Indeed, wet electrolytics are a particularly interesting topic.
There was a capacitor plague which was a design flaw causing premature failure in electrolytics. Nichicon, Rubycon and maybe Matsushita and Elna were affected. There was a formula change in the electrolytics in the '80s going to a water based electrolyte that did not go well. The exact period of this production problem is unclear (or the parties are not sharing it); but, the Japanese vendors probably had the problem fixed some time in the '90s. However, it appears that industrial espionage resulted in Rubycon's electrolyte formula being stolen and either the old formula was stolen or there was an error in copying the new formula which resulted in a second capacitor plague in the late '90s. This particular detail is anecdotal since there is no formal legal / documention trail related to the purported industrial espionage. Probably falls into the steal a bad / incorrect design, your bad; but, no prosecution because that just shines a brighter light on the original design problem. Most of these second wave failures were lower cost 'off-brand' capacitors used in consumer electronics and PCs originating from Taiwan.
The capacitor plague failure was typically a pop the can top failure on capacitors that were running closer to the design limits; but, seal failures were also attributed to the problem.
It is possible that some of the electronics on the NSX got caught in the tail end of the first capacitor plague. In addition to environmental conditions, this could be a partial explanation for the high variability around the failure rate. I preemptively replaced the electrolytics on the CCU from my 2000 which was a bit of a waste because none of the capacitors were showing any sign of deterioration. Also, most of the originals were 105C, not 85C capacitors which might represent a production change. When I had my door apart a couple of years ago I had a quick look at the door amps and they showed no sign of capacitor leakage. As I recall, the capacitors that came out of the CCU were mostly / all Nichicon. I didn't check to see where Bose sourced their capacitors.
There is another interesting age-out feature associated with oxidation of the aluminum electrode that occurs when the capacitor is not energized (in storage or not in use). You can find discussion of this if you look up reforming electrolytic capacitors. The oxidation ageing process does not occur when the capacitor is energized so extended periods of non use might actually contribute to earlier failure. You would have to analyse the design conditions to see whether the operating conditions are correct for the capacitors to reform themselves 'in use'.
As a final note about ECUs. Search around and you will find that the Mitsubishi 3000 is noted for capacitor failures which extended to the ECU. I didn't search around to find out whether there are other cars of this vintage suffering for capacitor related ECU failures.
