Why is Hadoop need to start the examples?
If you start an example you can choose what optimizer implementation to use. Because some of them are based on hadoop, the classes are having references to the Hadoop libraries. Therefore the jar have to be startet with
hadoop jar hadooptimizer.jar [ options ]
For the moment HadoOptimizers example are also using the special Hadoop like packaging mechanism, with dependencies includes in the lib folder of the archive.