Walter Hower originally proposed this problem as a benchmark in the Constraints Archive. He claims it has several interesting features: it concerns higher-ary constraints (not just binary ones), it is a real-world problem, it is not too complex, and it shows that local pre-processing may have little or no effect.
Revisiting Global Constraint Satisfaction
Inf. Process. Lett. 66(1), 41–48, apr 1998
Global Constraint Satisfaction Revisited
Technical Report TR-97-02, Department of Computer Science, University College Cork, 1997