Constraint Satisfaction and Rete
A few days ago, Daniel Selman of ILOG (now owned by IBM) published a solution to the Einstein puzzle. See http://blogs.ilog.com/brms/2009/02/16/einsteins-puzzle/. He did this in response to a challenge from James Owen who is one of the organisers of the October Rules Fest conference. James invited the various vendors who have involvement in the conference to provide a solution using whatever approach they deemed best. See http://orf2009.blogspot.com/2009/02/puzzle-1-for-conference.html.
Daniel suggests that, if ILOG had used a Rete rules engine, they would have had to implement 'convoluted rules' to solve the problem. Daniel is correct, but I thought it would be interesting to look a little deeper at why this is the case.
Read more at http://geekswithblogs.net/cyoung/archive/2009/02/24/129639.aspx.