I should note that I believe actual increases in force output potential THROUGH a weight training protocol MUST be related to actual cross-sectional increases (hypertrophy) if the advancement goes to be applied to a real world situation, e.g. a sport - any other advancements (i.e. more weight, more reps, without correlating increase in size) will be related to attunement (the refinining of specific qualities related to mechanics involved with the exercise as well as central nervous system adaptations to performing the necessary work with the lowest resources nessary relating to the principle of least effort) that is related to a specific skill (i.e. a military press) and does not transfer to real world situations.
So, in my best evaluation, power training to improve a football player is essential only if doing FOOTBALL tasks, and that goes to skill training. As I've said before, given my understanding of motor-learning research, our central nervous system learns complex tasks in a very linear way. Given the complexity of power cleans (several different steps that involve the firing and collaboration of THOUSANDS of motor units, muscle fibers, etc. - including the SPECIFIC adjustment of joint position, lever movement, isometric flexion, balance etc.) it makes sense that the body would learn these complexities SPECIFICALLY in order to avoid injury as the load increases (can you imagine trying to clean a 300 pound weight the first time you ever try it, just because your legs can provide enough force to??) as well as to figure out how to economically coordinate all these functions biologically.
Imagine you've been doing power cleans for years, and suddenly you're told to do the snatch. Similar exercises, indeed, but I can promise you that if you use a maximal load that you use with a power clean, you'll likely kill yourself if you try and snatch it over your head. Many things can go wrong even when adding a single additional step to something that is so complex. Balance issues, joint position, ancillary muscle involvement - a few extra seconds and steps in a complex movement can involve all the muscle groups in your body!
In any case, I don't believe that even marginally complex skills transfer to other completely unrelated tasks. If such were the case, quarterbacks could be good pitchers because of various commonalities between the tasks; such is not the case because they are not exact. So learning to be powerful in an exercise (clean) does not make one more powerful generally, unless a cross-sectional increase occurs over the muscle. This is possible, but the rate of speed that the actual exercise is performed is not relevant (see posts above). There is no evidence to suggest that power cleans are better at increasing hypertrophy of the hips, glutes, or quads beyond the capacity of a squat or a lunge, for example, and I'm not dimissing their potential for such things, but I do think it's quite unsafe.