First off, I'm not advocating for low reliability in favor of stats, but I need to point out that this math analysis is off.
If we are designing ships for a 30 year life, then every ~3% is going to mean one year of needing repair, and 1 drydock per 30 ships.
3%
failure repair rate means 95% reliability, due to 60% chance of at least serious failure upon hilarious breakdown event.
Also, starbases can do some minor repairs, and as research is pored into them, they'll be able to repair more.
edit: Also, it's likely that hilarious breakdowns are themselves events that can be passed, so the actual serious failure rate is probably lower.
An extra stat is going to give a +1/36 to +6/36 depending on how hard an encounter is. So consider it about +3/36 per stat, and if each stat is a 1/6 chance, then a plus one gives +3/216 or +1/72 bonus.
This isn't correct and it's way too simplistic. There are multiple rolls involved in an event, and there are at least two that are affected by the ship's stats: the response roll, and the event success roll(s).
Information we know so far (AFAIK):
Sci-Fi - To Boldly Go... (a Starfleet quest) | Page 301
Sci-Fi - To Boldly Go... (a Starfleet quest) | Page 302 (analysis)
The response roll for an unplanned event is a (2d6 + <defense> + <presence or science>) >= 3d6. For a measly Constellation, this is 2d6+5>=3d6, which is ~70%. Incrementing one of the relevant stats results in 2d6+6>=3d6, which is ~78%.
The event roll is 2d6 + <relevant stat> >= difficulty DC. Difficulty DC is highly variable, but the example shows a hard DC of 11 for non-EC-ships like Constellations. For science hard difficulty event, this is 2d6+2>=11, which is ~28%. Incrementing science results in 2d6+3>=11, which is ~42%.
This is all still a very incomplete analysis because it depends on the ship, all the possible difficulty DCs, how the event handles multiple ships, etc., and we don't even know what the response roll is for a planned event.
But clearly, even a single stat increment can have a large impact.
(Note: I used anydice.com for computing the probabilities.)
Now with all that said, I think explorers should have at least 98% reliability, because the EC explorers are guaranteed an event every quarter, the majority of which will net significant benefits, and sometimes EC explorers are sent on critical missions, like
Nash's deniable super secret mission.
For cruisers, I'm fine with 95% reliability, escorts probably lower than that, because of lower chances of sector events. This however depends on having sufficient number of starbases and berths for repairs, scaling with the number of ships.
Warp core and hull reliability obviously need to be
minimized maximized to prevent outright ship loss. 99-100% for explorers, maybe 98% for other ships.
Also, the whole "trade reliability for stats" point may be moot if we're aiming for canon ship designs. In those cases, we have to meet those stat lines to get all the vague benefits of canon designs (the only concrete known one being the reduced pp cost). So in those cases, it's only a question of sufficient reliability.
edit: fix stupid errors
