Life needs some kind of chemistry that doesn't lock up into compounds so stable they're hard to crack apart, but allows compounds stable enough to build structures.
Carbon, hydrogen, oxygen, and nitrogen are suck a system. That's why organic chemistry is a thing. There aren't that many families of elements with that property.
Ammonia and silicon based life have been suggested. But none of the alternatives have very promising chemistry. See [1]. Life is probably stuck with CHON, in the "goldilocks zone" where water exists as a liquid.
We now know that planets are not rare. Many extrasolar planets have been discovered. A few are promising. The systems with known extrasolar planets might have smaller, more interesting planets, too small to be detected at interstellar ranges.
But stars are a long way away. Unless FTL is possible (which it probably isn't, because causality would break), the most we can hope for is someone to talk to by radio or something similar.
See the Drake Equation.[2] There's been progress on firming up the numbers since the 1950s.
CHON are also very common in the universe. Most proposed alternatives (not all) depend on things less common. This doesn't rule them out but it makes the odds worse.
Radio communications between tween solar systems require more energy than we have. We couldn't detect earth level civilizations in the nearest solar system (which probably doesn't have earth like life) even if both of us by chance aimed at each other at the correct time.
> Anyone can imagine a programing language that is as fast as C and as dynamic as Lisp, but when you sit down and think through what those goals entail, you realize the design becomes contradictory.
Ignoring the usual LLM rant, that's an interesting observation. Those conflicting goals reflect a problem that comes up quite often - the conflict between efficient volume production and flexibility. It's solvable for programming languages. That's what just-in-time compilers are for. Anything can change, but in practice, most things don't change that often. It's a caching problem.
This hits much harder in manufacturing. An extreme case is what was once called "Detroit automation" - totally specialized lines of machine tools that could make V8 auto engines all day and all night with very little human attention. But that's all they could make. Even switching to a V6 or a different cylinder size required new equipment. The other extreme is 3D printing in metal. It works, but it's so slow it's only useful for high-value items. Space-X makes Raptor engines that way. Nobody makes auto engine blocks that way.
A decade ago, there was a huge enthusiasm for 3D printing for making everything. That's declined. It's become another machine in the machine shop. It works, but if you want to bang out thousands of something, injection molding or stamping is far faster. There's a sizable tooling cost, and then each item is cheap. This is the tradeoff between efficiency and dynamics.
A year or two ago, someone posted a link on HN to a video of someone making a small screw on a lathe. Nobody does that except out of desperate need for a non-standard part. Small screws are made by special purpose machines that bang them out at machine-gun speeds. American culture does not know this any more. Too few Americans today have been inside manufacturing plants. The culture has forgotten where stuff comes from.
The wgpu/winit/egui churn is a headache. It's worse if you're doing 3D work. Too many crates want to own the event loop. "We're a framework now!"
I'm dreading the "upgrade" from wgpu 24 to wgpu 29. I stayed with wgpu 24 for a year so I could get work done. Now I have to fix three programs and all their tests and examples.
Order of initialization can be supported at various levels:
- Completely random (OK if interdependence are locked out, otherwise bad)
- Consistent, but sorted by something such as alphabetical name (meh.)
- Manual, controlled in linker scripts (headache)
- True dependency tree order, including diagnosing loops (seen in the Modula family).
General comment: yes, you can, and you probably shouldn't unless you have profiling data that indicates a significant performance improvement for a critical use case.
I think these things are used more for developer experience than for performance, since you can always just do the initialization in main if you really have to.
It's something of an issue if you have some crate that needs to set itself up at load time without a call from main. But those are very rare. Even "simplelog" needs a call at startup to do anything.
Yeah, I think of avoiding the call from main as a devex consideration rather than a performance one, since either way the initialization code runs once at process startup.
But it can cause problems when there's multiple crates trying to do it. If the call is explicit, the application developer can sequence the calls appropriately (or at least deterministically), as opposed to having the order determined by details of the implementation (something that was learned from the C++ 'static initialization order fiasco')
I'm trying to think of any useful crate that really uses this. "Tracy", the profiling crate, used to self-start on loading, but that was changed to require a call to get it going.
This is a price the US pays for the right to keep and bear arms. US cops have to assume that everyone is armed. That leads to a paranoid style of policing. As gun laws have become less restrictive, cops have armed and armored up.
It’s possible it’s a bit of an “arms” race, the police are more aggro but so are the public. At least in public perception back before the 70s its was perceived that by and large there was “respect for authority” but that’s eroded over the decades for various reasons among them court cases asserting more rights for individuals where cops can’t just up and arrest willy nilly. But also movements like “sovereign citizen” leaks in places enough to affect behavior elsewhere.
