> Trying to look quickly for a hard number, it seems that the usual estimate (perhaps one study being widely reported rather than hard estimates) is about 5-10%

> Trying to look quickly for a hard number, it seems that the usual estimate (perhaps one study being widely reported rather than hard estimates) is about 5-10% of works surviving. I suspect the actual value is much smaller, probably less than 0.1%. Those estimates are based on surviving references and indices, and there's no way to know how complete they are. Fundamentally we're all just guessing - if something's completely lost, there's a very good chance that it's impossible to know centuries later. > In the modern day, I think we are doing a better of job of preservation and archival than existed in Classical times. I hope so, but I have my doubts. When Sumerian libraries burned, the clay tablets simply hardened. Today, most modern data storage systems only last a decade or two, they require copying for the data to be readable... and we have no idea if future generations will make those copies. Also, increasingly data is stored in dynamically-generated pages on websites and in single page applications (SPAs). If the website goes down, a lot of the data is essentially impossible to get. I really appreciate efforts like the GitHub Arctic Vault: <https://archiveprogram.github.com/arctic-vault/ https://archiveprogram.github.com/arctic-vault/>. That has a fighting chance to be recoverable in two thousand years. But the Library of Alexandria showed us that if all the archives are in one place, they can all be simultaneously lost. I certainly would like to this we're doing a better job of archiving. I want to be proven wrong :-).