> Whether the SSD is powered or not during the long periods where you're not accessing the data has minimal effect.
I've heard conflicting reports on this.
Ultimately, any storage technology is about storing data. This can be done magnetically (aka: bits are tiny little magnets pointing in certain directions), as long as the magnets don't change, you can read back the data fine.
Or it can be done optically (ex: the dye inside of CD-roms, DVD or BluRay). As long as the dye remains a certain color, you can read it back fine.
Magnets lose charge over time, dyes can change color (especially if they're exposed to the sun). You won't notice over a year or even 2 years, but after 5, 10, or 15 years, things change enough that you start getting errors.
SSDs store data by punching an electrical charge into a transistor. The barrier is such that the electrical charge doesn't want to leak out: but... what is the rate of that leakage? All materials "leak" charge over time. And if the voltage changes dramatically, the change-in-voltage will change the data (and thus you'll lose data over time).
--------------------------------
Some SSDs, such as the Samsung 830, were well known for leaking charge and causing issues. Today, we use stacked 3D NAND, which is a different transistor type than the planar NAND on the Samsung 830, but that doesn't fundamentally answer the question of "What is the leakage current" and "how long do we have before enough voltage leaves the transistor to corrupt the data?"