Amodei, and Musk, and Altman, and Zuckerberg, and all of the rest are vicious, dangerous thugs who don't care who they hurt or kill, or what they damage or destroy, as long as they can satiate their egos and their greed. They're utterly devoid of conscience, of compassion, of ethics, of responsibility, of everything that it means to be a functioning human being in a society. They are raw, pure evil -- and they keep proving it to us every time they speak or write.

Only a moron would trust people like this, and I use the term people loosely here.

We didn't make them this way: they chose to become this way. And now they're whining about the consequences of their decisions. Well, guess what: every time throughout history that sociopaths have done this, have put their gratification and enrichment ahead of the people, the people eventually decided en masse that they'd had enough and that it was time to put a stop to this nonsense. I see no reason why this time will be different -- or should be different.

"AI" could mean "statistical pattern recognition" or "image processing and object classification" or "flight control" or "target prioritization" or any number of other things that aren't quite what we usually mean in 2026 when we toss around the term "AI" in 2026. (Certainly there's no need for a missile to have anything resembling a LLM.)

For example, given Ukraine's steadily-improving defenses, it might make sense to equip a missile with enough data acquisition and processing capacity to select targets based on real-time conditions while en route -- e.g., to shift from a primary target to a secondary one if the defenses around the primary seem to be working too well. Given all the RF jamming in play, that'd have to be done by onboard processing with no reliance on external communications.

But this is all speculation. Until they get their hands on one that's intact enough to recover the code running it, we won't really know. But one thing that's clear is that the Russians didn't install it in a munition arbitrarily: they had a reason. (Insert quote from "The Hunt for Red October" about Russians and plans.)

I've done a lot of incident response, and was a certified trainer for over a decade. One of the things I've observed is that the higher the reliance on technology among the party, the higher the probability that they'll require rescue. People who can read a topo map, can use a compass, who understand the importance of carrying more water than they think they'll ever need, and who have a (tested, working) flashlight rarely get into trouble. People who rely on Google/Apple/whatever maps, who are glued to their cell phones, who can't read a map or use a compass, set themselves up for issues. Of course, thankfully, most of the time those issues don't arise or are minor. But when they're not...

In this particular instance: 6000 feet of elevation gain is rather a lot. I've done 5000 up/down in a single day, but that's a LONG day even for someone who's fairly fit and accustomed to the exertion. Anyone who's not in shape is going to really struggle not just with the exertion but with the altitude, and we see in this instance, as they fatigue they may stumble and/or fall. It's just not a good idea, even for (let's say) young people who are physically active, go to the gym, play a sport or two -- because it's a very different kind of physical activity and because you can't just stop when you've had enough.

Aside: experienced trail runners make all of us look silly. I remember slogging my way up a trail at 11,000 feet when I heard the quick tap-tap-tap of one behind me. He went by with a quick "Morning!" and just kept going. I was still slogging my way up at about 12,000 feet when I saw him coming toward me, on his way back down -- same tap-tap-tap at the same pace, just out for a casual morning trot while we ordinary mortals were trudging along. I'm willing to bet that he did that trail often enough to know every rock, to have every step mapped out in his brain, to just dance his way up and down.

This particular one was exquisitely susceptible because it was managed by a stupid, corrupt crypto bro (yes, that's redundant; all crypto bros are stupid and corrupt) but there are similar ones managed by allegedly sober people with decades of investment experience...and they're not immune to the same market forces. The hype/mania around AI valuations will eventually collide with reality -- and perhaps "eventually" is "soon". A few people will make a fortune (and already have, on paper at least) and a lot of other people will be screwed.

I say that because it was triggered by a single change that should have been easily discovered during basic testing, or if not then, during any audit conducted during the last 5+ years since that change was made. Sure, it's possible that this was typical crypto bro sloppiness and stupidity, but that seems like a reach given all the opportunities for it to be detected before it was exploited. Add to that the pervasive corruption that is an intrinsic part of all crypto operation and it seems to me that the most likely explanation is that someone involved making that change has been playing the long game. It would be very interesting to find out who was involved in that change and where they are now: if they're still in the same position that would argue against this hypothesis, if they're not then perhaps they're elsewhere -- and cashing in.

The attackers are assessing defenses: what are they, what are their response times, what do the responses look like, etc. The real attack(s) will come somewhere else and will be much more thorough, because they'll be informed by the intelligence gathered from these probes. Given the state of US infrastructure security ("miserable") any competent attacker should be able to do a great deal of physical, economic, and societal damage rather quickly...

...which is one of many, many reasons why starting a war with a capable adversary without any kind of plan was a profoundly stupid and reckless idea.

I suppose that's cheaper and easier than using their own AI slop detectors -- doubly so because that would require hiring actual real live people and paying them suitably.

But I think that's not the real goal here. I think the real goal is to find out which AI slop escapes detection and use that information to fine-tune their models to crank out even more AI slop, so that -- eventually -- LinkedIn has almost no human-generated content. A lot of people will leave, of course, but those who remain will be the ones most open to exploitation, and that is where the real money can be made.

