The OP argues that although programs written in dynamic languages are smaller, they have more bugs. But empirical studies find the opposite: bug counts grow superlinearly with program length irrespective of language. In fact, program length is the best predictor of error rate we know. So his argument contradicts the evidence.
Chapter 27 of Code Complete (2nd ed.) discusses this. So does http://news.ycombinator.com/item?id=3037293 among other HN threads. Also, if anyone has access to Capers Jones' "The Impact of Program Size" from Programming Productivity (1986) and is feeling generous, please cite the relevant findings. I can't find it online.