I like Mr. Wizards explanation as well. He noted it, but I think it needs to be stressed that the EGR tables have to be used in conjunction with additional vtuning.
The process is as basic as he mentioned-
step 1. verify where change is needed (cylinder specific). To do this you can use the egr analysis chart, but I just use the highlight feature in the Vtune software now. That will get you pretty close, and with egr table adjustments it falls in the horseshoe and had grenade category IMO (close is good enough).
step 2. make a change in the egr table (cylinder specific). The % of change needed is 2 or 3 times or more the % of difference that you may see in some cases.
step 3. perform a vtune run to dial in the ve tables to the change
step 4. review your work...rinse and repeat as necessary (I only shoot for EGR table affected cells being with in 5% of the non EGR table affected trend line).
I like and use this feature, for much of the reasons already mentioned. I think it also helps reduce the margin of error for a street tuner, since there’s less chance of holes in the ve tables. Another benefit that I have seen with egr adjustments is by using this to reduce peaks, the cubic inch of the calibration can be more in line with the actual cubic inches of the engine.
The toughest part of working the egr tables is they do not exactly line up with the ve tables. You just need to look at surrounding cells and make educated guesses on which egr table cell to adjust. You also need to consider that sometimes you will add value to the tables, and sometimes reduce values on the tables. I found in most cases that 4k and up rpms need reductions, but you should collect some data before making the change just in case. I think lambda MAP based VE table cals are much easier to tune the EGR tables than the TPS VE table cals, since those calibration are already segmented into MAP. It makes seeing where change is needed much more obvious.