I really loved this blog. At DockerCon the other week I was talking about how hard it is for computational scientists to get other computational scientists to repeat their results. The reason is that small changes in input data, the code and the environment would conspire to effect the results.
Now you can ship everything - P, Q & R - and just run again.