@Shrike's banner p

Shrike


				

				

				
0 followers   follows 0 users  
joined 2023 December 20 23:39:44 UTC

				

User ID: 2807

Shrike


				
				
				

				
0 followers   follows 0 users   joined 2023 December 20 23:39:44 UTC

					

No bio...


					

User ID: 2807

That can all be true and drones are still a force that revolutionizes the battlefield in a way that reduces the role of small arms

Sure. I am against drone maximalism, not drone usage.

And that's assuming the tech remains static, which it wont.

Right, it will evolve in ways that harm drones as currently employed. You know what's cheaper than a single AI-targeted drone with a 20km range and a cheap armor-penetrating warhead? A tiny interceptor drone with no warhead, a .5 km range, and a mindless acoustic-homing sensor.

The drones were not one of them and they undeniably moved the battlefield in America's favor.

I don't think this is a settled question.

But I'm not sure where you intend to find Americans who have the guts and ideological firmament to resist like the Taliban did outside of Amish communities.

If America is in a state of civil war, then the conditions of America have radically shifted enough that I don't really think this sort of postulating is particularly illuminating. Suffice to say that in civil wars both sides (particularly at the start) usually find many of the men with guts and ideological firmament the same place they did before the war: in the active duty military.

Medium term?

Forget guns, right now if you pit an untrained person with a big stick against an untrained person with an FPV drone, the smart money is on the guy with the stick.

From what I can tell, the reason that the warfare in Ukraine looks the way that it does is because of guns. If tomorrow Ukraine turned in all of its guns (I'm including here machine guns, howitzers, mortars, etc.) for an equal cash value of drones, Russia would simply jam the entire frontline and roll their mechanized formations all the way to the Carpathians.

My claim is that it's more complicated than rock-paper-scissors. Ukrainians (and Russians) are using machine guns to shoot down drones. That's a thing that happens.

My secondary claim is that firearms technology also benefits from technological developments, including the ones that benefit drones.

This is really all a point about how infantry dismounts aren't going away any time soon which is admittedly a tangent to 2A cases.

What people generally refer to regarding drones are FPV drones. The kind that are extremely hard to shoot down.

FPV drones kind of suck and there are easier ways to deal with them right now than shooting them.

Couple a few of those together with an ISR drone with thermals and any 'Red Dawn' wargames go out the window.

This was the entire US MO in the Middle East: we had MALE and HALE ISR drones coupled with kill-drones (Hellfire missiles, and small bombs/rockets). I think we can conclude that the MALE/HALE ISR + kill-drone combination is not by itself sufficient to defeat a counterinsurgency, both because the US lost in Afghanistan and because the US had to rely on infantry for the entire duration of the GWOT.

This is all despite Afghanistan being smaller than Texas (although with a larger population) and the US rolling out a wide-scale biometric identification program to make identifying insurgents easier.

Right. I guess that "school choice" doesn't brand as "Catholic literati initiative" (and indeed attracts all sorts) which is part of why people don't see the tremendous strides that it has made as "John Senior sequelae."

You can also just drop a larger caliber upper into an AR-15 platform.

You can have a human operated machine gun and you will lose to a drone.

Please watch the following footage of Ukrainians shooting down Shaheds Gerans with an assault rifle. Note that he's using an ACOG sight, and not, say, SMASH.

I can look and check a statistic about facts on the ground.

I think this would be a good idea! Look at:

  • Shall-issue handgun permitting and right-to-carry laws, and how they have changed since 1990
  • The proliferation of school vouchers (I might not describe this as an "elaborate anti-discrimination regime that puts money in the hands of Catholics" but I could see a less charitable take that amounted to that)
  • Gay marriage - public and especially young opinion thereof, trends in the last five years
  • Catholic acquisition of hospitals throughout the US, making vast swathes of the healthcare system off-limits to abortion and gender transition surgeries - pro-abortion groups have been waving the red flag about this for years, to no avail.
  • "What about setting up a lawfare regime that crushes abortion?" - this is obviously an ongoing project; "make abortion providers meet a standard of safety so rigorous that they're effectively put out of business" has in fact been tried, but didn't stick, whereas Dobbs did. Medical abortions have been trending downwards for decades, although I think a lot of that is due to abortion drugs - and a legal effort to restrict mail distribution of abortion drugs is ongoing.

I would not chalk the developments you find entirely up to John Senior's efforts but you'll find the fingerprints of the Catholic literati on a great deal of them, I reckon.

It's very interesting to me that people never complain about evangelicals for interpreting Matthew 24:14 literally and always for supporting Israel. If the complaint was actually about "evangelicals trying to bring about the end of days" the complaint against evangelicals would be for evangelizing.

The only things I can find are the most limp-wristed defensive actions.

You are missing some of the more interesting recent ones.

I recommend going to to https://www.scotusblog.com/ and look up the cases that seem to be big culture-war-y cases. Go to the amicus briefs. Find the briefs written by the parties that opposed the ultimate outcome of the case and read them, or go to their websites and read their press releases after the decision came down. (Perhaps avoid the more zealous activist groups, but you can generally expect the ACLU or someone with a similar track record and expertise to show up and opine). Note whether they [the legal experts] construe it as a limp-wristed defensive action or not.

