The NWO Big Software project MERITS is finished. It investigated whether refactoring of legacy software is possible by automatic learning. The conclusion is that this is a bridge too far, but learning techniques are an effective means to get structural insight in software that is otherwise hard to obtain. Read the the full report.