It's quite interesting, because it definitely shows a signal -- and as the figure notes, possibly a second signal -- in the noise.

Two considerations come to mind, however. The first is that this figure is based on 89 selected events, and while the criteria for those selections seem well-grounded, perhaps it should 62 or 117 events. In other words, have events been included that should not have been, or have events been omitted because they're not yet known? My own feeling is that as a first approximation this list is probably reasonable, but that it would be worth examining how including/excluding certain events changes this power spectrum.

The second of course is that this is an observation, not an explanation. There is no explanation, as yet, only various hypotheses. That's not really a criticism, a first step in understanding any process is to characterize the process as accurately as possible. I think a possible path forward here is to continue trying to do so without being overly concerned with the causative mechanism -- in other words, to try to understand that "what" as thoroughly as possible before delving into the "why".

Finally, I think it's worth remembering that when Alvarez et.al. posited an impact event 66 My ago as the proximate cause of a mass extinction event back in 1980, the very concept was difficult for many people to accept. We now do do because it's decades later and the substantiating evidence is overwhelming. Perhaps this is similar, and perhaps in time evidence will accrue that supports one (or more) of the explanatory hypotheses.

Agreed. I've been waffling on migrating from Github -- inertia, you know -- despite its increasing enshittification. This gives me at least some reassurance that Codeberg has its head screwed up straight, so I think I'm going to start moving projects over there. It probably won't be perfect (what is?) but at least I won't have to wake up every day and wonder what absolutely batshit crazy nonsense is going to inflicted on the site and thus my projects.

You do not understand. Let me quote Kevin Anderson, director of the Tyndall Energy Program. Keep in mind that (a) he's a scientist, and scientists are trained to write conservatively, and (b) he's British, and they are trained to write conservatively. In other words, when a British scientists puts something down on paper, it's quite likely to be the opposite of hyperbole. Here is what he wrote:

"A 4 degrees C future is incompatible with an organized global community, is likely to be beyond adaptation, is devastating to the majority of ecosystems, and has a high probability of not being stable."

I'm not British, but I'm certainly well-educated in science, so let me translate that for you. "Incompatible with an organized global community" means that it's unlikely countries will exist: the inevitable forced mass migrations, wars over water and food, and social unrest will make them untenable. "likely to be beyond adapation" means that we're not going to be able to engineer our way out of it. "devastating to the majority of ecosystems" means more floods, more fires, more superstorms, and even worse: no food. "high probability of not being stable" means that a runaway scenario is likely -- that is, it won't stop or even slow down at +4C, it will accelerate.

In this future, which is the one we're headed for, your "freedoms" will be meaningless and irrelevant. It's unlikely you'll survive, because it's unlikely most people will survive, and those that do will probably be those in remote areas and with the skills to survive without modern technology -- in other words, people who aren't posting on SlashDot today and who likely have no idea that the Internet exists.

So while you're certainly free to continue being selfish and ignorant, and my guess is that you'll make that choice, you might want to pause for just a moment and consider that your approach inevitably results in the deaths of billions of people -- including yours, your family's, your friend's, etc. -- and that maybe it might be a better idea to consider the merits of embracing science and cooperation. There's no other way, sadly, because all of the things that could have been "other ways" are no longer viable options.

Climatologists have been warning and warning and warning about this future. And yet humanity has not found a way to confront it and do the difficult things that have to be done in order to try to mitigate the consequences. Absent revolution on a global scale, I don't see a way out of this.

And the sad part, the tragic part, the awful part, is: this is only the beginning. It's the shallow part of an exponential curve.

Like every other unneeded and unwanted TLD that's been introduced in the last decade-plus, this will be overrun by spammers and scammers within days. There will be millions (if not tens of millions) of domains registered, used once/burned, and discarded. Since even attempting to deal with these on a case-by-case is simply unworkable, a much better choice is to preemptively blacklist them all permanently and get on with doing something productive.

...you need to make sure that any freelance criming that you do directly benefits them.

"Nothing major; I just handed them over to a stochastic parrot."

The phrase The design appears safer than [...] in this article is attempting to do an enormous amount of work and failing miserably.

Actually, I'm a Unix and Plan 9 enthusiast -- I merely focused on BSD Unix because it's the topic at hand. Having used, at one point or another, almost every Unix distribution that has existed since the mid-1970's, I have my preferences -- but since some of those are long-defunct, it's not practical to attempt to run them in 2026. And of course Plan 9 takes the Unix approach to its next level: many of the ideas it incorporates are advanced yet elegant, a quality that often gets lost in bloatware -- a canonical example of which is systemd, currently infesting any number of Linux distributions (but not the ones I'm running, which is in fact why I'm running them).

The point, which you seemed to have missed entirely, is that regardless of what you're running and what I'm running, the world is running rather a lot of Linux and Unix on all kinds of hardware, and it's often under the hood so it's not readily obvious which is in play.