Without a hostage there is no ransom!

I cannot agree more with this sentiment. It feels outright insulting to asked to read LLM output in a context where it is *supposed* to be human feedback. Tell me what you would have told the LLM to say, I can take it from there. I don't need you to LLM it up, because it will bury your point in a bunch of crap.

Could it provide useful info? Maybe, but I can do that myself if so. I want *your* thought on something, however incomplete it might be.

VMware manglement incompetent, film at 11

/ also, the popcorn you are eating has been pissed in

I went to school in VA. There was rarely a day that the heat would get turned on. They're big monolithic heat islands full of children, calorie powered.

We also didn't have A/C back then, either. Windows would be open all day. I doubt they do that now.

I know this comes off trollish, but: why can't these power companies (which are usually public utilities) just build more power production instead of jacking prices? What's the deal? Why is this a "hate on AI that runs in datacenters" problem, and not a "just produce more power" problem?

The dream is that the world is built for human limbs and the 'easiest' answer to claim the same versatility is to also have human limbs.

Stairs, cluttered terrain, a humble curb can all cause problems for the usually better answer of wheels.

The non-humanoid robots we already make those by the ton, and are, as one would predict, much more useful than human-like anatomy in their context. They however want to cover the underserved facet, banking hard on ML to make the humanoid design more viable while they traditionally are just infeasible to program.

Of course, that has proven a challenge, since the ML needs to instrument all the inputs and outputs of a human interaction system, and feeling is a huge part of human operation that cannot be instrumented. So they set people about trying to clumsily remote operate them in hopes of gaining training data, but it's low quality control and very low volume of data.

Middle-range strike drones are much cheaper than JDAMs (smaller payload, but you don't care about that against trucks), longer range, and let you operate in fully contested airspace or even when the enemy has air superiority.

Aerial bombs are for entirely different purposes; they're for destroying fortified positions. Whether the aircraft should be manned or not is an entirely separate question, but one thing is unambiguous, it needs to be big enough to carry said bomb (aerial bombs are very heavy).

But again, complete overkill for a transport vehicle.

The disconnect is the "promise" is that LLM brings expertise down to the masses. If AI is "too hard for Ford to get right", that dramatically undermines the messaging that drives the current expectations and levels of investment.

This is very much evidence that companies can't be as bullish as they might inclined to be, because whatever you may think of Ford, the typical company is probably worse.

It's less funny when you have people literally and sincerely saying this...

I have to take some solace in the fact that these are the same idiots who talk a lot and do nothing that I usually ignore who do this all the time, but management is extra entranced with them over the AI cheerleading this go around...

Being rich doesn't make you stupid, but being really rich starts to isolate you in a bubble of luxury and sycophants, and eventually you start to forget what the rest of the world is like, and start making decisions based on the unstated assumption that other people don't matter.

Re, the terrain of Donbas: compare, at the same zoom level:

Donbas

To a stereotypically flat place in the US, like, say:

Kansas

Unless you mean the "Smoky Hills" of Kansas:

Smoky Hills

Though their relief is only about 2/3rds that of that in Donbas. Donbas's relief is more like that of the Piedmont Province (the area west of the Appalachians), the dissected till-plains of southern Iowa / northern Missouri, the Tennessee / Kentucky western highland rim, or the low glaciated plateaus of the northeastern US (NE. Pennsylvania to southern NY).

It's not as forested as it used to be, but still has sizable patches left, such as along the Siversky Donetsk, mainly pine. Maybe the area east of the Appalachians would be a good reference for the mix of farmland with residual forest patches (well more than midwest states like e.g. Kansas). Defensive lines are commonly built in the forested areas, for greater cover.

Oh, and also (re: NERA) worth noting that there would be a brief boost in energy transfer to the generated gas from cell discharge. You wouldn't come close to fully discharging a cell (that requires lithium diffusion), but it can effectively instantaneously discharge the double-layer capacitance at the electrode-electrolyte interfaces, and very rapidly oxidize lithium at the anode surface (such as the SEI) / reduce species at the cathode. So in a way, not an entirely non-reactive armour, and somewhat reminiscent of the reactive-but-not-explosive NERA variants where they mix nitrate salts into the elastomer to make the reaction more energetic and gas-generating.

