I propose that A/B test strings should have their counterpart also included in the description.

For now, it's relatively easy to figure out each pair of strings, since their string IDs are adjacent. But in the future, after couple of itterations, that connection may quickly get lost, as well as the context of the pair of strings. Without knowing what the string is testing against, localizations may accidently change the core difference of the strings and mess up the test. So I think the easiest fix is just to also include the other string in the description.