There are two major types of burr as I see it. One is the bits of steel that form at the apex because the abrasive cannot cleanly remove every last bit of steel as it trails off into space (trailing), or as the abrasive enters into the steel (leading). A percentage is shoved out of position, either because the angle of the cutting mineral wasn't optimum or the steel deformed out of the way ahead of the mineral's passage (per the second type of burr below but on a much smaller scale). This is left behind per pass and over the course of a few passes it builds up to an appreciable level - heavily dependent on the manner in which the abrasive is presented etc.
The second type of burr is the curling that is formed when the amount of force applied is greater than the steel's ability to resist. There will be tailings at the edge per the first type of burr, and along the curving surface as it veers away from the abrasive surface. The curvature issue is mostly a product of force and what the steel can support - this can be eliminated by using an abrasive that cuts the steel more efficiently at lower force, using a steel that has more lateral strength, or increasing the edge angle.
The burring by abrasive contact has more components to it and has been covered to some extent in earlier posts. Factors that reduce this tendency - less applied force, greater friability of the abrasive, looser attachment of the abrasive to the underlying surface (mobility), conformability of the underlying surface, higher contact speed, with some steels a higher Rockwell value.
The mineral is basically a shaped ploughshare driving through the steel. The steel builds up and is cut free, builds up and cut free - abrasives work with a scratching mechanism for the most part (though CBN can actually shave curls off if the abrasive orientation is just right). Continuing with the analogy, the ploghshares to either side and behind all grind off the wake-like buildup of the abrasives to either side and any left in its path. You generally won't see burrs on a flat surface but will see grind troughs that terminate in mid stride if using a scrubbing pass or stopping cold in the middle of the stone - occasionally there will be a bit of debris still attached but not often.
If the ploughshare has mobility, can fold back on itself, sacrifice part of its physical structure (friability) or otherwise deflect when it encounters too much resistance etc, burring will be reduced. This is why lapping operations produce the frosted surface, and also reduced burring - likewise slurries, softer waterstones etc all tend to produce less burring - all other factors being equal. Speed reduces the burr formation because the ability of the steel to hang on as the minerals scratch away at it is partially a form of tensile strength per an elongation test - higher speeds result in less elongation before failure. At hand sharpening speeds I do not believe this is a factor.
I still have reservations about the conclusion reached in the earlier blog regarding coarser edges and inability to create high points that are as thin across as with finer abrasives (albeit over a smaller total percentage of the edge). The absence of burrs or burr precursors on the sample edges leave me curious about how they were prepared specifically. I have some optical images taken at 400 and 1000x of edges finished at medium grit value (800) and the high points appear to be quite acute - low sub-micron. Granted they are not with the clarity of a SEM at the same magnification, but even accounting for some lack of focal clarity they edges look mighty thin.
(post #59). 
