You can measure human emotion and engagement with simple patterns.
You just have to trace the pattern of the replies, to see how flat the structure is, how many people are involved in the different levels of the structure, and how grammatically complex and diverse the replies are.

GM
- GM
- GM
- GM
- GM
- GM
- GM
- GM
- GM
- GM
- GM
- GM
- GM

Groundbreaking topic
- Complex note from npub 1
-- Complex response to npub 1 from npub 2
- Complex response from npub 1 from npub 3
- Complex note from npub 4
- Complex note from npub 1
-- Complex response to npub 1 from npub 2
- Complex response from npub 1 from npub 3
- Complex note from npub 4
- Complex note from npub 1
-- Complex response to npub 1 from npub 2
- Complex response from npub 1 from npub 3
- Complex note from npub 4 

 Yes, I think you are right. By which I mean, I have no clue how automation / detection works in reality, but it sounds like you know what you're talking about.

Sounds like you could be capable of making some sort of interesting discussion / post highlighter bot? 🤔 

I would like to see primal replace their sidebar with something like this. 

 I'm more logistician than bot programmer. 

 I don't think they understand the math enough to do that. They've tried to mimic the effect with time-frames (1-hour, 2-hour, etc.), but that often is just a reflection of active time zones, rather than interesting discussion. Europeans tend to trend until Americans wake up and drown us out. 

 Tbh I would prefer any sort of customisable sidebar on primal. Even replacing it with the #asknostr hashtag, at least that is/was (mostly) used but people who have something to bring up / want answered.  

 The reason why the first note is obviously going to be a boring read is clear: nobody expects it to contain interesting infomation in the replies, so nobody opens the replies, they just respond with their own GM from their feed, or zap, and move on.

You want to highlight the notes where people have opened it up and started chatting.