I've just stumbled upon this paper :
Computational Reduction of Wilson's Primality Test for Modern Cryptosystems
Interested, I've read it through, used the final formula (eqn. 21) for a small prime and it failed. Is it me or is this paper heavily flawed, if not at the edge in terms of actual relevance ? By relevance, I mean, just look at the example : they proved the primality of 29 in 14 rather heavy steps. How many steps would've taken a trial check ? It seems a bit too optimistic ... and it's quite hard to read too.
What do you think ?