One thing we’re thinking about is how to determine natural breakpoints or “chapters” for long discussion topics – and we can eventually display these breakpoints as navigation selections on the expanded topic progress bar.
Say you have a 500+ post topic that you want to explore. How do you do that?
One way is to use our “summarize topic” function, which uses our internal topic hotness algorithm to produce the reader’s digest condensed version of the topic – by stripping out 80% of the posts.
The traditional answer here is pagination, but when have you ever seen a book with chapters like this?
- chapter 1: pages 1 - 50
- chapter 2: pages 51 - 100
- chapter 3: pages 101 - 150
- chapter 4: pages 151 - 200
Chapters in a book have titles, and tend to break on meaningful turning points or inflection points in the story.
So the question is, what exactly are those natural inflection points in a 500+ post topic?
We have some ideas.
Conversations tend to ebb and wane. There will be 50 posts one day, then everyone goes to sleep for the night. They wake up the next day and continue the conversation. The natural breakpoints are at lulls in the conversation:
- Activity period 1: Monday - Wednesday
- Activity period 2: Thursday - Friday
- Activity period 3: Sunday
- Activity period 4: May 2015
But what if the conversation never pauses, never wanes? How can you have meaningful breakpoints when the participants are in so many timezones, and the conversation so intense, that it never lets up?
Maybe you could use arbitrary time based breakpoints:
- Time period 1: May 2014
- Time period 2: June 2014
- Time period 3: First week of June 2014
Hotness or Significance breakpoints
We could use our existing in-topic hotness algorithm to determine which posts in the topic had the most “heat” and display some kind of visual heatmap of the topic, a map of interesting areas in the conversation where something significant happened as measured by likes, replies, internal/external links, read time, bookmarks, and so forth on each individual post in the topic.
So those heatmap breakpoints would look like:
- Significant event 1: three days ago
- Significant event 2: yesterday
- Significant event 3: six hours ago