In fact, according to Wikipedia, many of them appear to have masses similar to oceangoing ships of similar length. This gives reasonable results for ships that have a similarly long and thin shape, like the Phobos, which I estimate to have a density of .24 T/m^3 (a.k.a. g/cm^3).

But for wider ships, it's too low. For instance, the Xenophobe worldship is only 15% more massive than the Phobos, despite having nearly 4 times the volume by my estimate. And the Xenophobe ark is only 2.5 times as massive, despite having probably 15-30 times the volume. (It's a bit hard to judge how long the cylindrical section is and whether the bottom is rounded or flat.) The CSC is particularly hard to judge, but I think its mass is probably also low by around an order of magnitude.

I got to thinking about this after seeing this post showing off a ship that's 1000 m long and 9 million(!) tons. That seemed ridiculous compared to vanilla ships, but after estimating its volume I concluded it was reasonable.

Lengths taken from Human Space HD.

assumedpseudonym 18 May 2016:

 Actually, there’s a ship mass guidelines page (which is, unfortunately, not publicly available, though it probably should be) which I consulted for… most of my ships. George probably already had masses for a lot of his ships put in before writing it up, and it doesn’t help that several of them have changed sizes once or twice since then.

nms 18 May 2016:

Does it involve calculating (or getting your modelling program to calculate) the ship's volume?

assumedpseudonym 19 May 2016:

 It does take volume into account, yes. As for accuracy, however, I haven’t poked at that.

george moromisato 20 May 2016:

Many ships have incorrect masses, I definitely agree. I'm updating the masses on an ad hoc basis.

The relevant metric is that, on average, ships mass 0.04 tons per cubic meter (which assumes lots of empty space and very light-weight materials). The guidelines (which I agree do not need to be secret) are really just computations based on standard shapes. But if you can compute your volume, you just need to multiply by 0.04—you don't need the guidelines.

The CSC is probably OK at 200,000 metric tons. Though it is 400 meters across, it is not too high. Also, AssumedPseudonym's 1 km ship is correct at 9 million tons. Since mass grows with the cube of length, even a CSC upscaled to 1 km across would be 3 million tons. AP's ship is fatter, so 9 million tons feels right.

nms 21 May 2016:

That's less dense than I expected, but OK. So the Phobos is way too heavy. The Xenophobe ark might still be a bit too light.

I thought I recalled seeing a side-on view of a CSC after the redesign and that there was considerable volume below the engine block, but I can't find it now.

As for AP's ship, it's considerably smaller in cross section for its length than a CSC. It looks like it could fit (with a bit of dismantling) in a cylinder 300 m in diameter, suggesting a mass under 2.8 million tons at standard density.

But really, getting the volume directly from the modelling program would be ideal.