At the moment, the move is no doubt profitting from my rather rock-like table image, so that's definitely part of it--although people are seeing some rather unexpected moves these days, too...

Anyhow, here's another idea more for long-term, picking up a bit on Harringon's randomization ideas: I wonder if it would be close to optimal if you kept something like a 50-50 mix between legit raising hands and "other" from LP. More like 75-25 in MP.
Also, a certain number of times actually limping on what I'd call the true raising hands (AK, JJ-AA) maybe something like 25% of the time.
At the moment, I'm probably not far off from the general percentages in all probability, but the positional component does make my raising strategy at least in principle readable. I do think it would take a VERY long time for anyone to really figure it out, though.
Basically, it does have its advantages if, in any given situation, there's really no hand that your opponents can effectively exclude from the mix. I will say that I'm much more inclined to make "table factors" more the decisive component than just pure randomization, as Harrington appears to envision.
On a slightly different note: One thing I'm really liking is calling readable raisers with position on unusual holdings. Out of position, I'm extremely picky, normally just going with pairs and AK. But in position, if I have a read on the raiser (e.g., check or weak bet on a bad flop), I'm pretty sure I've turned a profit off of calls with hands like KJ, etc.