Ordering search results by total number of likes in topic
The current “Most Liked” search feature lists individual posts, whereas the desired functionality would display each topic only once, ordered by the total number of likes in that topic, including the original post and replies. cpinter hopes to see this change.
However, codinghorror estimates that implementing this feature would require a significant rewrite of the search functionality, taking around 1-4 weeks of engineering time, depending on familiarity with Discourse.
cpinter acknowledges the complexity and welcomes the feature if enough people request it, but lacks the capacity to develop it themselves.
Other users, tophee and fedorov, discuss the nuances of the requested feature, with fedorov describing it as “search for posts but sort the results by topics.” cpinter confirms this interpretation.
The current operation of “Most Liked” ordering in search is like this:
In topic A, there are N liked replies, each getting M likes (total N*M likes)
In the search result page topic A’s title appears N times, and each item in the result points to the individual replies
Instead, we are looking for an option to see the following behavior in case the condition described in 1 applies:
In the search result page every topic appears maximum once, ordered by the total number of likes in that topic, including the original post and the replies, which is in case of topic A is N*M. The link of the item in the search result page points to the original post in the topic (i.e. the whole topic).
I hope I was unambiguous. Would this be possible? Thanks!
This would require a major rewrite of the way search works. I’d estimate this would take a week of time for an engineer on our team who has been working with Discourse for years. For someone on the outside, who is not already intimately familiar with Discourse, probably more like 3 or 4 weeks of engineering time.
Understood, so this is not easily available. If enough people request this, and you decide to go for it, then we’d be very happy to have this feature. Otherwise I don’t think we have the capacity to work 3-4 weeks on this feature ourselves. Thanks again!