A couple things that would be useful.
First off, a setting for Genome that outputs transform changes. So we could use it to procedurally transform objects. Would be great for rigs.
Secondly, a way to get velocity data without manually creating an offset mesh. I wouldn’t mind if it introduced a performance bump, but production wise, it can get inconvenient to manage two nearly identical objects in order to maintain the velocity data. Not a huge thing, but definitely an inconvenience.
Transform, like a Genome Controller?
For the velocity, where is that coming from? Like is it a Skin or PointCache modifier or is something in the base object like an XMesh Loader or Frost?
Yeah, a Genome Controller. But I think I would actually prefer to have it stay applied as the modifier, but send data to the objects controllers, or to an under the hood Genome Controller. Basically, the Modifier makes it easy to see that Genome is involved. Sort of keeping it all in one place.
As for velocity…I am referring to the process where you create a copy of a mesh and offset it in time (ie through mesher or point cache) and referencing the vertex displacement between the two to get velocity. As it is, you cannot maintain live reference because of dependency issues, so you end up having to remember to update both meshes or things can break.
Actually, the idea of Genome sending data to controllers, opens a number of cool ideas…Like controlling things other than just transform controllers. Modifier parameters…material parameters…hmmmmmm…Thinkbox ICE here we come ? 
Unless there’s a solver involved, like Scene Genie, I wouldn’t want stuff running out of place. A controller should be listed as a controller. Same thing with a texmap.
I suppose without a solver a global Genome that just made named tracks might be intetesting. Don’t have a use case yet.
For velocity, you’d have to evaluate the mesh below the current Genome, right? So you didn’t get some recursion mess.
Yeah.
Velocity would be great.
I’d like to throw my vote in for it.