On harvester shields: how OP could they actually be, even if completely free and automatic? Not that I'm at all entertaining such kindness to the player But possibly some simplification: have them cost some moderate amount of K (1000 or 2000) such that you'd be rewarded for not needing them at all, and just balance their energy cost so that the direct m/c cost (which is an extremely rare mechanic) can go away. And if you're running them off the ZPG and paying almost nothing for them, well, yay for ZPGs.
Or is that just a really bad idea?
Does the AI actually target harvesters with any frequency? I've seen them pop a harvester as a drive-by on the way to a real target (Command Station, wormhole, FacIV, etc), but I can't remember them ever going out of the way to target a harvester unless there was literally nothing else to chase.
Under those circumstances, I don't think I'd be spending any of my precious Knowledge on a "Armored Harvester" that was indestructible. Even if it was an otherwise resource and cost free upgrade.
If the AI does start killing harvesters at a faster rate than the 8-10 minutes or so that Dazio and I came up with, then defending them becomes more important. Using Knowledge to replace metal/crystal is also something I'm not enthusiastic about, either. 750K is roughly 5 AIP. That's 1 gate raid. That's 5 distribution nodes. And so on.
The cost for losing a harvester is 80 seconds in resources + 150 seconds lost income for each occurance. Cost = ( 80sec + 150sec ) * x, where x = # times destroyed
The cost for the Exoshield is 80 seconds in resources + 0.5 * income for all the time it runs. Cost = 80 + 0.5y. where y = time the Exoshield exists.
Setting the two equations equal to each other and solving gives
y = 300 + 460(x-1)
x = (160 + y) / 460
Not that it matters, because we all know that this unit is a burden on the player, but I'm surprised you are measuring resources in seconds rather than straight efficiency, as the time might not be a big deal if you have a good economy or a lot of savings.
What would you consider as an alternate measure of efficiency? Since build time is a constant (ignoring Assist) no matter the state of the economy, the 'seconds' unit HAD to be in there somewhere. Resource costs translated nicely into time as well, so it made everything else simple.