OK, here are my two images...
The top one is probably cheating, but in 250 days I was able to get 6 learners up to 2510 in the 'know' score.
The second one was a bit trickier, but I found that when there was a good number of learners, and the learning objects were fairly difficult, and people were both helpful, but not patient, it seemed to cause quite a bit of question asking.
Like Kami, I found it interesting that the fewer the learners seemed to increase what was learned. I'm not sure why that is the case in this simulation. I would think in real life that the more learners you have, the more of an 'expertise' you have access too, but then I guess you also run the risk of information overload. So many questions being posed that the flow of information becomes too unwieldy.
Overall, similar to my experience with flocking, I found myself moving sliders just see what happened, then using that information to try and predict, and get the desired results. I would need to play with, or look at the calculations behind, the system more before I feel like I really understood, and could predict what would happen. Several times I would move a slider and get an exact opposite reaction than what I expected.
I like this kind of modeling. It reminds me of Isaac Asimov's Foundation series. The basic premise is that you cannot predict what a single person will do, or even a group of people, but you can predict what large numbers of people will do. While I'm not sure how closely this model represents real life, it would be interesting to see how close it comes, or what other factors are needed to make it a closer representation.