Also weapons are relatively cheaper today than decades ago.
There's an insane number of police shootings where it turns out the person was looking at an insanely large sentence and they "weren't going back to jail." From that perspective it's not even clear they're acting irrationally -- if the penalty for third-striking for stealing a TV and murder is the same then some criminals are going to make it worth their while.
Assault rifles were banned for most of that time. Now I can go buy 10,000 rounds of green tip XM85 ammo and an AR platform at a gun show in an hour. I’m not saying that justifies militarization (that’s mostly war profiteers selling to police departments and right wing alignment with law enforcement) but OP isn’t completely wrong.
There’s not much to back this up, at least that you’ve included as reference.
The bigger issue that comes to mind and that you can actually look in to is the practice of teaching police departments about “Killology”. This is (or was) a kind of seminar that taught departments this mindset of “everyone that an officer interacts with is a potential threat”. Add this to the “super criminal” bs that was popular in the 80s/90s, the constant right-wing fearmongering about dangerous criminals in blue cities, and the militarization of police, and it feels more like they’ve been primed for violence from the power structure more-so than any actual threat from the public.
At the range nearly all casual police interactions like traffic stop happen happen at (<20ft), a knife has to be treated just as deadly as a gun. So even if you remove the guns you'll still have to treat everyone as a deadly threat under such a model.
And yet mysteriously the British police seem to be able to deal with such situations with nothing more than a Taser - and often manage to avoid even using that. And yet, in spite of all the falsehoods being spread about knife crime levels, their mortality rate is much less than that in the US.
Do you have a reliable citation for this claim? [0] I disbelieve that this has been happening in any substantial way in the US. I expect that at very best, they've stayed roughly as restrictive as they have been for quite a long time.
> US cops have to assume that everyone is armed.
Weird. In San Francisco, California (a city of roughly 800->900k), the regular CompStat reports [1] have this to say about the number of incidents of firearm violence (whether fatal or non-fatal) in the city:
* 2022 -> 185 incidents
* 2023 -> 162 incidents
* 2024 -> 132 incidents
* 2025 -> 101 incidents
For fun, you can slap this pretty fucking shitty Power BI dashboard [2] around to compare those numbers to the number of times cops have either threatened to shoot or have shot someone each year.
Weirdly, I'm having great difficulty finding the city's officer injury reports. In the absence of those reports, I'll assume that policing still doesn't crack the top ten most hazardous jobs in the US, and that it's still roughly as hazardous as being a groundskeeper or professional athlete.
[0] If your supporting evidence is "spooooky ghost guns", I'll laugh my way out the door.
It's pretty clear to me that gun laws have become less restrictive. Stand your ground laws and open carry have become more commeon and more normalized. Tactical rifles (i.e. semi-automatic variants of military rifles like the AR) are less restricted since the "assault weapons" bans expired or were overturned. Perhaps even more importantly, ARs have become a much more prominant part of gun culture. Openly carrying guns at protests has become more normalized (although it did exist before; e.g. many Black Panther demonstrations).
> Openly carrying guns at protests has become more normalized...
Yes. For hundreds of years, open carry has been legal just about everywhere. More people choosing to do things that have been legal for ages doesn't mean that the relevant regs have been loosened.
I do note that over the past couple of decades it has become illegal to openly carry in many places, [0] so that's a substantial increase in the restrictiveness of firearms regulation.
> Tactical rifles (i.e. semi-automatic variants of military rifles like the AR)....
By this you mean to say "semi-automatic variants of rifles designed in the last sixty five-ish years". [1]
> ...are less restricted since the "assault weapons" bans expired or were overturned.
You should look into the status of state regs on firearms possession and notice how many of them have been enacted within the past decade. You should also look into the regs on ammunition production, sale, and possession. While the regulation of ammo possession, sale, and production is not literally the regulation of firearms, a firearm without its ammo is no fun to operate, unless you're really into swinging around a very expensive, poorly-balanced club.
[0] ...among them, schools, hospitals, wherever a private business owner places a legally-conformant notice...
[1] Seriously, go look at when the AR-15 was designed and first manufactured.
> a land border with a large continent that has a lot of guns and violence and criminality
That's a curious perception indeed, given that guns predominantly flow from the US to Mexico and not the other way around, and guns in Mexico are of mostly US origin.
I suspect this fear of guns largly explains the additional risks police face.
assuming everyone has a gun and is willing to use it, raises the stakes of every encounter.
so instead of a police encounter starting at a very low risk level (casual conversation), it starts a very close to deadly force risk.
This causes both sides to be a lot more tense, with a lot less room for mistakes.
It also makes any encounter feel very risky.
