What should we be building?


What do you think The Coral Project should be making? Why?

Tell us about your biggest community problems, and also where you think we can have the biggest impact.


This is regarding the First Project. Pardon and please move if this reply belongs elsewhere.

We’re doing rudimentary work along these lines. We’ve got Fb comments associated with articles on Hearst properties and we’re pulling in numbers of comments by post and attributing an engagement score to that post. (In our thinking, the post author, content, and comment author have distinct reputation quotients.)

Further quantitative analysis we plan to do from there:

  • Frequency by individual community members within a given property
  • Frequency by members across properties (each prop has their own Fb app, so each needs to be queried separately, pulled in [to somewhere], and cross-referenced)
  • Cross-referencing of commenting and web analytics (when comments are made, what correlation is there to traffic, especially w/r/t more recent comments on older articles?)

Qualitative analysis includes

  • Surfacing frequent commenters and allowing for editorial review of those comments
  • Simple sentiment analysis for positive / negative / neutral ratings of comments

We’d like to to define, refine, and automate this work. “Time is the limiting factor in journalistic community building” here among other places. I’ll dig into docs; if there’s guidance you can give to get more involved with putting this First Project work into use against our use case, please let me know.



It’s great to see that you are interested in looking at users across properties. Once of our guding principles is that “a community” is not a single thing. People interact with different individuals on various assets, sections, and, in your case, properties.

I like to think of this issue from a ux point of view. When an individual posts, they do so in reaction to what is around them in the interface. This points toward interesting insights surfacing from more and more granular slices of the whole general mishmash that is a large community. There will be a strong focus on this in our Trust product.

Cross-referencing with web analytics is another angle that we have in mind. This is a bit more challenging as analytics sources are another major dataset outside of community, but breaking down these divisions is on the agenda. Stay tuned.


Cool, thanks, David. There’s significant historical comment data to analyze from Facebook – even when we constrain it to discussions that have appeared on work published by our non-staff contributor community.

Early eyeballing doesn’t show a whole lot of repeated, cross-property participation by significant numbers of individuals. This is logical from the PoV, e.g., of Cosmo, Esquire, and Road & Track properties, e.g. To your point, we figure there’ll be more affinities within Fashion, Automotive, Home, etc. We want to dig in more.

I’m staying tuned and will also update as we find points of interest as we gather more of this member info from across discussions. Thanks again -