I wish cultural norms around documentation would shift to "pull" rather than "push" — generating "views" of organized knowledge on the fly instead of making endless rearrangements of the same information. It's become too cheap in terms of proof of (mental) work to spray endless pages of notes, reports, memos, decks, etc. but the "documentation is good" paradigm hasn't caught up yet.
Ideally AI would minimize excessive documentation. "Core knowledge" (first principles, human intent, tribal knowledge, data illegible to AI systems) would be documented by humans, while AI would be used to derive everything downstream (e.g. weekly progress updates, changelogs). But the temptation to use AI to pad that core knowledge is too pervasive, like all the meaningless LLM-generated fluff all too common in emails these days.
There is definitely going to be some Wirth's law-like [0] effect about the asymmetry of software complexity outpacing LLMs' abilities to untangle said software. Claude 9.2 Optimus Prime might be able to wrangle 1M LoC, but somehow YC 2035 will have some Series A startup with 1B+ LoC in prod — we'll always have software companies teetering on the very edge of unmaintainability.
It's the Peter principle for computers. Codebases expand to the limits of the organization's ability to manage them. If you make one person use ed to write code for a bare metal environment, you'll get a comparatively small, laser-focused codebase. If you task a hundred modern developers to solve the same problem, you'll get a Linux box device running a million lines of JavaScript.
Same thing happens in other fields. A rich country and a poor country might build equivalent roads, but they won't pay the same price for them.
> This would be like an NFL team drafting a quarterback based on how many instagram followers they have rather than a relevant metric like pass completion, or god forbid, doing some work and actually scouting candidates. Maybe the Cleveland Browns would do that
Not quite the same, but the New York Jets (one of the few NFL teams that can match the dysfunction of the Browns — they have the longest active playoff drought in big 4 North American sports) passed on a few successful players because the owner, Woody Johnson, reportedly didn't like their Madden (video game) ratings [0]:
> A few weeks later, Douglas and his Broncos counterpart, George Paton, were deep in negotiations for a trade that would have sent Jeudy to the Jets and given future Hall of Fame quarterback Aaron Rodgers another potential playmaker. The Broncos felt a deal was near. Then, abruptly, it all fell apart. In Denver’s executive offices, they couldn’t believe the reason why.
> Douglas told the Broncos that Johnson didn’t want to make the trade because the owner felt Jeudy’s player rating in “Madden NFL,” the popular video game, wasn’t high enough, according to multiple league sources. The Broncos ultimately traded the receiver to the Cleveland Browns. Last Sunday, Jeudy crossed the 1,000-yard receiving mark for the first time in his career.
...
> Johnson’s reference to Jeudy’s “Madden” rating was, to some in the Jets’ organization, a sign of Brick and Jack’s influence. Another example came when Johnson pushed back on signing free-agent guard John Simpson due to a lackluster “awareness” rating in Madden. The Jets signed Simpson anyway, and he has had a solid season: Pro Football Focus currently has him graded as the eighth-best guard in the NFL.
If you like space toilets you may be interested in this [0] HN submission where you can track how full the ISS urine tank is from the comfort of your menu bar.
Dude, I just looked a picture of old socks and urine stuffed together in a waste bag. I think I need some time before I can feel the delight to get interested in any kind of toilet.
That's the one, looks like I had some details muddled (it's a book club not a library, and so the fee is for the book which was in fact returned but perhaps lost in the post) but the outline and relevance here exactly correct. Thanks!
I would have thought so too but Naval Gazing has a short series [0] on why it's not as dire as one might think. An aircraft carrier's location being "secret" in this case is just one layer of the survivability onion [1] anyhow. (Caveat that as someone who takes a casual interest in this, I can't vouch for accurate this is at all.)
It is important to note the Naval Gazing article is specifically talking about the difficulties of actually targeting a ship for a successful kill rather than just tracking it. It's in response to the idea that satellites plus missiles would mean carriers could be instantly destroyed in a first round of hostilities with a sufficiently prepared opponent. Tracking is a lot easier to do than getting data fresh and precise enough to hit the ship with no other tools (eg ships already nearby that can get a live precise track vs terminal detection and guidance on the missile itself).
Also the capabilities of commercial and government geospatial systems has only continued to improve in the ~decade since the article was written.
It also seems worth considering that the article's view that "spending a lot of time searching for the carrier is a good way to get killed by defending fighters" is a distinctly pre-drone-ubiquity assumption.
