Gould sees simple to complex as some kind of statistical inevitability, moving from one form to another in a kinds of inevitable drunk walk of progress, ignoring all non random forcings which act on it, so that each species that evolves, from dinosaur to horse to human is just a bit further out to 'complexity' by chance (well, statistical inevitability to be fair). This ignores both the selective pressures which adapt species in specific and very non random ways (humans don't have big brains because drunken random evolution man happened to stumble that way, but because natural selection pushed them in that direction and favoured the right mutations to achieve it) and also the separate progress of different branches. Take the bombardier beetle discussed as developing complexity. Gould would have it that the drunken man just inevitably ended up making the simple beetles more complex. But in fact the more accepted theory would hold that they gain their increased defensive complexity through a wide variety of very non random forcings. Crucially too it has no effect on, if you like, the macro complexity of evolution a whole except where it produces selective pressure on other organisms. Interestingly though, studies like the big yeast experiment suggest that given similar circumstances, other life forms would evolve similarly, of by a different path.