Safer: Grouping waves best on expected swim times and starting the fastest waves first would lead to less traffic throughout the swim, and where there would still be inevitable traffic (e.g., toward the start) it would at least be more appropriate traffic given swimmers of a similar speed would be together. The less people being knocked into and swum over, the more everyone enjoys the race; from the beginner who might have a panic attack when all of a sudden someone is on top of them, to the athlete going for the win that is frustrated by needing to constantly fight through people.
More Fair: Fighting through crowded waters slows you down. If an athlete has the 'bad luck' of being in a later wave, a fast swimmer's time is going to almost always be slower than if they started earlier. There is no positive drafting effect since packs generally form early and within a wave.
No Additional Costs: At first I thought that perhaps some races start statistically slower age groups first in an effort to have the race finish earlier, and thus reduce permit, EMS, etc., costs. But then I looked into this a bit more, and it seems like there is almost always an athlete in every age-group that pushes the overall race time-limit. So, even if you are more likely to have 6 athletes in a 75-79 age group on the course for 8 hours, if you have just one 'bucket-list' athlete in the 25-29 age group go 8 hours, then the course stays open the same amount of time. Given these are not real-time decisions, I presume contracts/permits are signed ahead of time anyways, so starting slower waves first can't have anything to do with an attempt to reduce costs (which is why I am always befuddled when fast age-groups start last).
How it Would Work: During race-registration, simply ask athletes for a projected swim time (with some preamble about how one might do this). Take the distribution, and create waves based on this. Is self-reported data ever going to be truly accurate? Of course not! BUT, it will be a lot better than what we have today, and certainly addresses the biggest issue of safety since it will undoubtedly separate swimmers where there are big ability [and thus likely comfort] gaps. I would take this a step further and guess that swimmers more apt to 'deflate' their times are competitive (maybe they want a draft effect) so from a safety standpoint, I would be less concerned about having someone that swims 32 minutes start with the 28 guys if it prevents a 40 minute swimmer from being toppled 10 times. I am not arguing this system would be perfect, but it would be directionally correct and far better than what is currently in place.
Added Benefit: If anything, this would lead to a *faster overall race* since presumably, swim-start waves could be separated by less time since the faster swimmers would be going off first, creating a natural separation throughout the field.
I could certainly be missing something, but to me, this seems like a no-brainer.