FWIW, junk in LEO can still remain for dozens or even hundreds of years.

Most satellites also don't burn up completely. At least in the US, the current standard is that the odds of a casualty in case of uncontrolled reentry are less than 1 in 10,000. I think that as we see a proliferation of LEO constellations with tens of thousands of smallsats (some of which are almost surely going to be operated by companies that go bankrupt), that standard is going to bite someone in the ass.

“LEO” is vague, 100km lasts months and 500km could be decades but both called LEO.

A 1U cubesat at 100km will burn up in a few hours.

Small satellites have a high surface area to mass ratio and thus decay due to drag much faster. I.e. on the order of a few years at most.

What are you considering small or a few years? Smallsats like Starlink is using can take 100+ years to deorbit from drag in LEO (at the 1100km orbits they haven't started deploying to yet— for the orbits they're currently using the deorbit time is on the order of a few years). Even if you're talking about 1U cubesats (~1kg and 10cmx10cmx10cm), those can take decades. The first cubesats ever built are still on orbit.

Going even smaller, there are still small droplets of coolant in LEO from nuclear reactors that the USSR launched 50 years ago. The droplets are more or less spheres with diameters in the 5mm-5cm range.

This is fascinating. What are (or were) those reactors for?

There were used to power Soviet reconnaissance satellites that used radar to track ships. The satellites themselves were in really low orbits (~200km), and they didn't want to use solar panels due to additional drag and inability to generate power at night. When the satellites were deactivated, they were supposed to eject their nuclear reactors (actual nuclear reactors, not RTGs[1]) to a higher orbit before the satellite reentered. The program had a number of "oopsies". A lot of the reactors leaked, and are actually one of the biggest sources of space junk today. I think they were the #1 source of space junk before China's antisatellite weapon test.

The biggest oopsie was when some of the satellites malfunctioned and weren't able to eject their reactors. One of them landed in the ocean. The other didn't— it ended up spreading radioactive debris across a good chunk of northern Canada. Some of the pieces of debris where radioactive enough to cause serious harm, so it was fortunate that the junk landed in unpopulated areas.

[1]: RTGs are how most nuclear powered things in space generate electricity— https://en.wikipedia.org/wiki/Radioisotope_thermoelectric_ge...

I think 1 in 10,000 is high, there's a lot of land out there, and sea.

It almost seems like you would have to try really hard if you want a 0.0001 chance of hitting a person on deorbit.

I think one death total, to save megabucks on each of hundreds or thousands of satellites, is an OK situation.

The title is specifically about GEO satellites

Compact GEO.. GEOO..... (sigh) good point. So they're not in unregulated space (does that even exist?) and have to have a de-orbit or park policy and understand how to effect it, and are governed by the economy they launch from and the territory they fly over, and the risk of litigation from other GEO owners affected by their policy.