I know the "Links" section is recommended to track duplicates, but I feel it doesn't go far enough because it doesn't list all the duplicates, and moreover, it makes it hard to find the question.
Rather than link two questions together, a real duplicate link should add the question to a group of duplicates. Whenever a question is a member of such a duplicate group, all other duplicates would be listed in the right pane, along with the question vote count, sorted by the votes.
Advantages:
- Easy navigation among duplicates
- Extra incentive for voting on questions
- Extra incentive for properly linking real duplicates (example of unlinked that would benefit)
- Better motivation for only closing as exact duplicate when a question is an exact duplicate and not merely somewhat related.
Disadvantages:
- significant implementation effort
Example:
Example of a question that would benefit: What programming language for a beginner? [closed]
The Linked section contains a single other duplicate (out of 12 links, i.e. <10% signal to noise), and then recursively apply this algorithm. After a while I discovered these duplicates:
- Language for non-programmers to start learning programming
- Which language to choose to learn the basics of programming? [closed]
- what programming language to learn, php, asp.net or jsp? [closed]
- Recommendations for a language for a non-professional beginner [closed]
- What is a good programming language for beginners? [closed]
- Best language for a non-programmer tech support person to learn
- What language is best for a beginner to learn? [closed]
- What is the easiest language to start with?
At which point I started to get lost in my recursive search, because I kept stumbling across questions I've already found and the signal-to-noise (w.r.t duplicates!) in the Linked section was consistently very low.
This feature would have encouraged people to actually use the duplicates rather than re-ask a question after stumbling across one like this one.
Notes:
- I'm not suggesting that the Link section should go. I'm just saying its purpose is orthogonal (but superficially similar) to linking duplicates together.