Here are my thoughts:
1. There is no reason to use disposable fleets in Vanilla, because there is no mechanic to replace them without losing large amounts of time, even if we leave cost out of the equation.
Taking losses means losing time, and not in a fun way; when we're going to have to "defend an Outpost", if we lose, we're basically hosed; there being no way to replace the losses means that even if we have a million credits (as I've had, every time I've played Vanilla since the Supplies change, within three hours) there's no speedy way to spend it and recoup the real losses in specific weapons and ships. I think it's very problematic that I have money I can't spend, ever, in a game like this.
Uomoz's solution is intriguing in that regard, but I think relative cost matters, too; if it's cheaper / faster / more efficient to build big fleets that I know will take some losses vs. tanky super-fleets that don't but cannot afford a catastrophic loss... I'm going to do the former.
In Vacuum, I largely did away with the delay; if there's a ship you want, you just need to know what Faction makes it and you can obtain another one, pronto, assuming you're on good terms with the Faction. This, and having plenty of money by midgame onwards, if one plays smart, makes it possible to largely ignore this and take losses willy-nilly if it seems like a reasonably-fun thing.
So I don't mind taking losses or losing a battle- but in Vanilla, I really mind that I've lost the last five hours of my life and it's indeterminate how long it will take to replace those losses. It's not like taking losses in a RTS, where either that means I just lost the game or I'll shrug and order up another 100 Peewees or Marines; given the open-world nature of the game, it's mainly about time loss, and time loss isn't fun.
Coupled with the "slow movement after battles" and other factors... this factor leads, very naturally, to extremely conservative play. If you don't take losses, you don't have to travel to Askonia and hope that the ship you need, given your DPs, to fill a specific role, will exist. This is similar to Mount and Blade, where all of the serious players who wanted to conquer Calradia gravitated towards armies of the most-OP troops for sieges and avoided open-field battles like the plague. Losses there took serious time to replace as well, due to the need to re-level them up to the point where they were useful, and it encouraged a conservative attitude.
I think it's even worse in SS, though, because there is no sure way to replace a loss, especially of a rare ship type. If you lost an Astral, you may have to spend hours getting one again, unless you're farming Tri-Tach just to get one. That's not much fun.
Given that "Askonia" as a concept is going away, this may or may not get worse, depending, to a large extent, on what replaces it. If we can start placing orders for Autofacs to construct fleet replacements to replace losses, then the sting of loss is considerably less, so long as it wasn't a major strategic disaster; if you've managed to save that hapless Independent base from the scurvy Pirates of the Long Reach then that Hammerhead with custom gear you lost can be re-ordered and all is well- you've lost money, but not much time.
However, if things remain largely like they are, in terms of access to ships, or become even worse, in terms of time-investment per ship constructed, the opposite will happen; each battle is risk, risk might mean defeat, if not in detail then in irreplaceable time, and the tendency to only pick the sure thing unless there is absolutely no other choice won't work.
In terms of creating situations where there is absolutely no choice, there are mechanics problems atm. Basically, fleet movements are pretty fast, and there is no warning given to the player that a given location is a target; if that's going to become a Thing, then players need to have good warning, or it won't be much fun. Camping your Newbie Outpost and intercepting threats will be thrilling for about an hour, but after that, you're basically cutting yourself off from further adventures and expansion. At a certain point, either your little empire's somewhat self-defending vs. anything that's not a huge threat, or you're stuck.
In Vacuum, I made this work by making it quite difficult for the AI fleets to capture / destroy bases, and prevented them from wiping each other out, since that largely just helps the players; this kept things pretty simple, and players don't have to spend much time on forced defense. I'll try adding the occasional battle that players cannot avoid into the next build to see how that feels, though; it might be fun.
I thought that another fun idea would be to let the player take control of battles between their forces and that of the enemy, if they wished, becoming directly involved in the outcome and assuming a more managerial role.
2. Ship and weapon balance and relative costs matter a lot.
There is no reason to use anything smaller than Cruisers unless you're trying to solo a fleet, Megas-style.
Frigates and Destroyers don't have a point in big engagement, if handled by the AI; the difference in real capability vs. cost makes the bigger ships overwhelmingly more cost-efficient. This has a lot to do with the DP costs of ships; if an Onslaught costs 50 DP to deploy, I'm going to think twice about using it vs. a scad of Destroyers who cost 5 apiece.
The DP system doesn't have quite enough granularity (imo) in that regard; deploying a Wing vs. a Destroyer is usually a no-brainer (the Wing, unless it utterly sucks, because it's usually a lot more DPS in total and it's immortal). I've thought about playing with that in Vacuum, using a different range of DP values and splitting the ships much more widely based on how powerful they actually are... the issue there is largely one of making sure the AI can deploy more forces than the player, which isn't a big deal, and adjusting the DPs so that the gulf between lots of cheap ships and a few powerhouses is enough that the cheap-ship fleet is tempting.
I know there are Vanilla players who play with big fleets, but I agree wholeheartedly with Megas's analysis that it isn't efficient; it's cool and it's pretty, but would a serious player trying to be efficient use it right now? No.
Moreover... until it's not attractive to solo-kill fleets vs. using big disposable fleets, it's going to be the solution for practically all serious play. I think I've got Vacuum just about cured of that problem, finally, but it took quite a bit of rebal to make it happen. Vanilla's current balance remains at a point where I'd say that the frigate-fleet-of-doom approach is most efficient, followed by the Multi-Medusa Horde, and that is kind of worrying, because once again, the emphasis is on players flying solo, no losses, extremely conservative play.
In addition, big fleets run into some technical issues with the engine; viewing the Fleet screen when there are lots of ships is quite expensive computationally, to the point of lagging when large numbers of ships are present, largely because of engines rendering and other things like that, so far as I can determine.
One solution to that would be to render the ships with gear and all that to a FBO when the UI opens and then just render that result; it'd look exactly the same but be lightning-fast. There are other issues there of a UI nature, though; if really big fleets are the norm for high-end fleets going the cheap-DP route, then tools like being able to sort the ships becomes important, amongst other things. And there is the obvious tension between DPs and the total cost in Supplies as well.