But this is exactly the GP's argument, no? According to the agile philosophy (and my personal experience), we don't have enough information at the start of the project to make right decisions about all parts of the implementation - so we try to decide about just the most crucial architectural aspects and make the rest as easy to change at possible, learning as we go.
From this perspective, the argument is that dynamic languages make it easier to postpone some decisions and allow us to quickly experiment with the things we've left flexible. And then, over time, once we're more confident in our decisions, we can put in the effort to repay this technical debt and lock things down to some desired level.
that is exactly why I like static typing - I can just change stuff and ask the compiler to kindly list all the places that need to be updated to accomodate for the change.
As far as static types are feasible they are great to have. Clojure's spec is the gold standard I work to; if anyone has a better system it needs a lot more publicity.