Russ Tedrake

actually it was python3.9 on mac.

Sorry to be late to the party. Just saw the proposed PR. My initial reactions: - the name MathematicalProgram::Add() , and the need for a "suggested best practices" feels clunky...

Another deficit in the current best practices proposal (at the top) is that AddCost can only return a Binding with the general constraint type. So your suggested best practices could...

**RE: Add vs Add[Cost|Constraint].** I'm happy with using just `Add[Cost|Constraint]`, and removing `Add`. I think all calls would be unambiguous without the extra characters, but feel that this just adds...

fwiw -- i've just updated the deepnote tutorial to use Drake's JointSliders instead of the manipulation joint sliders, and removed the dependency on the manipulation repo.

This seems like the right issue to add one more idea/request: It would be nice if the authoring multibody plant tutorial could also show people how to author and visualize...

Also, from slack: [russtedrake](https://tri-internal.slack.com/archives/C02KXJ6LM54/p1654939112447029?thread_ts=1651202317.475869&cid=C02KXJ6LM54) in this tutorial, there are a few examples of having the sdf string directly in the jupyter cell... a workflow I like very much. But then...

Checking in on this one. I'm writing lots of trajectory optimization code for underactuated right now that looks like e.g.: ``` for j in range(num_q): # TODO: use vector form...

yes, please, if it fits. i think it would add a lot of value.

Possibly related to #9773. At least, I'd prefer to scrutinize the behavior of a formulation that we think is correct.