This a great example in terms of historical importance, but it is not as
challenging as others we'll come across - so it probably isn't the best
candidate for a "hard case". The reason is that a record of a battle has an
unchanging core of content that is browsed repeatedly- so it's a type 1
document. In theory, this record could be represented in a variety of
different formats. For example, each tank could have it's entire course
recorded, or there could be a time series of records, with each record
consisting of each tank's position at a given moment; the two methods are
approximately equivalent.
Deeply interactive or generative documents will be even harder to preserve.
While there are equivalent representations of interactive/generative
systems in theory, in practice it very hard to find them, and even harder
to confirm the equivalence, for the reasons I touched upon in the long
email this morning- the time domain and the interface with interaction
hardware are two areas that are particularly confounding.
Best,
Jaron