This began as an experiment with the New York Times API to see if I could bring reader comments back into the pages that spawned the original discussions. Currently the way the site works is that the comments are on a separate page from the article. Personally I prefer to read them all in the same place.
But I quickly discovered that having all the comments on the same page created a pretty long page. An article or blog post on nytimes.com can easily get 200 comments. 500 or 700 is not unheard of. So after the initial step of bringing them all into the same place, the challenge changed into surfacing the best of them. If there are 10 comments I might want to read out of 500, how can I find them?
Fortunately most Times writers read all the comments and choose the ones they feel are the best. And secondly, the readers themselves can recommend comments they think are good. So between these 2 flags, we have a good start at finding quality comments. This data is in the API and readily accessible, and the script uses some simple math to surface the most recommended comments for a particular post, the top 93rd percentile.
The third element I’ve added is the ability to “follow” a commenter I like. This is just a simple flag, and every time I come to a page where they have written a comment, theirs is highlighted.
So now that I know the best comments, as decided by Times editors, the readers, and myself, I highlight those, and minimize everything else. The rest of the comments are shortened into 140 character previews (or “tweets”) that can be expanded if they look interesting. Or you can forget about all this filtering and expand everything in one shot.
After using this script for a week or so, I eventually came across some commenters that were really great. And I became curious as to what else they were up to on nytimes.com. The API gave me access to a feed of their other comments, and from that feed I could extract their user ID, which is also used in TimesPeople. So my last step was to add the functionality to load their other comments and their TimesPeople recommendations into the page.