Because.it.can.work.better 
Some people enjoy that, myself I fall off way before that point - I just have the border where further improvement is no longer worth the effort.
I’d like in here point you out to the definition of word “optimal”.
There is no such thing like “local optimum” for every set of constraints there is only one “optimum”
Now, constraints can shift, for example following your analogies from behind - by one gaining more knowledge of the model upon which we are operating.
Optimal is the goal of the process of optimization indeed. But for me (personally) the fun is the process, not the goal itself.
Personally I’d swap to rattle just for the sake of flying cooler boat 
buuuut…
point is some WANTS to see how far they can push it - and the process of pushing it is where the fun is 
That’s true, as with any other branch of gameplay 
AH you anderson with yer edits xD
Well yes, but actually no. If you go by the fun version of the process - research and try - indeed proving it will be virtually impossible.
But once one manage to learn whole of current status of the model it is not so difficult to mathematically find and prove that optimal - at least until constraints shift - for example due to market prices shift. But it’s saping all the fun from it 
Improvement is altering solution to be better. Optimization is a process of changes to make your solution closer to the optimal.
The main difference to be honest in here is that “optimal” includes more variables to play with - cost efficiency for example is often taken into consideration when determining optimal, which can rule out better - but prohibitively more expensive solution out.
Well, except variables have been extensively researched and are available to public quite easily. Now sure if you go manually for it - as in start with example and then test it and improve upon it you can make the process virtually never ending - amount of steps will still be finite - it’s just you’ll run out of lifetime before you reach the count 