> In that case, there is no resolution. I hope that criteria can be
> established so that we are not "arbitrary". I have no hope that we can be
Well, social scientists use "sampling" as a valid means of representing
the population. As long as it's a random sample, one can generalize
For example, saving **every** newspaper in the U.S. probably isn't
necessary. But saving every third day might be 99% as useful, for
historians in the future.
At a certain point you get diminishing returns, and you want a certain
minimum "sample size."
But if, by "arbitrary" it is meant "random (and thus representative),"
then a cost-effective middle-ground can be reached. :)