# Study 000J Discussion

`000J` changes the mechanism branch in an important way.

After `000H`, it was tempting to treat embedding support as the dominant support variable everywhere.

`000J` shows that the archive is more complicated:

- in the direct divergence case, long settled bout structure appears to have outweighed weak embedding
- in the later specialized state, embedding remained a very strong organizer of low-burden versus high-burden expression

So the cleaner model is not:

`embedding always wins`

It is:

`the importance of embedding versus bout structure appears to depend on envelope state`

That is exactly why the adaptive-envelope theory from `000I` matters. A single global ranking would be too simple for the current archive.