Don't we just do:

worst_code_i_ve_ever_written.pl | openai "fix it"

Mind you, not fixing the code. Fix the output. Much more expensive.

That would mean "Tesla sales drop of 50%" news is in fact financial engineering to mask that it's in fact half an order of magnitude worse.

What about the criticism that China just won't do that and that the people you're asking to do this have a lot of power, but don't have the power to stop this?

Nobody is even still criticizing China for what they're doing in Xinjang, so this? Not a chance.

Make a compromise! Left or right, YOU want a powerful, prosperous US built on immigration, for education and for labor, sufficiently restricted to protect American jobs. That's where the US has always gotten it's "power" from. YOU want carefully monitored, safe, supported safety from Polio, TB, ...

Better autistic than paralyzed, and with better education, autism can be helped.

You can go along with MAGA a bit. There is a history. Early vaccines, were pretty bad and did a whole lot worse than cause autism, so yes, vaccines should be monitored. There are worries. Why disagree with that?

The key here is to see that we aren't far apart. We're far from fascists like Trump and Musk AND SO IS YOUR MAGA FAMILY.

Why have a discussion about "vaccines"? Of course that leads to the discussion breaking down. Take the Polio vaccine. If a Polio vaccination causes autism, that's an improvement!

Obviously education needs better support
Obviously Israel needs to be supported and protection from Palestinians
Obviously we need to support Ukraine and attack Russia economically (for Ukraine's sake AND for the climate's sake)
Obviously we need to prevent mobs from attacking Jews in the US

The danger to Trump (who, you're quite right, plays along with Putin and middle eastern islamofascists and other authoritarians) is that you'll go out to your "MAGA family", figure out a nice compromise agenda.

It's trivial to see even what such an agenda would be, in high level terms.

Dictators, terrorists and authoritarians really do believe the big weakness of "the West" is the massive divisions in the west, whether left vs right, immigrant divisions, rich vs poor

The danger for them is not that you'll convince everyone left or right talking points are right. That won't happen

Why? Well, one of the big peddlers of misinformation is Putin. Any reasonable agenda will be very anti Putin, for a dozen different reasons. So he wants to sabotage the very forming of an agenda, of a compromise. So do a whole lot of other authoritarian regimes, middle eastern ones, China, ...

The goal of the sponsors of this is to prevent a reasonable agenda from getting wide support. You see, in reality Trump and Biden don't differ much. They both want a powerful economy in the US, built on immigration, prevent the rise of China, and both want Israel to exist as it currently does.

NEITHER party actually does what they preach. The left, when/where in power, does not sabotage Israel and protects Jews and the right, when/where in power, does not restrict immigration and protects immigrants. To give US examples: Trump did not immigration quotas. Biden did not sanction Netanyahu.

The goal of doing this is not advancing either agenda. It's "attacking the center", as was done before and during WW2. The goal is to create 2 parties that WILL NOT compromise, because they cannot. The left is constantly pushed to antisemitism and the right to anti-immigration. BUT ...

I don't understand why this idea keeps getting pushed. What Trump and Putin and ... are doing is "attacking the center". For example, if you haven't noticed Putin supports Hamas and Hezbollah, both with weapons and know how AND racist EU anti-muslim parties (and Trump's agenda, if you must)

Recessions are always preceded by contractions in the money supply. Trump's "big, beautiful bill" (double social security for the rich) expands the money supply by ~2%. EU will arm Germany, expanding money supply by 4%. China is desperately expanding it's money supply.

An economic boom is coming.

If you think network speed and latency for teleoperation are tough ... wait until you quantify what you need for AI.

I was calculating last week, and these absurd figures keep popping up. 100kw PER RACK. 500 Gbit/sec disk read/write PER RACK. 200 gbit PER MACHINE network ...

It keeps coming.

Well, border check areas in airports are almost designed to spread diseases: lots of people, lines (strangers VERY close together), badly ventilated and in most of America: humid.

The videos about how hard they're working to "keep you safe from disease" are a nice touch.

They Criticized Musk on X. Then Their Reach Collapsed. Three users who disagreed with the site’s owner saw views for their posts plummet.

Right-wing accounts criticized Musk on X. Their engagement tanked practically overnight. www.nytimes.com/interactive/...

It's telling to investigate exactly what scientific management is, what EXACTLY Taylor researched: his claims are:

1) management does not need to be experts (in the work)
2) neither do workers (management will solve it and describe, down to individual hand movements, what to do)

This is bullshit.

The origin of scientific management was fraud.

Frederick Winslow Taylor was a fraud, he falsified data. He didn't care, published books on it. And, the people who taught his theories knew he did that.

archive.ph/c4ETs

And yes, this "detail" keeps getting erased from Wikipedia. See the history.

And what about distillation? Chinese models will never respect this, or tell you, so what if a Chinese model uses your book and I use the Chinese model as a "teacher" for my model?

Because if we're talking about humans, this is free.

Sure, but for starters, it's hidden. You don't see it when looking at the spreadsheet, you have to click around.

And then everything in the table is A5 + B88 - C3 * D88 with indices evolving in different directions.

Python notebooks with a pandas datatable pd.show() at the end, that's the stuff.

Issue is sharing knowledge ...

Time I need to figure out what a pivot table by someone else does: half an hour, minimum. And I often make or don't see mistakes.

Time I need to figure out what a python notebook does: 5 minutes, maybe.

And GUESS what they're showing in the video: "amazing" recovery from balance loss, after tricks that are clearly not balancing correctly (robot rotates after jump). And if you look carefully ... they recover by rolling the wheels!

This thing can jump, BUT it cannot control where it's jumping!

You might look at the math though. Wheeled legs (powered wheels) are 10x simpler than just legs. So it's cherry picked in the design of the robot.

Essentially wheels on the legs mean you can use an algorithm to balance the robot that's older than me (I'm over 40)