What this means is that tomorrow if I have to suggest to someone that their problem can be formulated as a convex program, this not only gives them a solution "in theory", but gives them a solution "in practice", that they can actually implement. Such happenings are surprisingly rare when applying theoretical methods to real-world problems; it is not often that one can invoke a sophisticated algorithm that has known worst-case bounds, and find an implementation for it.
Today is the FOCS panel discussion on 'Exciting the public about (theoretical) computer science'. One way (though certainly not the only way) of exciting people about theory is to demonstrate the power of mathematical rigor and worst-case analysis by showing how algorithms that have good theoretical guarantees actually work well in practice: in other words, showing that algorithms can become "technologies".
This is a really long and roundabout way of mentioning a call for papers: ESA 2006 will be held in Zurich from Sep 11-15, 2006. ESA has a both a theoretical and applied track (disclaimer: I am on the PC for the applied track), and this year the applied track is trying a new approach to submissions:
Authors of papers that include an experimental aspect can make supporting source code or data sets available on a webpage (referenced in the submitted paper) or send them by e-mail to firstname.lastname@example.org; such supporting files will be considered by the programcommittee members at their discretion.So get your source code ready ! And remember this is not about theory making itself "useful", but about demonstrating that the formal approach to algorithm design is far more effective than any other approach.