Fascinating look at the way the multiplicative nature of quality in complex tasks leads to exponential effects in overall productivity. The video is about 20 minutes long, but worth the time.

The video is fairly general, but I think anyone involved with software development will see the implications right away. One is that only small drops in productivity of one member of a team can be disastrous. For jobs that require getting many things right (made from lots of individual tasks) then your chance of getting a satisfactory result drops to almost zero with only a few low quality results on individual tasks. Everyone has to be on top of their game to successfully complete complex tasks.

Another is that good people want to work with good people. This leads to virtuous or vicious cycles that are hard to get our of. Hiring and keeping good people is vital. And poor performers hurt everyone. If you can't get rid of them, best to isolate them.

For a firm, hiring the best people matters...a lot. Putting those people in good teams and a good environment matters...a lot. For an economy, training people to do high quality jobs matters...a lot.

Adrian Colyer applies this to DevOps in The O-Ring Theory of DevOps. He points out that O-ring theory implies that every stage in the DevOps pipeline needs to be performing at high quality and failure of even on stage (e.g. failure of tests to catch problems) can cause big problems. Conversely, automating steps well implies that the quality will be consistent.