But that's the whole point... Making ships obsolete is fun. It gives a sense of progression, the thrill of growth.
I think that's where we disagree. I'm going for strategy here, not wow. In chess, a pawn always remains relevant, so does the knight, bishop, etc. They all serve a purpose.
My reference games for this project also emulate this very well: weak scouts, that you can manage to build early in game, remain useful at the end. Even some smaller warships, with better mobility, tend to be there until the very end.
In strategy, an ambiguous choice is defined by the fact that its neither fundamentally a good or bad choice. Its really up to the player to make this choice relevant or not.
In this case, building such ship or such other ship isn't specifically a dominant strategy, its an actual choice. To make these choices relevant, it needs to correlate and support the player's plan, and hinges on the success of that plan.
It is possible that the player's plan is flawed, but in essence, all ships are 'worth building'.
While some will clearly cost more, you'll get what you've paid for, and for certain encounters, having one big ship at the right place will create an advantage: for example, the Death Star could kill a bunch of massed X-Wings, and could for this encounter by moving towards a critical planet, forcing the enemy to concentrate his forces, against his will. Since the Death Star could be made in such a way that it can really eat through numbers of light/mid ships, in this particular encounter, the Death Star would be better than the X-Wings.
Assuming however that the player chooses to let the very slow Death Star claim this planet, despite its value, and instead spreads his X-Wings to take over a dozen of loosely defended worlds at the same time. In this case, its quite possible that the X-Wing player would have gained an important economic advantage, or at least minimized the lost of his planets, and, hasn't lost a single X-Wing in the process.
The example is a bit cheesy, but you probably see what I mean by this now.
I guess that's probably why we also disagree on the research aspect of the game ;)