If you cannot choose which ship will be shot but can assume the enemy will focus a weak one then the optimal one on one fit is also the optimal fleet fit. *
You want your ship to be tanky so it survives being shot by a bunch of other ships but do not realize that this means you will get shot by multiple ships more, as compared to dumping your flux, because your allies cannot kill a ship they’re ganging up on as fast if they do not dump their flux. Your allies cannot force ships to backup if they do not dump their flux. Your ship cannot force an enemy to back up if it does not win the flux war and dumping flux wins the flux war.
This is not how the AI behaves. It doesn't try to find your weakest ship, it just engages the nearest enemy and does some basic pivoting around allies and enemies to try and flank or retreat based on relative flux levels (as far as I know). I think this is really the main issue here: nothing behaves optimally, and you have to design your strategy around how things actually work (particularly the ship AI, and weapon targeting AI). You cannot make any assumptions about what ships will focus on or fire at, you have to essentially plan for engagements of random subsets of ships from each fleet (although you do have some control with escorts that you can work around). Any ship from your fleet could randomly end up fighting any ship from the enemy fleet, and you want to maximizer the chance of survival in those random engagements.
If my wolf randomly engages a capital ship alone, I want it spend exactly 0 of it's capacity on dealing damage and spend it all on blocking damage while it runs away, because it has no ability to deal meaningful damage in this engagement, but if it survives, it could deal meaningful damage in a different engagement. Think about it this way: if I am currently losing the flux war (meaning that even if I dump all my flux into damage I still overload first), then firing my weapons over dissipation is just increasing the rate my flux is increasing without any benefit. Simply spending capacity on tanking damage with shields cause my ship to survive longer, even if it is 'falling behind' in the flux war, and that extra time is what will allow ships to reposition to change the terms of the engagement favorably.
A simple example:
My ship is locally outmatched in a 1v1 and has no chance of wining the flux war, but it has allies nearby that together would win or at least stalemate an engagement. If my ship dumps all its flux and loses the flux war faster (but by a closer margin), that's much worse than if it just shield tanked and ran to its allies who could cover it while it vented.
Obviously this is a simple example and there are certainly cases where dumping flux would be the right play, but the point is that doing as well as you can in the flux war will not always result in the best outcome, so you can't just say 'this weapon makes me better at winning the flux war so it's better'. If that weapons also leaves you on high flux more often, then you're more likely to end up in random unfavorable engagements where you take damage (or die).
My goal is to minimize how often ships die while still winning the fight.
I choose to design ships that are objectively somewhat worse at winning the flux war in a theoretically optimal scenario because they are less likely to die/take damage due to the random interactions of combat. I might then need to deploy more of these ships to win the same engagement compared to the theoretical best performance of some optimized ships, but in my experience, the expected value of spending some additional supplies to deploy additional 'safe' ships is much higher than the expected value of deploying fewer 'optimized' ships because the cost of losing ships is much much higher than the cost of deploying more. Basically my ships are optimized more for survivability than damage output, which isn't to say I don't try to improve flux war performance, it's just not my highest objective. In my experience, ships designed safely tend to 'stalemate' a lot of situations where they are objectively at a disadvantage anyway because of the AI, so the strategy often outperforms in terms of required deployment cost because my flagship can be damage oriented to win local engagements while my fleet is survivability oriented to avoid losses elsewhere.
Similarly a weapon that did 7500 kin dps and used 6500 flux a second for 28 OP would be amazing.
This is a perfect example. While this weapon in a vacuum with perfectly optimal play would be incredible, it would be extremely volatile and the AI would almost certainly miss while firing it at a fighter or frigate and overload itself for virtually no gain. The AI can target the wrong target or miss entirely, which it does frequently due to weapon inaccuracy, target leading, ship rotation, or range issues. Planning for randomness in weapon usage is also an important part of ship design. IMO, that weapon would be borderline unusable in practice, because while it would sometimes result in instantly overloading an enemy, it would also sometimes result in instantly overloading yourself for no benefit and dying, which is much worse than any positive outcome of overloading the enemy. Using other weapons would reliably result in winning without that chance of dying.