No machine is smart enough to condense an article and extract an author's subtle ways and meanings. This is a tweet oriented, instant fix world where fluff just doesn't stick. My guess is that a lot of articles today are speed written and that's an issue too - people are the new computers! Yet, readers still want and appreciate quality - so introducing micropayments into the equation is where this idea differs. One computational problem posed may be determining what summary to serve up first, and what writer to assign to which article - if that becomes part of the scope.
there are manual methods (i put up http://www.istldr.com/) and automated ones (http://techcrunch.com/2010/10/18/tldr/)
another point to consider is that with a reputable article (think nytimes), the title oftentimes does an excellent job as the TL;DR