The thing about scientific progress is that concentration of knowledge and people discussing it tends to drive knowledge much more than isolated geniuses or lone insights. For example, Chrysippos of Soli, the founded of propositional calculus, was said to have written 300 books on linguistics and logic—and we get barely anything.
To balance the attitude you express, the notion that the ancients knew diddly-squat is far more pervasive. Until about three decades ago, the prevailing scholarly consensus was that the Greeks had no non-trivial combinatorics whatsoever. That is, until in the 1990s some cryptic lines of Plutarch were decoded to refer to Hipparkhos' calculation of the 10th Schröder number in critiquing the Chrysippos' claim about the number of molecular propositions that can be constructed from ten atomic propositions. This is not a trivial thing, yet Plutarch claimed that it is well-known to all arithmeticians.
That's an entire branch of mathematics that was previously thought to be completely unknown to the ancient Greeks, yet it has existed, but was lost. Along with Archimedean protocalculus, which was rediscovered very early in the XX century. And scholars of Greek antiquities made all kinds of just-so stories why the Greeks were totally dumb about combinatorics.
Again, the value of such a library is not that would necessarily that it is full of correct revolutionary insights to drive science and technology, but that a record of arguments and critiques, as long as it is available (and so exists as an institution rather than a tomb) is itself a driver of such development. Even primitive and wrong things can be very valuable to that end, as long as people think about and discuss it.
A very good example of that is Newtonian physics, which through Galileo owes a lot to medieval scholastic commentaries on Aristotelian physics.