>>30
but in practice the more functional your code is, the fewer bugs it has. How have you written a lot of software and not noticed this? The original quote is about C++ design patterns and template metaprogramming, which are overly complex and retarded and cause bugs. Reducing side-effects simplifies your code (almost invariably.)
You can find-replace this comment too. It won't change the reality that you can go out and measure yourself. Functional style always reduces complexity and bugs. I'm actually confident in using the word "always" here.
>>31
Fine. You can still use monads in a dynamically typed language. They're just lambdas that take and return lambdas. (which is what everything is.) You can even write a nice macro to reinvent the "do" syntax from Haskell. You can write a bunch of macros to reinvent the useful bits of Haskell, actually, and call it Qi.
I think that Haskell's attitude toward IO and side-effects is one of those useful things that is interesting and worth learning the language for further insight on. Haskell has something interesting to say, is all I'm saying.
>>32
The point is that IO of whatever sort can be sectioned away. A large part of your program is unconcerned with it and only transforms I to O. Those parts don't need side-effects, and when you get right down to it, implementing those parts with side-effects isn't even particularly easier. In fact, it's entirely an optimization concern.