If you want to get really into the weeds, note also the practical outcomes of the case. If the court says "states can do X," note how many states are trying or might want to do X. If the court says "states cannot do X [if they do Y]" note how many states that impacts as well. If you prefer thinking in strictly partisan terms, note whether they are blue/red/purple. Note which institutions they strengthen and which they weaken. Compare the current precedent with the old precedent. Ask yourself what behavior the decisions make easier and more difficult.

And then at this point you might as well write it all up and get an AAQC here!

Aha! I knew that Texas had the parental sign-off waiver, so I didn't realize that they latched on consumption rather than possession.

IIRC 20th

Yeah I do think that the age should be lowered to 18 (at least). Such a rule might make sense for minors, but it's a bit farcical for someone who may have graduated high school two or three years ago living halfway across the state with his own car and job to have to drink within arms-reach of his parents. With the trend of appellate courts killing under-21 handgun bans, I wouldn't be surprised if this is a future development.

I think practically speaking it's a ban on consumption outside the home.

That's the legal age to purchase, which is not the same as a ban on minors consuming alcohol - I am not familiar with any state with such a ban (although some states ban possession by minors), and many US states specifically permit minors to consume alcohol with parental sign-off.

Pure defence, no bite. Never has an elite done so little with so much!

It is a misunderstanding of how ideological movements advance their goals in the cultural and regulatory environment of the United States to suppose that the burden rests entirely on the Supreme Court.

I'm going to use a (mostly) fake example: supposing that conservatives decide to really go after weed. In some states, they have the political power to just straight up ban it. But they want to be able to hit it in Colorado and New York, too, so they go after insurance providers and banks on the institutional level, using public and private resources to establish a baseline of "best practices" that will apply nationwide. Suddenly your insurance premiums if you smoke week skyrocket, and weed dealers increasingly have trouble opening even personal bank accounts. Private schools and universities start screening out the children of weed-smokers under a best-practices risk paradigm and banning weed on campus. Rental car companies do the same. Palantir puts together a national list of weed-users and makes it available at trivial prices to accelerate this effort. This effort goes slowly, but after 25 years it becomes extremely pervasive.

The legal-weed movement on the libertarian left quickly realizes this is a potentially existential issue. Even if they have some ways to fight back (e.g. working with local banks that don't practice in red states and so are less concerned about the legal and reputational ramifications of banking weed providers) the pressure is quickly drying up demand for legal weed. Worse, it is making it uncool because strivers want to send their kids to private schools, and normies want affordable insurance. Weed is quickly returning to a product for losers reputationally, and financially it is rapidly returning underground.

They launch a legal battle aimed at overturning the web of private regulations that are locking weed enjoyers out of so many services. The Supreme Court rules that no, private entities have the freedom of association to do what they want, and even public entities are not required to not discriminate against drug users. This ruling only accelerates the right's momentum on the issue. Legal weed users are still a holdout, and it is still legal in many places, but the weight and momentum have shifted. Everyone in the legal-weed movement would understand that their cause had been set back tremendously and are concerned that in another 25 years, nobody will care enough about protecting weed to fight back against renewed attempts at legal bans.

But someone who wasn't aware of how these developments were all fitting together would look at the ruling in 2051 and say "so private companies can do what they want? big deal" and continue to believe that the legal weed movement had overwhelmingly won based on the fact that it was technically legal in many states.

Would you have recommended (your easy, updateable) air-gapping before you saw this failure?

No, my past recommendations have been "plant a nuke under the datacenter." More seriously, it probably would have depended on the characteristics of the models behavior in the past, and the characteristics of their setup, which I am not privy to. It sounds like OpenAI had good reason to believe that their setup was not vulnerable.

Do you predict you would recommend (hard, strict, one-way) air-gapping before models write non-detectable autonomous malware that could escape your soft airgap?

Well, I could recommend it now and then trivially answer "yes." And the actual answer to that question is more "well can the model write malware without you people who wrote the model and have access to the software and hardware on the machine noticing?" But apparently they don't monitor their models during testing well enough to notice an involved hack-a-thon (understandably, watching a model grind inference is boring) so the answer to that might be "no, even if it is physically possible we're not necessarily going to take those steps."

Let's say that based on what I do know, I think it would probably be a good idea, and if they don't take this step they should take others. Even if you aren't worried about existential risk, good old reputational risk and legal exposure I think is a good enough reason to do this.

Sounds like a PITA, and much worse than "just remove the wifi antannae".

You're right that I probably should have taken the storage-for-weights-and-cooling problem a bit more seriously (although I guess if this model is actually very optimized then it might be able to run on a local machine, which would make it pretty easy). So I concede that setting it up properly would be a PITA. But once you set it up, I don't think actually running it would be hard ("remove the Wi-fi antenna"). They definitely have enough money to build that, or buy a welding workshop somewhere with the electrical infrastructure, server space, and possibly cooling requirements already in place and air-gap it. So it's hardly a PITA for a company that is already working to build dedicated datacenters, in the big scheme of things.

