Bot.10 uses optimistic search. As Scott says, it helps pull the bot toward unexplored territory. Since Bot.10 has no other form of exploration this makes a lot of sense.

Scott's comments on optimistic search prompted these thoughts.  Suppose we can see a food pellet. We can find can find an optimistic but not a pessimistic path to it for the closest ant.  What would we want that ant to do?  The ant could be assigned to the optimistic search path, which in the worst case will be a form of exploration.  The alternative is to devote the ant to some other task.  That other task, whatever it is, may be more important that the exploration task created by optimistic search. Suppose even that the other task is to be part of the bot's general exploration algorithm.  If the bot's overall exploration  algorithm is a good one, it is presumably a better use of the ant than to assign it to the ad hoc exploration task generated by optimistic search.

One could argue that this concedes the food pellet to the enemy. Perhaps so. The question then becomes what our policy should be about food pellets that we can see but can't see a path to? Perhaps we should devote additional resources to them. Perhaps we should have all near-by ants move toward them using optimistic search on the grounds that we don't want the enemy to get them. That may be a good idea. It seems like a separate policy decision, though.