
That is, to some extent, what the F-35 is; the mass-produced plane that incorporates what we learned from the F-117 and F-22 and whatnot. We've already made 10x as many as the F-22's production run.
In WW2 the US would send a 1,000 bombers to hit a target and still miss. That's why they needed so many. Now a single attack jet can hit multiple targets with very high probability.
Cheap drones are extremely limited in the kinds of targets they can reach and damage while evading air defenses. I understand this domain well.
Upgrading drones so that they have sufficient range and carry a sufficiently capable warhead and have a decent probability of surviving a modern air defense environment has been done many times by many countries. The price always comes in ~$1M/drone. It doesn't matter who builds it. Those economics get expensive fast for a weapon system you can't reuse. Much cheaper drones either have no useful range or are susceptible to even cheaper defenses; in either case they don't do any meaningful damage. That point on the price-performance curve wasn't picked at random by competent weapon designers.
Even the Ukrainian FP-5 is ~$0.5M, and it is significantly less capable than some western weapons with a similar profile.
The US has assumed drone swarm attacks would be a thing for decades and has both tested and fielded many systems purpose-built for those scenarios.
Exactly, drones enable quantity and precision. Geran type drones can easily fly 1000nm, and that kind of range needs wide area sensing and patrols to intercept, really expensive at present.
I don't know that a loss right now would be likely, probably a stalemate which would be ruinously expensive for everyone.
Drones favor defenders by making movement costly, there is a considerable advantage to being dug in. Air dominance no longer guarantees being free from low altitude aerial threat. Long range drones require basing further away, which means A2A refuelling, or a massive innovation in drone defence (cheap missiles, autonomous drone interceptors, sensor nets).
> If China produces 100 times or 1000 times their current numbers (and they can), marginal differences in capability are not going to matter.
If china somehow learnes magic and produced 10,000 f16 equivalents and got into a major non-nuclear shooting war with the united states... they'd lose 10,000 planes. At some point there is such a qualitative difference that numbers don't really matter.
You are aware that China is producing two fifth gen stealth fighters, and is flight testing two sixth gen platforms? And that Chinese AAMs are world class? Read up on how Pakistan crushed the Indian Air Force recently flying fourth gen Chinese fighters using their current AAMs.
There is no evidence to show that Pakistan crushed the Indian Air force infact it's the other way around. A lot of Chinese equipment was blown out by Indian Brahmos.
The idea that China would lose 10k fighters when we barely have enough AAMs and aircraft in comparison is silly. The days of F-22s clubbing baby seals is long past.
This is delusional. The PLAAF is a capable force and innovates more quickly than the USAF. Chinese A2A weapons are very good kinetically, and while EW and stealth would have an advantage, engagement geometry means an 4:1 fight is always going to be costly. We could expect significant attrition in EW and stealth advantage over the first few weeks as their RADARs and seekers adapt.
> engagement geometry means an 4:1 fight is always going to be costly
That's not how 6th gen fighter combats work. You get hit by missiles and explode without ever even detecting the opponent.
Does china have better stuff than f16s? Sure (and modern f16s are not the same as 1970s f16s which makes my point harder to understand in the first place anyways) but at some point, with some military technologies, you can't beat them with quantity.
That's how your imagination of 5th fighters work (because there are no 6th gen fighters in service), as if they are somehow invisible. This is a misunderstanding of RADAR.
Lower frequency RADAR will pick up F-35s, but not with enough precision to generate a target track. Pilots spend a lot of thought on the problem of signature management.
A Chinese Wedgetail would be extremely dangerous, as it could provide a very good detection, and with a close enough X-band RADAR you will get a target, and then it is up to kinetic escape/EW/decoys. That is a bad situation to be in during a large force engagement.
The PLAAF is of course working on longer range and faster AEW&C and jam resistant data link and expendable sensors. It is just a matter of time.
I'm not trying to say that f35s are immune to modern chinese technology, I'm trying to say they're immune to previous generation stuff. They don't need to be literally invisible, you can know where they are all you want if you can't actually hit them with a missile.
There's a lot of things that complicate actual fights, people get lucky, people get unlucky, you do clever things like predict incoming sortie directions and saturate that area, but while the f35 is not produced in the same numbers as say, the p51, it's not "the last starfighter" either, the opponent has to get lucky a whole bunch of times before their ability to resist has been removed.
They don't seem to have fared all that well against US/Israeli F-35s in Iran, and seem to struggle in Ukraine against far less sophisticated opponents.
> seem to struggle in Ukraine against far less sophisticated opponents
Yes, that's the recurring theme in modern warfare. Cheap mass produced shit makes sophisticated tech struggle. The US is now going to be immune from this, especially since we have stopped mass producing anything for thirty years.
Ukraine used cheap drones to destroy a few S400 setups.
The previous poster was claiming that not only can the S400s not detect f35s, they can't adequately combat previous generation stuff either.
