You wouldn’t want to pick any deliberately perverse formats, obviously saner is better to the degree that space allows; but even fairly primitive techniques should at least be able to tell you that something was purposefully written there, even if deciphering it eludes you. Al-Kindi was, obviously, a really sharp guy, so his results probably aren’t typical; but he had worked out the use of frequency analysis back in the 9th century. More sophisticated statistics came later, and information theory is comparatively new; but a sanely chosen digital format should be fairly clearly and intriguingly neither random nor merely decorative based on frequency analysis alone.
Doesn’t much matter if The Future only cares about chipping the archival glass into crude spear-points in order to fight off the radroaches that come out of the wastes to carry off the weak and unwary every night; but that’s not a data encoding problem.
