When I see NASA spacecraft doing such extraordinary things, long beyond their expected lifespan, I always feel a little guilty when a simple function I wrote can't even process a test file without crashing!
So, while still impressive, they're unfortunately not working as well as they could be.
[1] https://en.wikipedia.org/wiki/Radioisotope_Thermoelectric_Ge...
Are there copies of these RTGs on Earth for analyst or it was defined by lack of any other explanation?
There are three factors that reduce the output of an RTG over time. The first, and most obvious, is the radioactive decay of the fuel source. (As an aside, potentially this can be more complicated than one might think because as the primary isotope decays it builds up other radioactive isotopes which also produce heat and have different half-lives, so you need to solve a series of differential equations to determine the heat output over time, but for Pu-238 it's actually fairly simple since the main decay product is a much longer lived isotope, U-234) Anyway, the half-life of Pu-238 is about 88 years, so you'd expect it would take about that long for the heat output of the RTG to drop by half.
However, the radioactivity of the RTG fuel source also degrades the thermocouples used to generate electricity over time, and this reduces the power output on top of the decay of the fuel.
Additionally, as the amount of Pu-238 remaining falls the amount of heat generated by it also falls, lowering the temperature of the RTG and reducing the Carnot efficiency of the device, though only marginally.