When I first read the Nova rules, I started wondering about armor. It looked like a nice, straightforward defense, so I ran a little math the next day to see how the shipbuilding numbers work out. I came to some conclusions that I thought it might be best to keep quiet. The forums (namely BeowulfJB and a few others) reached similar conclusions recently, so I figured I might as well share here.
In short: the trouble with armor is that it costs the same as buying more hull, but it provides you with very little benefit over just buying more hull. Having a larger hull provides you with more of those delicious SU, whereas armor actually costs you SU. As a result, if you want a ship that can take n hits, you're almost certainly better off going with just hull n rather than hull m and armor p such that n = m+p.
There are, however, two possible uses for armor that I can see. The first has to do with thrust. A hull n ship has a higher thrust factor than a hull m armor p ship. How much higher? Thrust factor scales as hull to the 1.7th power, so thrust t on a hull n requires 2*t*(n)^1.6 SU. If you look just at the case of a hull n/2 armor n/2, then the armor consumes 30*(n/2)^.7 SUs, though for the same thrust you'll likewise need to use 2*t*(n/2)^1.6. I did some tables in OOCalc, and I'm having trouble creating useful graphs of my findings since they're in three dimensions and relative, but it is possible to generate an SU savings via armor. The trouble is that the circumstances must be somewhat extreme; for example, if you really, really want a ship that can take 68 hits and has thrust 8, you can save 222 SU by making it Hull 34 Armor 34 compared to straight 68 hull. But, for 4, 8, and 12 hits, half-armor-half-hull generates no SU savings for arbitrary thrust. This is also true of any multiple-of-four number of hits less than 68 for thrusts less than 8.
The other thing armor does is that it lets you ever-so-slightly manipulate the way your systems degrade. For example, a hull 5 / armor 1 ship can take 3 hits before being damaged, then another 2 to crippled and 1 more to destroy, while a hull 6 ship takes 2 for each step of damage. Likewise, a hull 4 / armor 4 can take 4 hits before being damaged and 6 to cripple, while a hull 8 takes 3 and 6. This lets you front-load your damage, and retain fighting effectiveness for slightly longer at the price of a more rapid degradation once you hit damaged or crippled. However, as you move to larger hull sizes, this effect is diminished; you'll only ever be able to front-load one extra point, which is much more significant on small ships (where armor is basically guaranteed to not generate an SU savings on thrust). I think this may be a worthwhile use of armor on smaller vessels, though; may have to try it out. But for like... a ship that you want to take 16 hits and have thrust 5? Avoid armor. You could go hull 14 armor 2 for front-loading (6 to damage, 12 to cripple, 16 to destroy vs 6 to damage, 11 to cripple, 16 to destroy), though it's not clear if that's really advantageous, and but other than that there seems to be little to no advantage to be gained from it in the majority of cases. Flavor / conversion reasons naturally remain valid.
The really annoying thing is that this is, to some extent, true of ECM and shields as well. These both have actual advantages, though - ECM can turn long range fire into no fire, and shields can beat the odds if your dice like you. On expectation, though, both of these will likewise increase your effective hull by some fraction, but use up your SU, generating an effective hull n with less SU and higher potential for thrust than a hull n should have. I have yet to actually run the numbers on them, though.