The role of directed methods in the protein field is analyzed. A crite
rion is formulated which suggests the necessary conditions for the suc
cess or the failure of the ab initio direct procedures. The distributi
on of quantity z = [alpha]/sigma(alpha) is central in our consideratio
ns. Most of the experimental protein data do not satisfy such a criter
ion, therefore their ab-initio solution is a quite improbable event.