Put the Han dynasty and the Roman Empire side by side and the resemblance is almost uncanny.
Both are continental empires governing enormous, diverse populations. Both run sophisticated administrative systems over long distances. Both build road networks that outlast them. They overlap in time. Neither knew much about the other.
Then look at what each one did with knowledge, and they diverge completely.
Rome consumed Greek learning and produced almost none of its own. The Han generated new knowledge in mathematics, astronomy, medicine and technology, and then did the harder thing: built institutions to keep producing it after the people who started it were dead.
The Nine Chapters on the Mathematical Art, compiled under the Han, contains systematic solutions to linear equations by a method equivalent to Gaussian elimination, roughly eighteen centuries before Gauss. It handles negative numbers as ordinary quantities to be calculated with, which European mathematics was still refusing to do in the seventeenth century.
Zhang Heng, court astronomer in the second century, built a water-powered armillary sphere and a seismograph that detected an earthquake several hundred kilometres away, in a direction nobody at court believed until the messengers arrived days later.
And the imperial court employed him to do it. That is the part that matters more than any individual result.
There was an imperial academy, an imperial library with a cataloguing system, a state medical college, an astronomical bureau taking continuous observations, and an examination system that recruited officials on the basis of what they knew.
Its effect was to make learning the route to status and security. In a world where every other society allocated power by birth or by force, China made it, imperfectly and with plenty of corruption, allocable by study.
Its cost was that it channelled the ablest minds toward a fixed canon and a fixed examination, and the syllabus was administrative and literary rather than natural-philosophical.
But hold both halves of that at once, because the lesson is not "the examination system was a mistake".
The lesson is that institutions decide outcomes, and that they do it whether or not anyone intended them to. Rome had no institutions for producing knowledge and produced none. China built institutions that produced an enormous amount and simultaneously constrained what kind.
We do not currently decide these things by examination. We decide them by grant cycle, by publication metric, by what a funder can be persuaded will yield an application within five years. Those are institutions too. They are producing an enormous amount of a certain kind of work and, quietly, almost none of another kind.
Han China and Rome had the same resources and made opposite choices, and the difference shows up a thousand years later in what survived and what had to be recovered from somebody else's translation.
Nobody in either empire experienced that as a choice. They simply decided, year after year, what to fund.