tenpo pini lili la mi sitelen e toki pi musi mute en pi sona mute. nena 'Submit' li lon poka wan. sitelen "Accept" li lon poka ante. lon ni la ilo li weka e toki mi. pakala ! kepeken sitelen kalama suli "F".
Anyhoo, what I said was (without the charm or sagacity of the original)
1. position systems are more complex than tp's because they are at least one level of abstraction further away from things: tp may have piles, but it doesn't have piles of piles. Roman numeral are complex because they are half a dozen systems functioning more or less together (old IE quarternary, quinquenary, decimal, vigesimal all with a mixture of additions and subtractions and multiplications). Ternary numbers are hard because we don't know them and can't work them fast in our heads.
2. 'nanpa Inli "64"' refers to a number but does not obviously work as a quantifier: 'kulupu pi nanpa Inli "64"' is not obviously 64 groups nor a group with 64 things in it nor a group called "The 64(th)" or the 64th group in some list, but 'toki pona la nanpa Inli 64 li nanpa suli" is true.
3. 'nanpa mute' can mean 'manieth,' some fairly far along ordinal place (i.e third or more), but it can't mean "big number" in the appropriate sense. I've tried to think of some way of doing that other than using 'suli' but all are hideously complex -- and 'nanpa suli' is established usage for at least seven years now.
4. All of this detail work ignores the fundamental point, that tp actively discourages enumerating multitudes, as a matter of peace of mind, if nothing else. And it does seem we don't really need them for much real stuff -- except for matching numerals. Which is why I suggested (again, I should note) that Sonja needs to give us a way of reading numerals without any ordinal or cardinal significance -- just as we have (or could use) a way of reading letters without any phonological significance