It seems that the most effective "drone defense" thusfar has been "moving in small groups or individually, at night or in bad weather, and then hiding in a basement until there's enough people / supplies to push further".

It's clear that armoured vehicle design needs to change. But hangar/turtle tanks hardly seem a durable approach either (even in Ukraine their use has fallen off). I'm still very much a believer in hybrid armoured vehicles, where you have a battery pack with several dozen km of range, and one or more generators powering it.

From a direct survivability perspective, if you use a non-flammable li-ion chemistry (there are plenty, it just means sacrificing some energy density - still requires managed venting / air control systems, however) and have cells in parallel connected by multiple busses, spread out across the vehicle's footprint, it becomes almost impossible to take out the entire power supply, just individual cells. Likewise, since electric motors are compact, you can have 1-2 motors on each axle, and again it becomes almost impossible for a drone to get a mobility kill that way (reducing approaches only to trying to disable the tracks themselves). If they take out the generator/generators, the vehicle still has its electric power to fall back on, and while it's not going to be making some deep push anymore, it can still keep fighting, and retreat when needed.

From an indirect survivability perspective, you have the ability to advance silently when needed (no engine noise, greatly reduced thermal signature), and since modern batteries have so much power density, you have the ability to have a higher top speed, which has proven critical for safety in drone-dense environments. You also have a lot of electrical power, for drone-detecting radars, drone jammers, anti-drone weaponry (lasers, microwave, etc), and so forth.

The mass and volume of the battery pack (we're talking maybe ~250kWh for a rugged heavy armoured offroad tracked vehicle, ~60kWh that for a lighter-armoured road-optimized vehicle) isn't wasted. Cell cans are steel, and between the inner plate and outer armour you're basically forming a honeycomb structure (good for dissipating shocks and spray) with a lot of thermal capacity (cells are organics, e.g. generally high specific heats). With a proper design, you might even be able to get it to function as non-explosive reactive armour. Specifically, contrary to misconceptions that NERA requires elasticity, NERA works instead by a vapor pressure-bulging effect: the interlayer vaporizes and expands violently outward, causing bending of the metal plates it's sandwiched between, so the incoming metal jet is constantly hitting a different location as the bending progresses. NERA normally uses, but does not in any way require, elastomers for this role, simply because they're easiest to package between metal layers, but a properly engineered battery pack should be able to serve the same role. In NERA, you want as much gas pressure generated as rapidly as possible; the copper plasma jet effectively instantly converts e.g. ethylene carbonate, graphite, etc to gaseous CO2, H2O, etc (plus vaporized metals along with the vaporized steel). The keys that matters are that cells that (A) cells that are in parallel are distributed throughout the footprint of the vehicle (not concentrated in a single location), (B) shared buses create multiple distinct parallel paths between the cells within a given parallel group, and to the next series group; and (C) (required for any NERA) that generated gases are properly vented / handled.

A number of next-generation armoured vehicle designs are pursuing hybrid propulsion.

. I believe the record for most FPV drone hits survived by a single tank in Ukraine is now thirty-two,

The value of 32 small FPV drones is way less than the value of one tank. And in general you're likely to be talking about "hangar tanks" / "turtle tanks" when talking about things like that, but these take on *massive* disadvantages, including being extremely visible and easily targeted by larger drones or artillery (as well as bogging down easily, difficulty getting through confined areas, poor or no gun maneuverability, etc).

it's mostly flat and open.

Donbas is mostly rolling hills, much of it forested. And much of the combat has been in urban environments, which is about as complex terrain as you can get.

cheap FPV drones don't work if the other guy is on the far side the hill you're hiding behind, for example.

Uh, yes they do? They literally fly.