For a long time I've wondered - why doesn't power scale exponentially upward as reactors get bigger? So, as an arbitrary example, two 100-block reactors make 200 power (100 each) but one 200-block reactor makes 250 power.
This would be both intuitive and realistic, and would overcome the whole problem of multiple reactors because you'd have the choice of two or more small ones or one big one that produces proportionally more power. For that matter, go ahead and put 200 separate one-block reactors on your ship, but it will be extremely inefficient.[...]
That'd be a completely viable and good design decision if players couldn't go wide instead of going tall.
By that, I mean you need to have superlinear power costs in addition to your superlinear power yields, however it's easy building Nx N-times-smaller weapons rather than a single weapon which means you can actually get away with something resembling a linear power cost (while favorising something heavy on the engine btw).
That might be fixed though. For instance, let's assume you have three cannon groups on your ship (let's say, respectively 200, 50, and 50 blocks), and that costs are supposed to be quadratic in the number of blocks (ie C*[Blocks]^2 where C is an arbitrary constant), ideally the total cost of the three groups should be C*300^2, from there it'd be logical to assign a (200/300)*C*300^2 cost to the first group (and (50/300)*C*300^2 for the last two groups).
However, there are a few problems with this:
- it makes each group dependent on each other, which makes things a bit more complex
- if you have both cannons and missiles, should there be two or one global number of modules? what about thrust and cannons? shields and thrust?
- you'd have to count turrets and docked entities in for balance (as long as they can draw power from you, they should also have an increased power cost), which starts to make things very complicated ("my station works fine, until someone docks on it")
Edit: a better/simpler way to "fix" that, but also a more restrictive one, would be to have a hard limit of simultaneous system activations (firing a cannon would have an activation window equal to its cooldown for instance, and only K windows could overlap at a time, trying to activate something if you have already K things activated would simply fail) so that it's useless to go too wide... there isn't the complexity issue anymore (as you don't need to have a global count of each thing and they are somewhat independent again), but the restriction alone might be reason enough to not use that (although it could be a very interesting thing to do, you could even imagine having chambers to increase the activation count of things)