True … let’s use the random word IDs then
True, the “(9) Implement that tagging refers to a revision.” would be just a poor surrogate for the same purpose as quasi-random word IDs. So let’s drop that point. What’s inferior about referring to revisions plus numeric word ranges is that after an edit, codings / taggings cannot be shown correctly in the latest version any more. But the latest version will be shown to website visitors by default, so they wouldn’t see any codings in edited content then. Which would be really unfortunate, as a great benefit of Open Ethnographer is its effortless side effect of giving semantic web integration (rendering tags as RDFa into posts …), semantic in-site search abilities etc…
One could dream up a mechanism to translate a mechanism to dream up translating old numeric word ranges when a new revision is created, but I guess the word IDs are a cleaner way then. Even though they create more messy HTML. (You’re welcome to disagree of course, and the software design is not set in stone so far.)