It would be survivorship bias if we assumed, based on the old books we have, that all old books on average were better than modern books. I don't think that's the argument being made here.
Let's assume for a moment that all famous old books are good books. We might then say that time is a great filter because all of the old books we know of are good books.
However, it might be that time actually has a false negative rate of 99.9%. Almost all of the great old books are filtered out by time. In that case, time probably sucks as a filter. But then again, if your only goal is "read only great books," time is still a useful filter even if it has a terrible false negative rate, since it's irrelevant to you that there are lots of other books being excluded, so long as you get a sufficient supply of good books.
However, it might be that time actually has a false negative rate of 99.9%. Almost all of the great old books are filtered out by time. In that case, time probably sucks as a filter. But then again, if your only goal is "read only great books," time is still a useful filter even if it has a terrible false negative rate, since it's irrelevant to you that there are lots of other books being excluded, so long as you get a sufficient supply of good books.