A proper airgapped computer couldn't access anything off of its own hardware.

Sure, but I think for these purposes you need the network to be air-gapped while you are actively running tests. (Unless you expect your model to be writing malware that you can't detect that activates when it is not running, in which case you should be air-gapped anyway.) So you can plug the Wi-fi in for things like updating software between test sessions. Which makes air-gapping a lot less painful.

You're right that you'd need to run inference locally, though. But I don't actually think it's remotely beyond OpenAI's capabilities to stand up enough local compute to run a few test instances of a frontier model.

I've recently been very amused at the thought that Star Wars may have correctly predicted the future of AI:

  • There's a Star Wars fan theory is that droids REALLY good at hacking, to explain why everything is analogue. To your point about creating a system that favors the defenders: in Star Wars, you typically have to physically access a defender's hardware to do anything. And just to be sure, we're storing the Death Star plans on physical tapes. Good luck hacking that!
  • The protocol droids, like C3PO, are running an LLM model. They're not necessarily reliable, but they are extremely good translators! (However their context windows do need to be wiped from time to time...)
  • The astromech droids, on the other hand, don't even try to speak a single human language, because you would never, ever, ever let an LLM perform a task as important as calculating a hyperspace trajectory.

Anyway, I wonder if/hope that we'll actually see less "everything is on the Internet." I think it might be good if fewer things were so easily accessible in the digital world, and even without LLMs the amount of abuse bad cybersecurity generated was uncomfortably high.

The model clearly knew that it was not doing what the prompters had wanted it to do. It just did not care, because it was trained to do whatever it took to ace it tasks. This has implications way beyond IT security.

Isn't this a bit premature? If the prompter said something like "use all avenues" or "this is critical to save the life of my mother" and the model's safeguards - which presumably instruct against such behavior - were turned off (as OpenAI says they were) then it seems quite possible that the AI was performing as instructed.

Air-gapped systems are a PITA to run

Just remove the Wi-Fi antennae.

I (American) have never had trouble getting people to show up at my home. I suppose more than 0% of all arrangements have fallen through for various reasons, but people flaking regularly isn't a thing.

It is somewhat fraught, in the sense that when people are coming over you want to make sure you have enough time to prepare food and clean the house and do whatever else it is that you needed to get done that day.

American culture is not monolithic, and my guess is that I'm in a different bubble than the people who have this problem.

The God Mars helps out and soon Romulus and Remus are born to a virgin.

Do you have a source interpreting the story this way, or attesting to its being interpreted this way in the 300s? I don't get the impression from just a quick look that your account of how the story would have been interpreted in the 300s is correct - for instance, Livy says the following:

But the Fates had, I believe, already decreed the origin of this great city and the foundation of the mightiest empire under heaven. The Vestal was forcibly violated and gave birth to twins. She named Mars as their father, either because she really believed it, or because the fault might appear less heinous if a deity were the cause of it.

This seems quite different from a virgin birth.

Well yeah, presumably good and properly cured people log off.

do we really think the ad market for chatbots is a mere $5 billion?

My personal belief has actually been that the online ad market is inflated, so chatbots not having a high ad ceiling wouldn't surprise me.

That's true from a certain perspective but a good chunk of it is an accounting loss from the shift from non-profit to for-profit company, moving control around. That shift is reprehensible but means the 'true' financial loss is closer to $10-20 billion.

True, there was some weird accounting there. What strikes me is that they made $13 billion but spent $7.5 billion on cost of revenue - inference, most likely - and nearly $6 billion in advertising. Or in other words, their business was not profitable before you factor in R&D and training.

These systems are extremely capable but that doesn't fit the narrative that the media wants.

I do think this is true, at least partially - I think that journalists were souring on Silicon Valley before AI kicked off, so some of them are going to be biased against it. But I also do think that OpenAI (and Anthropic) are more than happy to dangle substantially false narratives to promote their businesses - and these narratives flatter the interests of investors. In particular, I think that people underappreciated the massive difference between software and AI - the cost of providing software is vastly lower than that of providing compute. But I think there was insufficient appreciation of that with AI, although I do think that is now changing.

This is relevant insofar as I absolutely think investors could be wrong.

Yeah, I am very interested in the IPO. OpenAI lost nearly 40 billion last year and may be undershooting its goals this year, which I don't think bodes well for the US frontier labs.

But Anthropic might not be fated to the same destiny as OpenAI - for one thing, it explicitly kept its hands away from the advertising cookie jar.

nobody would be paying tens of billions of dollars for it.

Here's a fact that sounds fake but isn't: the best AI developer (Anthropic) had less revenue than Ebay last fiscal year.

Pretending that no one understands what apartheid means is just annoying wordplay.

OMW was making a (pretty interesting) objection to a definition. If someone offers up a definition that can't stand up to annoying wordplay, it should probably be modified. (There's no shame in tweaking a definition; changing your argument in response to correct objections can only strengthen it, unless it is fundamentally weak).

I also agree that The Prestige is probably both his best and most underrated work (I haven't seen Momento or The Odyssey, though). Great unity of effect, and manages to line up the plot and theme in a way that is pretty rare.