[mattknox]

I ran twitter's new engineer training for a long time, and much of this rings pretty true to me. I would only add 3 things: -1 the skill of teaching is approximately orthogonal to domain knowledge, and the difference between a decent and very good teacher is huge, both in terms of student-reported happiness and in terms of retention of the material covered. -2 history with the company tends to be very valuable, in that new hires are often curious about _why_ the company decided to build system X as it was. My goal was that teachers should be able to answer most such questions accurately, and truthfully say that they didn't know in the other cases. I found (anecdotally, because this data is hard to capture) that the student-reported quality of a class was generally proportional to the maximum length of question/answer/followup q/answer/... chains. -3 the onboarding process is an extremely powerful propaganda platform. During twitter's long transition from ruby to scala, (before I ran it) there were a series of presentations that were forward looking to the point of inaccuracy: they described as existing in the present that which we hoped would one day exist. This confused a bunch of new engineers when the rainbow unicorn scala world promised them did not materialize. On the positive side, many of them were then motivated to help build a scala world.