CRACKING A WHIP MADE OF SMALL ROBOTS JOINED END TO END into a long, flexible chain was neither an especially bad nor an especially good way of engaging a foe in ambot-based combat. Extensive studies conducted within Blue military research labs had concluded that, on average, it was somewhat less effective than the more obvious procedure of just shooting individual ambots out of katapults. A dissenting opinion held that such studies were flawed because they failed to take into account two factors that were important in actual battle: One, the psychological impact on a defender who knew that the attack might literally whip around and come at him from any direction, including around corners or over barricades. Two, the element of skill, which was difficult to measure scientifically; the test subjects wielding those things in the lab were unlikely to have the same knack for it as Neoanders who had grown up using them and who had access to an ancient body of lore--a martial art, in effect--that they were disinclined to share with anyone else. If the whip was allowed to dissociate in midcrack, then its component ambots would be flung toward the target at supersonic velocity, which was as good as could be achieved by shooting the same objects out of a katapult. If it made contact with the target, direct physical damage would be inflicted and the ambots that had inflicted it could decouple themselves and carry out their usual programs. And if the whipcrack was off target, the chain could be recovered in full with no waste of ammunition. All the ambots came back for another attempt: something that certainly could not be said of ones that had been fired out of kats.