Today I attended the 20th birthday celebrations for DIMACS at Rutgers University. There were some nice talks: Ron Graham on addressing in graphs; Peter Winkler on combinatorics in statistical physics, Joan Feigenbaum on approximate privacy, Michael Trick on the DIMACS Challenges, Richard Karp on implicit set algorithms, Eva Tardos on games in networks, etc. There was also an interesting panel discussion on education (de facto national standards coming in the US are strongly opposed by some here, because of the de-emphasizing of discrete math topics) and industry (the environment is much meaner now, and managers are loath to fund projects without clearly stated benefits to the company). This was my first visit and I can see now why DIMACS has been so successful. Thanks to all the organizers and speakers for a very enjoyable day.
Note: another report by Muthu