What is the longest lasting power supply we can practically make for an interstellar space probe?

Factual Questions
21

I was just speculating.

235U does have a very long half-life (although enriched 235U will degrade faster<<

I don’t think that is correct? The 235 isotope will decay at its natural rate regardless of what other isotopes it it mixed with. You may be thinking about the overall decay rate of the mix of uranium we see today?

Thermionic systems would be more robust to even high levels of radiation but would be very limited in computing capability, would consume an enormous amount of power, and vacuum tube computers of any useful size had uptimes measured in hours and were very sensitive to dynamics (shock, vibration). <<

True, though that was in the days when they were all we had. There has, I think, been some speculation about microfabulated vacuum tube systems which could possibly be made today. They would (probably) not rival the density of semiconductor circuits, but might have been a way technology could have developed?

22

Here is a cite about this:

https://www.science.org/content/article/return-vacuum-tube

Now that I look closer, it’s several years old. But a search does show more recent work; it does appear to be something that people are working on, though probably not close to working systems so far.