I've read and re-read this: Why do votes get locked? I'm not at all against the concept of post locking to avoid anomalous (malicious) behaviour. My concern is that does adversely affect vote tallys on some specific answers.
In the most recent example, an answer to my own Linux networking question stated several things that I know to be correct (or so I thought). I'm relatively experienced with Linux and the answer tied together things I already knew. It gave a good direction for further reading and research. Ultimately reading that answer did lead me to discover the right answer. I gave it an upvote and began my further reading.
The problem: it helped me find the right answer by being entirely wrong! It led me to discover the entire configuration system (NSS) didn't exist in the particular Linux I was using. And yes I did state which Linux distribution I was using in the question.
The answer was wrong.
A more knowledgeable person could have known it was wrong from my question, this isn't about guesswork.
This is by no-means common. I have cast 400 votes on that site and this is only the second occasion I have really wanted to reverse my vote. But it does tend to focus on particular type of answers, meaning some have really skewed voting:
- well written (coherent)
- reference a commonly held belief and knowledge
- entirely wrong
They draw up-votes too fast which, when the reader discovers that the answer was wrong, are already locked in. At-best they can only upvote a correct answer and leave it side-by-side with a wrong answer. There are times when we up / down vote an aswer because our knowledge is wrong rather than a misclick.
The suggestion that the wrong answer will over time be lost in upvotes on other answers ignores the reality of most stack-exchange questions.
I wonder if it would be possible / be a good idea, to allow users to unlock a (very low) percentage of votes.
It's hard to think of how unlocking 0.5% of votes could cause a significant problem. It could easily be monitored to ensure unlocking is not directed at a single user. But while the percentage is low, it might have the effect of re-balancing some answers.