I don't think this has much to do with the principles outlined in Bret's talk. It provides no immediate feedback regarding what the code is doing and how the individual parts affect the output, just that the output passes the tests.
You're right, it's not quite the same. I talked about this on the blog post. When it comes to general-purpose programming (not drawings or simulations), test results seem to me the best way to get visibility into what the code is doing.
Bret demonstrates what immediacy means for general purpose programming at 16.45 in the presentation. It is like a perpetual debug mode.