China or Russia do not move their red lines. US concerns about TikTok were always exclusively the lack of Zionazi supremacist control over the messaging. This is what is getting "fixed" by buyer group. China will also receive other unrelated trade concessions, though falling short of "normal", they will do better than "allies" who have no red lines on bending over.

I'm not speaking for anyone else, just solving OP's truth with UBI/Freedom dividends as the solution. Free and fair markets are not evil. Complex systems to have random people enslaved to be linemen for a day seem categorically unworkable. A system that ensures enough linemen willing to receive great pay to be linemen is workable.

UBI/Freedom dividends means minarchism power redistribution. People need to be well below idiocracy intelligence to not prefer higher dividend to demonic warmongering budget. I disapprove of ultra centralized allocations, if only that any socialist, or other idealist, win to implement it, in an Israel first fascist media environment, leads to right wing fascist takeover of the centralization. US corrupt political/media system requires disempowerment. UBI/Freedom dividend only election platform is only possibility of ending the corruption. It's much more important than voting itself.

There's a Tom Cruise movie of historical value.

The plane was registered to Zambia

checks out. https://en.wikipedia.org/wiki/South_Africa%E2%80%93Zambia_relations

Huwei and SMIC are iterating too fast. As they get good enough yields on current/last generation, they move up/announce new generations that freeze a bit of demand. I think they are doing this to get good pricing for SMIC and local designs. It doesn't help that all the Trump sycophants are calling him brilliant for enslaving China to "NVIDIA's 4th best chips and software stack".

UBI removes the slavery. If you, or strawman, insists on a labour supremacist society, rather than a shortage of willing linemen, there might be a shortage of employers investing in a working power distribution system. There are many policies in between our current supremacist slavery that eliminate the structural slavery.

You need linemen? Call enough people to train them with promised starting salary when they complete training. Or pay them some salary during training. Or increase the promised starting salary. Enough people will say yes if you keep improving offer.

Your examples are already "good jobs" relative to say roofing (statistically most dangerous occupation).

ROI exists/dominates investment decisions under private or social/communal ownership. Let's say your country imposes a fixed price to pay you per bushel of wheat. The decision to buy a tractor is completely independent of how equal the ownership in the farm(s) is.

Adam Smith's justification for capitalism relied on an imaginary "perfect information" mechanism. He did see some centralized governance need to prevent the market corruption of monopoly/cartels, but was relatively silent on need for regulation (say health inspections of restaurants) as a way to avoid food poisoning, or everyone magically knowing that a food poisoning victim was poisoned at a specific restaurant. Though, its possible to theorize that restaurant inspection can be done by multiple private/social organizations. none of the options are immune to corruption if there is a "free market" for politicians/bribery.

Not enough people want to deal with waste systems or sanitation yet those are critical to any society.

These are often highly paid, highly desirable, union jobs. Many have missed OP's point that the coercive slavery that exists in our society is that the threat of starvation/homelessness means less power for individual labour vs employers or vs competing with employers. There is some $ offer that will get me to unclog your toilet.

MoE

This optimization is about smaller matrix multiplications. Experts will specialize on input token types, and while it is better at being split accross resources (GPUs), it is not really specialization on "output domain" (type of work). All experts need to be in memory.

Deepseek made a 7b math focused LLM that beat all other models on math benchmarks, even 540b math specialist LLMs. More than any internal speed/structure "tricks", they achieved this through highly curated training data.

The small models we get now tend to just be pruned from larger generalist models. Paper/video is suggesting smaller models that are "large tuned" post trained to be domain specialists. Large models could select from domain specialist models and only load those in memory or act as a judge in combining outputs of "sub models"

Where an LLM is a giant probabilistic classifier, there are much faster/accurate/less compute intensive deterministic classifiers (expert/rule systems). Where SLMs have advantages, using even cheaper classification steps is going in the same direction. A smaller LLM is automatically a faster classifier, as a hammer to bang on everything alternative.

If the country was "fixed"/doing awesome, forecasts for 3 fed cuts by December wouldn't be current. Labour Statistics department head wouldn't be fired for giving bad news.

The bigger problem no matter economic health is "our hate speech is free speech. Your/any criticism, or just concern about the right of criticism, is radical left damaging the country"

to be fair, I'd assume the same meaning. Just because the answer you'd prefer is "you stupid moron, this phrase have never been uttered in human history, and so no one has ever a right to make it" doesn't mean I'd prefer that response from either AI or a person.