tokens are a concept specific to pico8, whereas character or byte count is understood by everyone and has precedents with tweetcarts, the sizecoding scene, and so on.
But this is a PICO-8 jam? So surely those people using it would be aware what tokens are? (or at least, they should learn!)
Fair enough though. Rules is rules. Just seems a bit inefficient given the way PICO-8 works! I've personally found the challenge in PICO-8 to be keeping within the token count, not the 64K.