If Gavel only has a 6% success rate in simulations, how can it be feasible for HackMIT to use? Surely there must be other factors that make it work better in practice..?

UPL!

Thank you!