I can't comment on how accurate that is, but as for "cheap mass produced shit", I refer you to both times the US invaded iraq. Kill ratios on the order of literally 100 to 1. This is the point at which you can't actually overwhelm better technology with quantity.
Wars are, of course, incredibly more complicated than just measuring kill ratios and highscores as if it were a video game. There's usually multiple factions involved, all with differing objectives and those factions and objectives can change and mutate during the course of the war anyways.
If we look at a war the US unequivocally lost, defending south vietnam from the north vietnam invasion, there are a whole bunch of factors but exactly none of them involved some kind of "high tech too advanced" weapon system being defeated by "lower tech but more mass produced" systems.
Doing some incredibly basic research shows that the actual dollar cost of the vietnam war was, relative to the american economy and so on, pretty small, and not a major factor in the ultimate result of the war. Producing fighter jets or whatever that were 10% cheaper doesn't seem like it would have made much of a difference.
This actually kind of proves my point, we can "go to war" with iraq or iran and as long as we're not implementing literal conscription, the US is willing to maintain those wars for a very long time. There's a gigantic difference in public response between "2 pilots died when f35s were ambushed" and "10,000 american soldiers died in this month's battles around Baghdad". One of those gets you some headlines and a few protests, the other gets you voted out of office.
Practically speaking there are between 0 and 1 S-400 systems to worry about, and they will be one of the very first targets. Prompt global strike, JASSM-ER/LR, Tomahawk, AARGM-ER, JATM, even AIM-174B and 260s, they're all coming for it, and all the lesser S-300 (SA-10) / HQ-15, HQ-16/FK-3s etc.
The S-300 systems in Iran were completely ineffective against IDF F-35s. Doubtful they even got a detection.
You fire all your missiles then.. you turn around and leave. The enemy isn't in range to engage you, you're flying away at least the same speed they're chasing, what are they going to do?
The primary purpose of something like the F-35 program is not producing a bunch of jets ... It is to investigate new technologies
I thought the F-22 investigated the technologies and the F-35 is the mass-produced version.
When we have to fight the next serious war ... it's going to be something built on a similar platform in larger numbers to specifically address challenges of that era.
Not if every jet takes 20 years to develop.
If it can not be made cheap enough, whatever contractors involved are going to be nationalized.
Which would accomplish nothing since the rot is so deep.
The F-35 was designed to be a partially-nerfed export version of some of the capabilities in the F-22. It was anticipated that the large production rate would significantly reduce the unit costs, which seems to have panned out. They probably shouldn't have tried to produce three significantly different variations of the same design, since that added materially to the development cost.
The 6th gen platforms appear to be coming in at significantly reduced cost relatively to what they are replacing, which was a major objective.
> I thought the F-22 investigated the technologies and the F-35 is the mass-produced version.
Sure, I'd think of it as a mass^2 produced version then ;)
> Not if every jet takes 20 years to develop.
Think of F-35 variants, not entirely new platforms. If I have to guess, one reduced to a barebones autonomous version built for the purpose to commanding drone swarms and dealing with incoming drone swarms.
Do not try to solve an unsolvable problem, you'll end up hurting real users quite a bit more than you might imagine. Imagine new enthusiastic users trying your platform getting hit with an AI label because of inevitable false positives.
'Detecting AI' is not a problem that has real solutions, the only avenue is something supply side like synthid. But that harms users too, by introducing further barriers for indie users.
I train music generation models. They are very trivial to detect. In fact, detecting them then training them to evade detection by the detection model is a big part of training them! But the detectors win instantly without some hardcore regularization. Simply turn that off and you've instantly got a perfect classifier.
This isn't like text classification, the signal many orders of magnitude higher bitrate and so many more corners need to be cut. It's likely going to be nearly impossible or at least not remotely worth it to generate an audio signal that is truly undetectable in the foreseeable future.
You are right, the output of a model that generates music directly is, for now, easy to categorize as AI.
What this big flux of AI generated music online isn't really that. It'a a tiny bit autogenerated stuff and a whole lot of automatically remixed stuff. The reason it can not be easily classified as AI is because quite a bit of human produced music is also that, and you'd just shut out real users.
Today. Trying to detect AI is like extracting water from puddles in a lake that is quickly drying up. What is the point in the short term if it's impractical in the long term? It will catch some low-hanging fruit in the best case, and will find false positives in the worst.
My point is you should consider creating truly undetectable audio end to end with AI to be effectively impossible for the foreseeable future (i.e., I would bet money it is still trivially detectable five years from now). It won't be detectable to humans, though, only models.
in the broad strokes of ai generated, i wouldnt be so sure.
if the ai picked a bunch of samples and combined them together and mastered using an mcp to a DAW, how is that particularly distinguishable vs a person doing the same thing badly?
i can see how the llm generation pictures of spectrograms is essy to spot, but much less so with tool following.
even worse of you using a vla to have it actually play the guitar and use the recording as a sample.
theres some time and setup to make it happen sure, but somebody put that all in a studio and expose an mcp
reply