A prollyfill is attached to a proposal/draft for new or improved native features and it provides a way of changing the evolutionary model of the Web...
Let me use the popularity of selector engines as an example: Only after a giant swath of the community of developers decided that this was a useful thing and even extended a few selectors did it get real consideration as something that was needed natively - and at that point it really had no draft/spec to it beyond: 'It matches CSS'. It didn't have use-cases or tests or prose or anything --- So we got querySelectorAll, and really didn't match a lot of expectations or needs - but it did 'match CSS'. Likewise, some things in most of the selector engines proved to be kind of problematic if the idea was to 'match CSS' - but since there was no draft and no one looking at that, it just kind of slipped under the radar.
See if this helps at all: http://briankardell.wordpress.com/2012/11/17/w3c-extensible-...
