On Mon, Dec 12, 2011 at 12:27 PM, Nate Vack <[log in to unmask]> wrote:

> On Mon, Dec 12, 2011 at 2:06 PM, Brian Tingle
> <[log in to unmask]> wrote:
>
> > Potential contributors of specimens would have to be okay with the fact
> > that a determined person could recreate their original records.
>
> To make things simpler, you might just see how many contributors would
> just be OK with the original records, and skip the obfuscation.


true; but I'm also worried about end user support questions if we end up
have something like an ead-demo.xtf.cdlib.org

plus I'm also using this as an excuse to play with nltk (natural language
toolkit) and learn more python

but yes, I'm sure I'm prematurely optimizing this problem

On Mon, Dec 12, 2011 at 12:48 PM, Joe Hourcle <[log in to unmask]
> wrote:

> If the list of missions / agencies / etc is rather small, it'd be possible
> to
> just come up with a random list of nouns, and make a sort of secret
> decoder ring, assigning each mission name that needs to be replaced
> with a random (but consistent) word.


This is a great idea.  I think if I reset the pseudo-random seed based on
the input; then I don't even have to worry about keeping a decoder ring,
and it will work with any noun.  As long as the results look so silly that
no end user might mistake it for real this might work.

maybe I'll create an option switch for the text replacement method;
pig-latin, vowel/consonant-sensitive random letters, or random dictionary
word