There are two main reasons:
We don't know the language. Linear B recorded an archaic Greek. Linear A shares a few symbols with B, but substituting them produces irregular nonsense.
Totaling all known texts, we only have about 7,500 extant characters. Imagine trying to decipher English using 60 tweets.
context for the uninformed?
(possibly difficult to answer) followup: is it plausible that Linear A could ever be deciphered without uncovering additional texts, or is there something about having fewer than ten thousand characters of data that makes an undeciphered writing system nearly impossible to crack?
hi! you'll find more in this post