Are there some simple, unifying and convincing models about the properties we expect for the prime-numbers :
that would be much stronger than the usual probabilistic model $\mathcal{P}(n {\scriptstyle\text{ is prime}}) \approx \frac{1}{\ln n}$,
and that would answer at least asymptotically to most of the conjectures about prime numbers ? (in particular : generalized Riemann hypothesis, Goldbach conjecture, density of twin/factorial/Mersenne primes, least prime in subsets of the integers ...)
Or is it really in the nature of prime numbers, that the number of very difficult problems about them will be high forever ?