"ONLY code when you can't think of ANY other way to validate your hypothesis"

That's not really good advice. I would change it to "Only code when you can't think of any <i>faster</i> way to (in)validate your hypothesis".

I think that Paul Buchheit disagree: <a href="http://paulbuchheit.blogspot.com.ar/2009/01/communicating-with-code.html" rel="nofollow">http://paulbuchheit.blogspot.com.ar/2009/01/communicating-wi...</a>

Nah. Hard to believe, but some people think in code. They should continue to work the way that works best for them.