Oh! I listened to Scott Hanselman interview one of the authors of this paper on his podcast the other day. It might interest some of you as she explains it all in a very accessible way: https://www.hanselminutes.com/763/tldr-extreme-summarization...
(hey peter!) I did get very excited about it when I heard about it, but cooled significantly when I learned that it was trained only on cs papers and required the full abstract plus paper text. nice proof of concept but going to need significant work to generalize for, say, a newsletter business like yours
Ah yeah, I definitely wouldn't want to use this sort of thing on newsletters. I'm against automated curation generally. It's not what we're about :-)