As Joe noted, the type of edge that is optimal depends on if you are push cutting or slicing. However while the optimal edge for push cuts is the very highest polish possible, the optimum edge for slicing is not always the most coarse finish that you can apply. A coarse finish acts just like a saw, and thus there is a maximum amount of "hunger" that is practical. Now you might say just make it as hungry as possible so it cuts with the maximum depth per stroke. However the more hungry a saw, the more difficult it becomes to pull/push it through the material. In general you want the most coarse finish which allows a practical amount of force on the pull/push. This depends both on the medium being cut and the person doing the cutting. In general the more stiffer the medium and stronger the user, the more coarse the finish they will find optimal.
Serrations often consist of a somewhat-polished edge, arranged in arcs, and the cutting is enhanced by the geometry.
I think Steve Harvey hit upon a very critical point in regards to serrations in that one of the main reasons that they cut better than plain edges is that they are ground much more acute, often by a factor of 2-3. As well, the serrations that I have checked recently under magnification are very rough (Mission, SOG), I don't think they are polished at all, just ground, this also greatly enhances the edge life for slicing. The changing of cutting angles argument isn't really the dominant effect.
Serrations are little polished edge arcs, protected by and enhanced by the "points". Coarse-ground edges are uniformly-coarse inconsistencies that can act like serrations sometimes, but don't have the "protective" effect of a purpose-built serrated geometry.
I thought the same for quite some time, and in fact this is true up until a point. Basically in order for the micro-teeth to have the protective nature of a serration pattern they have to be of a decent size. For example I recently reprofiled a SOG SEAL 2000 on a 100 grit AO belt (primary and secondary grind). After I was finished altering the geometry I checked the edge under magnification and found that the teeth were up to 0.15 mm deep and up to 0.25 mm long. These teeth were large enough to actually have teeth inside them. I cleaned up the edge by stropping on canvax loaded with wax. This removed the debris and did a slight aligning, the blade would now shave evenly on both sides, but was rough.
I did an edge retention test by slicing 3/8" hemp through 2" of blade length, to check how the blade cut, and how the edge held up to extended use. Sharpness was tested by standard means I have described before (thread, poly, and on the hemp itself). The blade started out by cutting the hemp with between 11-13 lbs. After 510 slices the performance had degraded to between 19-21 lbs. I then did 150 slices through dirty 5/16" poly, which was actually leaving dust on the cutting block. The blade had then degraded to 25-27 lbs on the 3/8" hemp. After five alternate strokes on a smooth steel the performance jumped back to 17-19 lbs and was stable through 14 further test cuts through the 3/8" hemp.
Throughout the cutting I was checking the edge degredation under 20x magnification. I found that just like a serration pattern, these large micro-teeth would first wear along the tops, and thus the inner scallops were protected from abrasion for an extended period of time. The deeper the teeth, the more protection as the more material has to be worn away (or deformed) for the edge to stop cutting, thus the more coarse the finish the longer it lasts and the more protection from really gritty materials.
If you try similar work with a coarse edge that is much finer, say 600 DMT, you will not see the same level of edge retention. The teeth formed by the 600 grit DMT are much smaller, around 0.035 +/- 0.05 mm in depth, just visible under 20x magnification. While this finish has much better slicing ability than a high polish, and much better edge retention during such work, it wears down much faster than the x-coarse belt finish, and does not have the protective nature of the much larger teeth.
I am going to get some coarser belts and see how they work, a 40 grit should produce micro-teeth that are just barely visible and bump the slicing performance up another level.
Serrations however do retain a number of advantages of such coarse ground plain edges, primarily you can do really abusive cutting to a serration pattern like slice on a ceramic plate. Because the point on the serration pattern are so much thicker than those on the micro-teeth, they will protect the scallops whereas the microteeth will all get squashed. The microteeth finish however is trivial to sharpen, the serrations are not.
-Cliff