Comment by cyberax

> This in fact makes the problem worse, because a) you admit the metric is shit and does not reflect work

It's just like in physics: "all models are wrong, some are useful". All metrics are wrong, but some are useful when they are correctly applied.