For reasons of semantic interoperability, open source is absolutely essential for creating a free (as in unconstrained rather than captured) utility market whether the providers all run the same system or conform to an open source reference model.
It has to be code (i.e. an expression) rather than a principal in specification document as switching needs to be perfect and the normal interpretation errors of documented standards by one vendor or another would prevent switching.
However, the value to the users is the market i.e. ask most large AWS users whether they want lots of different IaaS cloud or lots of AWS clones ... they'll say AWS clones i.e. they want a market of AWS clones.
So, spot on - the value to users is not how "open" or not this cloud is but instead the existence of a utility market with easy switching. The latter does require an open approach but that is secondary to the eyes of the user and the value of "a market".
I've long supported the attempts to create open source AWS clones in order to form that market. The whole "open API" brigade and how EC2 / S3 / EBS weren't open suffered from a lack of understanding of law. APIs are (and have always been) principles and not copyrightable. Nothing has changed. Of course, the next counter will be the process itself of how AMZN builds those APIs is not open ... ignoring that Android is equally a closed wall process.
So two cheers for writing an article which emphasises that what's importance is user need, in this case the existence of a free (unconstrained) competitive market, and not how "open" something is.
As Benjamin Black would say "Solve user problems or become irrelevant". Being "more open" doesn't solve a user problem, creating "a market" with easy switching does.