Well, use of those would make the mean meaningless.
It wouldn’t be a problem if the polls had upper and lower bounds, because then you could exclude them (but you could also make the upper bound infinite if you wanted to). I don’t think there’s a need for them, though.
You don’t need to use infinities to make the mean meaningless: giving answers such as 1e100 will suffice. On the other hand, NANs are traditionally just disregarded when computing means (i.e., the mean of 1, 2, 3 and NAN is taken to be 2) -- essentially they would amount to a blank vote.
MP did not want it to accept either of those things; the notation used suggests “the largest integer less than or equal to infinity”, which doesn’t exist.
Should it (or at least, should it accept
inf
and/orNaN
)? [pollid:41]Well, use of those would make the mean meaningless.
It wouldn’t be a problem if the polls had upper and lower bounds, because then you could exclude them (but you could also make the upper bound infinite if you wanted to). I don’t think there’s a need for them, though.
You don’t need to use infinities to make the mean meaningless: giving answers such as 1e100 will suffice. On the other hand, NANs are traditionally just disregarded when computing means (i.e., the mean of 1, 2, 3 and NAN is taken to be 2) -- essentially they would amount to a blank vote.
MP did not want it to accept either of those things; the notation used suggests “the largest integer less than or equal to infinity”, which doesn’t exist.