When asking Joe Sadusk about it he suggested it was the time spend on each layer or number of shells, since I was printing with just 1. So I thought I'd do a little experiment and see what I can see. I tried different shells and different minimum layer times. The results were mostly the same, changing minimum layer times didn't fix a thing, adding shells only fixed a little.
employed something similar before, so I thought I'd give it a try this time.
I was impressed that in Makerware I was able to add a 20mm test cube from the menu then scale it non-uniformly to make it as tall as the piece but thin so it doesn't waste much plastic.
The result, the print on the right, was the best of the lot. You can see the sacrificial pillar there too, it kinda detached and if it weren't for the anchor would have been useless, but the anchor kept it in the general area where it could serve as a sort of brush while the layer cooled. This is a filthy workaround, but it works for now. The real solution in my opinion would be if the bot were equipt with a brush and the minimum layer time didn't slow down the print but instead moved the nozzle away for any extra time. But until then I think I'll keep the sacrificial tower in mind.