1) The point is exactly the typework of such deltas.

Sorry - didn't understand this. Typework?

2) Multiple controllers can be at work (and I know in the future you'll have multiple fx columns, but this automatic controller thing would work now! (and then!))

I'm not wild about expending effort on something I'm already planning to make obsolete.

3) It adds to non-destructive editing. Since I'm not hard-defining the depth in the fx column but in the definition of such an automatic controller. If I need less crescendo eb/flow, what would be easier? to change all the deltas across 30 patterns for every channel that uses them, or simply go to that auto.ctrl editor and change some values there? I'd say the latter is, even from a code point of view, the correct and most easiest way.

You're introducing an abstraction that could vary between command, file, project, user, etc. As before, I mentioned this inhibits memorisation and learning - you would have to look up the delta values before making an edit because they are no longer evident in the command itself. Abstractions can be useful, but they take away fine control and should be used with caution. It makes it a higher-reasoning task which must be thought out before hand, rather than a "get stuck in" organic approach to laying down tracks.

4) Experimenting is easy. Want another flow (e.g. starting later and ending earlier)? Just change one command, using hard-placed deltas or values you'd have to change the whole shebang for each and every experiment.

Isn't this the same as 3)?

I'll definitely look at MIDI effect slides, but will probably opt for fixed delta (like Dxx, etc.). Take solace though, as any "scripting feature" would probably allow you to set up your desired functionality.

Chris