Childcare is the one that comes to mind, but really anything with enough value that doesn't have an explicit exemption will be treated as taxable income on its value.

And then an AI to fire the guys who signed off on the AI signing off on the decisions middle managers that signed off on the AI signing off on AI code output.
I'm very interested in LLMs, and have been using them in all kinds of cool places... but seriously, this shit is just clown shoes.

No, services, for example can be a compensation, and no you do not own them.
What's important to remember, is that you are taxed at the fair market value of the compensation- regardless of whether you own it.

Pretty solid pondering.
I'm inclined to agree as someone pretty deeply embedded in the bullshit currently, 20 years into my career as a software engineer.
Though only thing I'm not so sure about is the large paycheques part.
Mostly, we've been looking at this semi-sentient codegen operators as disposable, and paying them disposable wages.

Not as long as the SOTA models retain a significant advantage over the open models- people will keep paying for them.
And they do, and every time it looks like they might eventually fall behind, they leap forward.

The real question isn't about the inference- it's when do the SOTA models hit the wall of what it costs to train them. That's the magic that the bubble is fueling. Inference is a legitimate commodity. People paid for compute before AI, and they will after AI. But nobody at the front of the pack has figured out how to make training pay for itself.

Correct. Because it's all that matters when computing an orbit.

12 to leave the solar system. 30 to reach the sun.

No, speed is a reaction to force. I can give you the values in force if you like, but they're less helpful. Orbital parameters care not what amount of force it took to move something (which is dependent upon the mass of the thing you're moving), only its velocity.

I'm not getting through to you, but you fundamentally misunderstand how orbital mechanics work.
You can't "drop" something into the sun. You have to decelerate it from the current velocity of the Earth by enough velocity for it to fall into the sun. That value is 30km/s. Decelerate it any less than that, and it will not hit the sun, it will merely enter a highly elliptical orbit around it.

How much force will it take to move an 11kT rock into an orbit that will intersect the sun? 20,000 Starships of force, because each Starship can only accelerate the rock by ~1.5m/s.
It's all math, man. Your intuition is leading you to misunderstand how it all works.
If you'd like to try to learn about how it all works.

Sorry, this is simply not true.
You can stretch the argument to absurdity, of course. But deterministic behavior does not imply "can survival all problems", it merely means "reacts to them as programmed, in a deterministic fashion".

Computers are deterministic.
Even in the case of an unrecoverable ECC error, they react deterministically. Hell, that's the entire point of ECC memory, so that the machine can react deterministically- i.e., it can know when memory has been corrupted, and react accordingly.

External factors are not non-determinism.
They're simply uncontrolled factors.

Unless you're using an OS that doesn't.

Unless it is.

Won't happen to you in your lifetime (uncorrectable).

Absolutely. And all things can be accounted for, and corrected for. Things that aren't corrected for are not an indictment of the determinism of your programming language or computer, they're an indictment of your programming.
And that isn't meant as a slight- in 99.9% of cases, correcting for all factors simply isn't necessary, so isn't done.

Hyperbolic or parabolic, sorry.

And just for completeness (and shits and giggles), you could kick a small space rock out of the solar system using only 8,000 doubly-refueled Starships.

I explained this to you.
It requires more energy to make it intersect with the sun (over twice as much) that it does to make it leave the solar system.
And even the amount that it takes to leave the solar system is more than all of the rocket fuel mankind has ever used for even a small space rock.

Delta-v is the measurement of how much change of velocity is required to transition one orbit to another.
To create any orbit that impacts the sun, it requires a minimum of 30km/s of delta-v.
To create an orbit that is hyperbolic (never comes back), it only takes 12km/s. It would make no sense to try to put something into the sun.
As explained before, even the 12km/s is far too much to move a space rock. We simply don't have the capability.
Starship, refueled entirely in orbit, then also refueled entirely at the asteroid, then using 100% of its fuel to push the asteroid would impart 1.5 meters/s of delta-v on that rock.
Only 20,000 more Starships to go before you've put it into the sun.

Armageddon-style is a mixed bag.
For a truly large impactor, it won't really make a difference. The planet surface is going to burn either way- either by the atmosphere being raised to the temperature of a broiling oven across its entire surface, or it spreading out from a central impact point.
For small impactors, blowing the thing to fucking hell can make a lot of sense, since it doesn't have enough mass to cause planetary scale effects.

Computer programs are 100% deterministic. If they fail to be so, then your CPU is broken.
Don't confused "these interactions are effing too complicated to fully grok" with "sometimes 1+1=3."

Well, to be fair- there are different kinds of determinism.
Classical physics are deterministic in principle, but not in practice.
I.e., there's enough thermodynamic noise and chaotic interactions that even in a deterministic classical regime, you have to zoom out quite a bit for things to be deterministic.

Quantum mechanics have a fundamentally unknowable amount of uncertainty. In the Copenhagen interpretation, that's handled stochastically- i.e., we model that uncertainty with random numbers.

Transistors, though, at least in the regime of a CPU? Completely deterministic, just as neurons are almost certainly deterministic.

Purely apologetic argument.

It is a crime to subvert the DPA like that, punishable by a year in prison per offense.

If Ford and GM has built defective tanks during WW2, people responsible would have gone to prison.

This is a punitive measure.

If you're curious about the exact numbers, it takes ~30km/s of delta-v to drop something in the sun, and about 12km/s of delta-v to send it out of the solar system.

Applying even 12km/s of delta-v to something with tens of thousands- up to millions of metric tons of mass would require more thrust than all rockets mankind has ever launched, ever.

It requires more energy to dump something into the sun than it does to send it out of the solar system, and sending it out of the solar system is far more energy than we can put up there.

Hence the "It can't be eliminated."