Specifically, it modifies the file in place (-i),
Matches lines between paragraph tags(/PatternStart/,/PatternEnd/),
Then, doesn't delete those lines matching the pattern range (!d), which has the effect of deleting all other lines outside the pattern range when passed the in-place argument.
Being partially sed stupid myself (or just plain stupid), I think its supposed to pull only the text between paragraph tags and save that to an htm file; then launch firefox and display that file, giving you only the pertinent text of the article (sans ads)