Our results for when there are 100 pricebots and wA=0.5 and wB=0.5 in total are the following. What we have found is that among the game-theoretic, myopically-optimal and derivative-following algorithms, the game-theoretic always does best. When game-theoretic plays only derivative-following, the game-theoretic ends up with all of the pricebots. When the game-theoretic plays only the myopically-optimal, the game-theoretic ends up with most but not all of the pricebots, and ends up in a sort of steady state. When all three algorithms, game-theoretic, myopically-optimal, and derivative-following, play against each other, the game-theoretic and myopically-optimal force the derivative-following to lose all its pricebots. Once there are no more derivative-following pricebots, then it is the same situation as before with the game-theoretic and myopically-optimal and we end up back at a steady state with the game-theoretic having most of the pricebots. When the Q-learning pricebots also play and wA=0.25 and wB=0.75 then the GT still do the best if there are 100 pricebots but if there are only 28 pricebots then the MY do the best. We see then that the game-theoretic algorithm, although the least complex, with a large enough initial general pricebot poplulation, manages to do the best of all of the algorithms. If we compare our results to those found in [1] we see some discrepancies. The myopically-optimal tend to dominate in [1] whereas that is not what we find. Also, [1] find that game-theoretic is the worst algorithm and that derivative-following is a fairly good algorithm, again conflicting with our results. It would be worthwhile to figure out why the myopically-optimal algorithm someimes does extremely well, such as when there are few pricebots in the general population, and sometimes does quite poorly, such as when there are many pricebots in the general population. The simultaneous qlearning needs to be debugged, and more testing is needed to determine whether the sequential Q-learning could do better than the other algorithms. It would also be interesting to see how the Q-learning algorithm fares against the other algorithms when it only plays against one of the other algorithms.