"Solved it. Deleting."

The "And?" is that an LLM by definition produces low-entropy output. For spicy autocomplete this is not just acceptable but desirable. But it does mean you will not get novel responses.

They want to avoid consuming AI output as training material for their AI. They don't want to pay money for classification. This way they can get users to do the classification in exchange for the hope that it will be hidden.

Except prehistoric productivity growth was nearly zero. No one got rich using those methods.

Instantaneous power is the most important metric when you are designing an system for instantaneous power support. The use case is when there is a failure in the local power plant and you need to keep everything in the data center up for 10 minutes to get the hot standby running and synchronized. It doesn't matter how long you can run only half the data center.

Abuse of employees wastes company time and doesn't improve performance. Even doing nothing has a better ROI.

Deming repeatedly noted that main factor in selection of "employee of the month" is either statistical variance or a process not under statistical control, and the individual has little to do with it. The same applies to the bottom employee as well as the top. This is just as wasteful as stack ranking that requires firing the bottom n% of staff.

But the median businessman has no idea how capitalism works, and at the 90th percentile they are actively opposed.

The problem had its origins in the Civil War, where the Union war bonds basically invented a consumer market for financial instruments that did nor previously exist. The demand they created soon could no longer met by the government so the US would end up essentially backing railroad company bonds. The railroads were allowed to issue bonds backed by track yet to be built that were convertible to government bonds. The government imposed requirement that a railroad could only borrow up to their market cap meant that the railroads could keep issuing more stock which it gave to the construction companies formed by the people that ran the railroads who priced their work to consume all of the money coming in to the railroad.

In some cases the government loaned actual government bonds to the railroads to sell. There was a court ruling that the law only required the railroads to pay back simple interest on the bonds, not compounded semiannually, on top of the principle and only pay it to the government at maturity.

As you note, the government subsidized the railroads with grant of land... which the government did not actually own.

And this is just what happered while they were still building the railroads, not running anything

" it was just because the capital investment required was massive and the payback didn't come as fast as the financial structures built to fund the construction required."

The real problem was that much of the financial structure was bogus, propped up by government shenanigans around the financial instruments rather than addressing the industry that the instruments supposedly represented. The required payback was never going to be possible.

The *useful* assets were snapped up, but a lot of it was garbage.

Did you ask it to integrate new findings?

"These are completely different things."

Unless they're not. Correlation may not be a *sufficient* criterion for causation, but it is a *necessary* criterion. Lack of correlation would mean you could stop looking causal relationship. There may still be no causal relation, but it would be absurd to reject on the *basis* of correlation.

Sure if the events were independent and had a uniform distribution of periods, you'd still expect *some* to be close to each other. See "Birthday Pproblem": Given a sample 23 people, there's a 50% chance that at least two people have the same birthday. However, half of the sample being close to each other randomly is significantly less likely and it decreases as the sample size increases.

" When you take the average of *ANY* set of numbers"

-1, off topic. That's not their analysis.

" If your hypothesized pattern of geologic activity also explains the random trajectories of interstellar debris, it has no real predictive power."

If it makes predictions that match reality, it has predictive power.

You are assuming that that geologic activity and trajectories of interstellar debris are independent, and rejecting out of hand any root cause that could affect both.

But even before that, you are assuming that trajectories of interstellar debris are random and rejecting the existence of gravity.

Correlation is not a *sufficient* condition for causation, but it is a *necessary* condition.

"happen somewhat randomly with a normal distribution"

Why would it be normal rather than some other distribution? That's actually significant!

"There is nothing special about 37 million years."

It's the mean of what you claim is a normal distribution. That's actually significant!