On the topic of warfare, wars are fought differently now. Compute will be mentioned in the same breath as total manufacturing output if a global war between superpowers erupts. In highly competitive industries this is already the case. Compute will be part of industrial mobilization in the same way that physical manufacturing or transportation capacity were mobilized in WWII. I’m not an expert on military computing but my intuition is that FLOPS are probably even more easily fungible into wartime compute than widget makers, and the US was able to go widgets->weapons on an unbelievable scale last time.
There are plenty of military uses for computing, but I also find it hard to believe anything but a handful of datacenters are or could be a major factor in anything but a completely 1 sided war. They are very vulnerable targets that are easy to locate and require large amounts of power and cooling. I also just don't see the application, encryption capabilities far exceed the compute available needed for decryption and computing precision and speed with even 20 year old tech far exceeds the precision of anything you would want to control. Even with tangible banefits, say 10% more or less casualties than there would be otherwise, in an exchange with anything resembling a peer military force im not sure it matters because everybody already loses.
They’re unclassified public cloud GPUs today, much the same as the massive industrial base of the United States was churning out harmless consumer widgets in 1939. Those widget makers happened to be reconfigurable into weapon makers, and so wartime production exploded from 2% to 40% of GDP in 5 years [1]. But the total industrial output of course didn’t expand by nearly that much.
I think it’s maybe plausible that private compute feels similar in the next do-or-die global war.
The United States has almost no domestic capability to produce advanced semiconductors. There is no abundance of industrial capacity cranking out GPUs that can be quickly diverted from AI companies into weapon systems.
Even if private compute was at a level of maturity where you could use it for classified workloads, knowing that the infrastructure is being managed by someone in India or China, securely getting data into and out of that infrastructure is still a mostly unsolvable problem.
My point is the existing private DCs can be reconfigured for a different use. Building new gpus is not required to on-shore compute. We already have it. Obviously if the military started contracting out compute onto the hyperscalar clusters it would involve a host of changes. I wasn’t aware that they were letting India and China manage their infrastructure… That seems exceedingly unlikely? That relationship would obviously be severed if the compute was reconfigured for the military.
yields are constantly improving on monthly basis, according to executives around 7% per month, so the capability is definitely there, but yields still needs some time
A gated, premium-tier product differentiation strategy only works when you sell the differentiated product. They went to market with 4.7 nerfed at security work and aren’t letting even large, vetted corporations pay more for the Mythos model… sentiment is quite negative where I work right now. There’s a real possibility that open source will give them a hair cut in the interim. And if the SWEs start modifying their CLI flows to avoid lock in to `claude`, it’s probable that the hair just never grows back. Losing strategy.
It's going to be quite a while until open source models catch up. And, as long as Anthropic maintains the perception that Opus is even slightly better than the best OSS models, they'll still be the preferred tool for professional developers.
Even if the best OSS model is only 1% worse than Claude, do you want to risk your codebase on it? When you're working through a tough bug in your code, and an OSS model just isn't grokking it, wouldn't it be only natural to want to cast it away and say "I should only be using the very best tools, dammit! My time is too valuable!"
That said, I agree with your point about SWEs modifying their workflows to avoid lock-in. That's a good idea, no matter what.
Have you ever reviewed an AI-generated commit from someone with insufficient competence that was more compelling than their work would be if it was done unassisted? In my experience it’s exactly the opposite. AI-generation aggravates existing blindspots. This is because, excluding malicious incompetence, devs will generally try to understand what they’re doing if they’re doing it without AI
I have. It's always more compelling in a web diff. These guys are the first coworkers for which it became absolutely necessary for me to review their work by pulling down all their code and inspecting every line myself in the context of the full codebase.
I try to understand what the llm is doing when it generates code. I understand that I'm still responsible for the code I commit even if it's llm generated so I may as well own it.
This article could’ve been written 20 years ago with only minor revisions, and it would’ve been true then. But it’s not now. It is trivial, literally a day of work, to set up a build system and CICD environment using Verilator if you are already proficient with your build system of choice. Learning TCL to script a bitfile generation target using your FPGA vendor’s tools is a few extra days of work. And regarding IDE support, the authors complain about the experience of writing code in the vendor GUI. They should look at one of the numerous fully featured systemverilog LSPs available in e.g. VS Code.
The real argument for open source toolchains is much narrower in scope and implying its requirement for fixing a nonexistent tool problem is absurd
I agree about with your claim, but the answer to your question is that “weeds” is a set of species that contains both invasive, ecologically harmful species, and crucial native annual and perennial forbs+grasses.