Can a carrier group's point defense weapons and fighters reliably counter a swarm of hundreds of cheap drones, flying lower than cloud cover, that are programmed to look for carriers over a wide area, confirm their shape optically, paint them for missiles, and take the disconnection/destruction of any one of them as an indication of possible activity and automated retasking? It's a scary world to be a slow-moving vehicle, these days.
That's why standard carrier doctrine is to stand off from shore, out of range of cheap missiles and drones. To strike a carrier, an adversary would need large, expensive missiles or drones plus an effective detection and targeting system.
Couldn't they just send a boat/plane/balloon/zepplin with a charger on it out launch the drones from there. The would come back when low on power and recharge in waves. It took me 30 seconds to think of this so I am sure there are a lot of better ideas out there already.
> Can a carrier group's point defense weapons and fighters reliably counter a swarm of hundreds of cheap drone
Hundreds of cheap drones would have negligible impact on a modern warship's integrity. An aircraft carrier is designed to have an actual airplane crash into it and continue operating. These boats still have armor. It's not purely an information war.
Less than $20 million each - assuming build capacity and plans ...
High-Altitude Pseudo-Satellites Are Ready for Launch (2023)
Editor's note: [ ... ] Airbus contacted Proceedings to note that the 2016 pricing estimates were correct at the time but that the company will be releasing new, lower estimates soon.
After an astounding 64 days aloft and a travelling a total more than 30,000nm, a British-built solar-powered UAV crashed just hours before it was due to break the ultimate world endurance record.
The aircraft was the British-built solar-powered Airbus Zephyr UAV – one of a new breed of HAPS (high altitude, pseudo-satellites) – a new category of UAVs that are aiming for zero-emission, ultra-long- endurance flight as a kind of terrestrial satellite – able to loiter in the stratosphere for weeks or months at a time to monitor borders, watch shipping, relay communications or conduct atmospheric science.
Not viable for a non-superpower to create and deploy a successful hostile drone, and imagination is cheaper than reality anyway.
Aside from the hostile drone command and launch being found and destroyed, the drone itself would either be shot down by a missile or disabled by a direct energy weapon.
If the drone were to fault on it's own and was designed to float, it will be expensive to retreive it. Cheaper at scale to launch the sensors into orbit or deploy bouys.
interceptors are much shorter range than attack/scouting drones because they need to go a lot faster and be more manuverable than the target they are intercepting. Cameras are cheap and really light compared to ordinance, and ziplime was able to make a fleet of fairly cheap drones with 200 mile range (as a private company a decade ago). Cheap drones definitely can maintain targeting of a carrier within a couple hundred miles of the coast (and if you can get to 5-600 miles you keep most carrier based aircraft out of range of your shores)
Not hidden from nation states with access to real-time satellite imagery, but more rustic guerilla operations usually don't have such sophisticated access
Actually probably even cheaper, a generic scan to spot all the ships, and when it's done, just need to get images around the last location. Probably can use something like the Planet API
Did you really get so salty by my comment (https://news.ycombinator.com/item?id=47423992) that now you just have to spam HN with the same? Suck it up and move on, healthier for everyone.
I didn't want to nitpick terminology, but yes, the tile-placement algorithm here is just a way of solving constraint satisfaction problems with DFS using a "minimum remaining values" heuristic [0]. The original use case for generating textures [1] is different in that the constraints are implicit in the input bitmap, but this project is a more straightforward tile placement with explicit constraints.
I think this algorithm is more efficient for generating maps with only local (adjacency) constraints, but setting this up as an integer linear program and plugging it into a constraint solver is more generalizable (say, if you wanted to enforce a constraint that rivers had to flow across the whole map and could not loop).
But I agree "wave function collapse" is not really the best name, for two reasons:
- the original repository mentions "it doesn't do the actual quantum mechanics, but it was inspired by QM", but it implies something QM-related.
- as an ORIE major in college that loved optimization, I think constraint satisfaction problems are really cool and actually somewhat approachable! So calling the heuristic something else like "wave function collapse" might limit people from finding previous work and known improvements (e.g. forward checking).
Ideally AI would minimize excessive documentation. "Core knowledge" (first principles, human intent, tribal knowledge, data illegible to AI systems) would be documented by humans, while AI would be used to derive everything downstream (e.g. weekly progress updates, changelogs). But the temptation to use AI to pad that core knowledge is too pervasive, like all the meaningless LLM-generated fluff all too common in emails these days.
reply