What do you do to ensure that your stereo downmixes are intelligible? I did not make a separate stereo mix on a recent project, opting instead to let the viewers' hardware take care of it, and I regretted it. Is it possible to get quality stereo sound simply by properly setting the metadata while encoding the surround mix, or must one make a separate stereo mix?

For reference, this is what Dolby says (www.dolby.com/assets/pdf/tech_library/18_Metadata.Guide.pdf). I want to know what works for you.

We do a separate mix in *most* cases.
We do a fair number of industrials, so we leave those to collapse at the playback device since there is little in the rear channels outside of machine and forklift noise.
Outside of those, we generally do a separate mix option for the DVD.