This is a truly awesome post, of the kind that the web should have more of. And, of course, in keeping with all internet traditions, having lavished praise upon the post, I'm now going to rip holes in it.
Drum's quick calculations ignore two very important aspects of Death Star construction. First, assuming a constant increase in GDP over 500 years is all well and good, but it fails to take into account that over those five centuries, steel will become more expensive, not just due to inflation (which I assume Drum is accounting for by assuming a "real increase"), but due to the increasing scarcity of the metal itself. Even with a super-advanced recycling system, we'd start to face some fairly steep rises in the price of steel once we start building interstellar fleets.
More importantly, Drum has come up with the estimated cost of building 0.01% of a Death Star on ten thousand different planets. Unless he's hoping to launch those pieces at pre-arranged times calculated to allow the entire set to arrive in the same system thousands of years later, to then be assembled using superglue and sellotape, I'd suggest there are some fairly major costs that are being waved away here.