I just spent the last few hours implementing and testing a HTML parsing solution based on a library that was pointed to in a top answer, only to find out that the recommended HTML tidying library is inflexible and does't handle HTML containing undeclared XML namespaces; for example the nice fancy microformats everyone is using today.
The answer was good in 2008 and I've added a better answer now, but I am unable to remove my upvote and others are doomed to walk this path in the future.
I'm sure that this will be closed as "as designed" or a duplicate, but I'm venting anyway because it sucks.
If votes are immutable, should we refrain from voting on a seemingly helpful answer in future until we've actually implemented and fully tested the solution? What about four years later when the technologies change and this bad information is polluting the Internet?