From the universalizability principle, if everyone merely let “weeds” propagate, because of the ecology of invasives that are in that set, we would be MUCH worse off for the next few millennia than we are now. Until the ecosystems healed and the “invasives” become “keystone species”. Not sure how long that would take but we won’t see it :)
It depends on the target and the surrounding soil. It’s often easier to pull especially for the random weed that sprouts up around your landscaping. However if you are trying to manage an infestation of invasive species, where the surrounding soil will have a seed bank heavily contaminated with seeds from the years of invasive reproduction, it’s usually a bad idea to merely pull. You can expose soil to sunlight and cause an explosion of dormant seeds. And some nasty invasives are nearly impossible to remove by hand because of their root structure — some species even leave little rhizomes broken off in the soil along the root structure when you pull off the foliage causing a hydra effect.
tl;dr targeted herbicide is a much less evolutionarily selected-for offense, as opposed to hand cultivation which mimics attacks plants have evolved to survive for eons
Asimov's Zeroth Law of Robotics: "A robot may not harm humanity, or by inaction, allow humanity to come to harm."
This is an addition to the other three laws embedded in positronic brains:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
To me the zeroth law echoes the paternalism built into LLMs, where they take on the role of shepherd rather than tool.
The other day I asked one a question, and didn't get an answer, but did get a lecture about how misleading the answer could be. I really don't want my encyclopedia to have an opinion about which facts I shouldn't know.
You’re just stating a trait of animal predation. Predation often eliminates the weakest animals in a population. The point is that invasive species by definition of invasive predate in a way that their prey has not had adequate evolutionary space to adapt to because they are introduced. All the words in that sentence have precise definitions in ecology and I don’t think we share the same framework or definitions
Coyotes do not deserve to be mentioned in the same breath as a lion. And people should celebrate anytime a native predator is able to carve out a niche in an urban environment as long as it doesn’t involve murdering children. Which coyotes don’t do. They eat small mammals and also sometimes invasive feral cats.
The reason coyotes don't eat children is that they're kept separate. Same reason sharks don't eat children. Left to their own devices, coyotes obviously will eat children.
> I used to go for six mile walks in the High Desert for exercise. I routinely ran into coyotes. They never bothered me.
> I did tell my children to not go out alone at night. There had been two or three attacks on children in the previous five years.
Another story that came up on HN was someone describing how he used to adventure through the wild areas near his home, and was never bothered by the lone coyotes he saw, but that on one occasion he met two coyotes together, and they began circling him. I didn't manage to find that one today.
My dad lives in Silverlake LA and there’s a coyote den up the hill a bit from his house. He’s got two huge dogs, but sometimes when you walk em they’ll get super excited to go after a coyote standing in the middle of the street. Thing is, they’re damn smart, and you can be sure that if you see one in the street, there’s another one the opposite way down the street, usually hiding under a car. They 100% know when they have numbers on a larger animal and when they don’t.
Coyotes attacked a small dog in the park near my house last fall (when the juveniles leave the den and try to stake out territory) so I understand the concern. But comparing them to lions in Zimbabwe doesn’t resonate with me and I live in Chicago (we have tons, one was under my porch last year).
Think of all the samples of the interaction function between humans and the >1mm coyotes (often unbeknownst to the humans) in American cities each day. The list of all attacks recorded in modern times has a Wikipedia page. In human-created spaces we make very little separation from the habitats coyotes live in. They choose not to predate the defenseless babies they encounter in backyards because it is not the ecological niche they have carved out.
I will let my older children play unsupervised in my backyard despite knowing there are dozens of coyotes in my city because no creature has made a niche out of killing them. The same is not true for my very young children but that’s because toddlers have made an evolutionary niche out of killing themselves :)
> but that on one occasion he met two coyotes together, and they began circling him.
He must have done something to react fearfully. I've charged at literal packs of 10 to 15 of them, yelling and enjoying myself, only to have them scatter in complete chaos. With children and no adult present, coyotes are a real danger, but when they meet adults that show no fear, they're amazingly nervous. I know of only one North American attack on a human adult in recent history (a rather petite woman) that was fatal.
While walking my dog on leash some years back, we were surrounded and stalked by 5 very hungry coyotes (it was a drought year and they were very scrawny) for about 20 minutes.
They got increasingly more bold, eventually only just out of kicking range. As I would charge and threaten the one or two in front of me, the others would try to approach from behind. We were being hunted.
All this while carrying my dog (they definitely would have killed him if he were on the ground) and wearing flip flops.
This was in Santa Cruz, late in the summer, at dusk. Terrifying. They tracked us all the way back to the car, but once I found a good stick they became much less daring.
reply