The fact that Google use MapReduce has no bearing on whether it is a "major step backwards" to the database community. The original authors essentially argue that MapReduce ignores 40 years of database technology, and that it is the worse for it; whether MapReduce is essential at Google isn't the point.