urbleflan

This looks really cool.

This is why cycle time is a much more useful metric. The book "Accelerate" shows what a mature conversation around software metrics should look like.

EDIT: To be fair, my point may not directly address yours. "Value" is still a tough thing to determine.

Taleb has some good comments about this, which is that just as you can make a prediction and have it come true by pure chance, we can view evidence from history and fit a pattern to it (even though that pattern occurred only by chance). Personally I think the best way to view history is as a series of poorly controlled experiments. As with scientific experiments, the evidence may be suggestive of some pattern, but we should embrace an attitude of skepticism and openness to counterexamples. The patterns we see may tend towards "laws," but we should assume some level of imperfect resolution and control and therefore associate a confidence percentage with our inferences.

It's a little strange to use the phrases "tend to" and "iron law" in the same sentence.