I don't think people having a gun prevents police from starting an encounter at a casual level. But the assumption everyone is out to harm them, and has the means to do so, does.
Plausible sounding theory, but in reality what kills cops is traffic accidents. Even there it's not people running them over. Cops are seldom charged with DUI where other dangerous behavior on the roads.
> requiring transistors that can efficiently switch on and off at high speed under high power.
Right. Switching power supplies need to go from off to on fast. In the full on state, the resistance is near zero (millohms with modern MOSFETs), and there's little heat dissipation. In the full off state, the resistance is near infinite (megohms with modern MOSFETs), with little heat dissipation because the current is near zero. During the transition, the switching element dissipates power as a resistor. The less time spend in transition, the less heat generated and the higher the efficiency.
Today the components are so good this is easy and efficiencies have passed 90%. That wasn't true in the Apple 2 era. Power transistors had higher OFF resistances, lower ON resistances, and slower slew rates. The better power transistors cost more. A cost-effective power supply that wouldn't overheat was tough to engineer.
(I've designed and built a switching power supply. Worked fine, could handle no-load and a dead short,
and didn't blither all over the RF spectrum. Probably had twice the parts cost a good designer could achieve.)
Correction: lower OFF resistances, higher ON resistances. There was heat generation even outside the transition period.
Every modern locomotive motor is powered by a switching power supply. The transistor packages are about 10x20cm. You could hold one in your hand. They're not super-expensive. They're smaller than a mechanical switch of the same rating.
MOSFETs are insanely good switches. It's amazing that's physically possible.
The brains of corvids are not closely related to mammalian brains. All the mammals have roughly the same brain, but corvids have a different architecture.[1]
Intelligence seems to have evolved three times on this planet - mammals, corvids, and octopuses. Octopuses have a distributed system rather than one central brain.
They all have neurons, but the higher level architecture differs drastically.
Knowing that several different architectures can work is important for AI.
There's apparently more than one way to do it.
I appreciate the linked article. I wonder if the the list should be expanded to 'at least three times', and I start to think about intelligence in plants.
My point here is that if intelligence developed more than once, it didn't come from some one-time random event or divine intervention. It looks like once there are connected neurons, it evolves via continuous improvement.
In some environments, that evolution hits limits. Flying birds are limited in brain mass or they can't get airborne. Which may be why corvids don't rule the world.
Oh yes absolutely. Intelligence certainly developed more than once. Bird brain is no slur.
Before your comment I would have said it emerged twice, but then I had not considered octopuses, they are wicked smart and so unlike other intelligent animals we know.
TechShop used to have a medium sized vacuum forming machine, but it was lost in one of their moves. Those are useful for tool trays. Lay down all the tools for some kit, vacuum-form a tray, and put the tray in a case for the kit. Often used in aerospace, where you want to make sure nobody left a wrench inside the engine or fuel tank.
It is much easier and faster. My friends father had a vacuum forming business, mostly making blister packs. I even worked with him fixing a big packaging machine in Brooklyn. You can make a primitive version with a plywood box with drilled holes and a vacuum cleaner.
People who have been in manufacturing for a long time understand all the methods used and the common theme is simplicity. There are tons of great uses for 3D printing but like robot arms, I see them misused time and time again to perform tasks a much simpler process/mechanism can perform. Go watch the TV show "How It's Made" and some of the machinery and mechanisms are delightfully simple. This is what universities don't teach their students and they graduate only knowing how to wield giant, complex hammers.
I wouldn’t say 3D printing is a misuse, even in this case, vacuum forming a tray insert is much faster, but I don’t have space for a vacuum former in my apartment but I can easily fit a 3D printer which is much more versatile.
Life needs some kind of chemistry that doesn't lock up into compounds so stable they're hard to crack apart, but allows compounds stable enough to build structures. Carbon, hydrogen, oxygen, and nitrogen are suck a system. That's why organic chemistry is a thing. There aren't that many families of elements with that property. Ammonia and silicon based life have been suggested. But none of the alternatives have very promising chemistry. See [1]. Life is probably stuck with CHON, in the "goldilocks zone" where water exists as a liquid.
We now know that planets are not rare. Many extrasolar planets have been discovered. A few are promising. The systems with known extrasolar planets might have smaller, more interesting planets, too small to be detected at interstellar ranges.
But stars are a long way away. Unless FTL is possible (which it probably isn't, because causality would break), the most we can hope for is someone to talk to by radio or something similar.
See the Drake Equation.[2] There's been progress on firming up the numbers since the 1950s.
[1] https://en.wikipedia.org/wiki/Hypothetical_types_of_biochemi...
[2] https://en.wikipedia.org/wiki/Drake_equation
reply