I've resumed dreaming about on-the-fly playlist creation by querying the "attribute tag" database. This quickly leads to questions which could kill the idea:

Would it be possible to expand the empeg ID3 tag database to include additional fields? Might as well leverage the player's ability to read and store the database. Wouldn't need to see them in emplode or jEmplode, unless those programs rebuild the database.

In conjunction, how horribly would the database size expand if additional tag fields were added to EVERY record? For my dream program, maybe 15 more fields would be needed.

Or, instead of adding fields, could an existing field be used to store an encoded value which represents 15 additional fields. ie: Genre contains 15 character where each position in the string represents a tag and each character represents a predefined tag value. Then software could parse it (and desktop software could encode it).

If these things aren't remotely possible, then I'll just continue listening to the "Rock" genre on shuffle. I'd much rather query Genre=ProgressiveRock Vocal=Male Tempo=Fast Tonality=Minor Mood=Serious Timbre=Electric, if it fit my mood.