Why Yieldbot chose Cascalog over Pig for Hadoop processing

Sometimes I feel like we are living in a game of Mad Libs.

Why ______ chose _____ over ______ for _____.

And then some 5 year old is making up words to fill in the gaps.

while my experience with cascalog is limited to "hello world", even that served to show how game-changing a REPL is for interacting with hadoop, let alone all of the other features which i am surely failing to appreciate. i would not want to be in direct competition with backtype.

I think a key observation is that working with a REPL just makes dev feel so light and quick.

I love it and have found it to be a big part of the joy of programming.