Epoxy offers two advantages -- it is an electrical insulator, and it conducts heat better than air. Transformers and inductors are generally potted with epoxy for this reason. [[ref][1]] Perhaps the cause of the high failure rate is that Commodore engineers decided they could use cheaper components to build the power supplies, depending on the epoxy properties to balance out any deficiencies.

In 2012 Bil Herd (Commodore engineer) [speculated][2]:

> They were made for CBM by the boatload, they got warm and were not rated for things like the CBM cartridge *[Bil probably meant CP/M cartridge]*, etc, though they would continue to make voltage they would just run hotter reducing their life.
Potting them was probably to protect CBM as it's hard to start a fire from within a pound of epoxy. The potting would have made the hottest components slightly less hot and everything else too hot.
 
>Remember that the VIC 1 had started at least one fire, I think something like 3. The case used to melt and sage over one of the heat sinks. Supposedly a programmer tried to design apiece of the power supply without really knowing heat and wattage calculations. So a good reaction would be a supply that never caught fire, was impervious to static and even water, and was a bitch to air ship.


 [1]: https://en.wikipedia.org/wiki/Epoxy#Electrical_systems_and_electronics
 [2]: http://c128.com/did-commodore-64-use-surplus-power-supplies