Backwards compatibility is something that you only ditch for extremely good reasons, for instance to deal with a security issue that can not be handled in any other way. Breaking it for reasons of aesthetics is just plain dumb. The python ecosystem is still suffering from these major mistakes many years later and python's image as a reliable and trustworthy tool suffered right along with it. Such changes are fine pre 1.0, but once you ask people to base their production systems on it you maintain backwards compatibility at a cost, even if that means holding your nose occasionally. And my suggested fix would at least allow for the old syntax to live happily along with the new one, which given the price (two stupid brackets) could have easily been dealt with. Focusing on purity vs the responsibility towards people maintaining 100's of millions of lines of code in aggregate (if not another order of magnitude more) means you don't have your priorities straight.

Imagine we'd drop HTTPS for a new protocol because it fixes the 'Referrer' spelling error.