i actually hate this paper already just from the second page because
However, the introduction of some smoothness criteria for group elements immediately allows the introduction of an index calculus method and a subexponential complexity.
the "smoothness criteria" is clearly not applicable to the standard case or they would say that
Section 4 considers recent results on the statistical distribution of random versus DH triples. Clearly, any possibility of distinguishing such triples would yield information that might be exploited to break the problem.
type of guy who thinks statistics is gonna prove P = NP