I haven't mentioned working on the DPM version 2 glossary in awhile, but there's been steady progress. From my initial group of over 80 terms, we were able to delete almost 20% of them without even breaking a sweat.
(I just realized that guesstimate of mine bears suspicious resemblance to the 80/20 rule.)
That still left quite a few terms to look up. ("Wait," you protest, "How can you look up definitions for your new terms if you haven't written them yet?" Good question.) Unless we invent a new word, then somebody somewhere has written a definition for the ones we're using. So I chase them down to see if it's a definition we can use.
I don't want just any definition, however. First, I check an internal tool that we're using at Microsoft to gather all glossary terms into one database. If I strike out with the tool, I search for the term in published documentation (term site:microsoft.com). I really want to find a definition that one of our products is already using.
But if I can't, the search goes webwide, along with a few other keywords to keep the results relevant to the general topic. On this search, however, I don't want to find a single definition; I want several that are worded differently, and I want descriptions rather than just definitions, so I can understand how the term is being used in the industry. And from that, I'll form our own definition.
That's the stage most of our terms are at now. Next, the terms and definitions go to the reviewers: program managers, developers, testers, editors, and other writers. The glossary gauntlet, you might say...