So complexity theory is confirming your intuition, which is that 'optimization type problems' are hard.
What about sequencing life? Supposedly we share a lot at the genomic level with animals we are vastly different from. Could a large degree of the polynomial that is life explain that divergence?
I am assuming you are pushing the same angle as the other poster. That is, most optimization problems are actually not "large constants" in the exponents. So, since most of what developers think of as "hard" are almost all classical NP problems, maybe it makes sense to look at other things we don't typically think of as computational.
Could just be nonsensical, though. I fully accept that.
Edit: I meant "small" coefficients above. Apologies.
NP: probably O(e^kn), k>1, where k close to 1 makes it 'fast', but still exponential
The objection is that k can be big for some polynomial algorithms but small (near 1) for other exponential ones, so we can have 'slow' polynomial algorithms and 'fast' exponential ones.
But in practice, k is usually small for polynomial algorithms and not close to 1 for exponential ones so simply knowing whether an algorithm is exponential or polynomial tells you something.
The 'hard' problems you gave me all (probably) fall into the exponential runtime class, whereas the easier ones (like finding any solution to a Rubiks) fall into the polynomial runtime. So your intuition of 'hard' vs. 'easy' matches complexity theory's evaluation, even though theoretically there could be really slow 'easy' problems and really fast 'hard' ones.
As an example, an algorithm to find the optimal solution to a Rubik's cube would actually be very difficult to do. However, finding a solution in a short enough time frame is quite easy. This is more true the larger of a cube you try and solve.
Contrast this with problems such as encryption, where we have specifically made problems where there is not a "best answer", but rather there is only a single answer that matters.