I found a GE 170W panel for $760. It should end up in a diminishing return situation -- you spend about $1500 for panels to power a high end PC. So, you'd essentially need about 380W grid power to make 340W of solar power. I think on average, 9 cent/kWh is about right. So, there is 5 years of 24/7 up time at max load you've already purchased in the panels. BTW, a low power system like this, if the panel is about $750, would mean that THG's also just bought 5 years of grid power for that low power PC to be on 24/7. (And we haven't priced in batteries
Also, found this in the article: "The reason is that the voltage levels are low, which means the current levels are high, and loss is proportional to current."
V=IR (we'll as-sume (you really need to stop filtering that) no internal resistance and all that good stuff)
So, if V is low then I is low; if V is high, then I is high. You pretty much implied that the resistance of your device depends on what it's doing, which I find wrong.
Loss, is in terms of power. P = IV = I(IR) = I^2*R. So, loss is proportional to current SQUARED.
Perhaps you know something I don't in regards to this (after all, I've only had introductory E&M, and we didn't really do alternating currents, though it looks like you were talking about DC).