The blogger needs more evidence.

.

He should write a script to harvest a subsection of Amazon. Then he should parse the results and see how closely correlated the Vine-to-Bad-Review there is.

.

A couple of data points here or there are not very convincing, but if he sampled say 10,000 products he could say:

* How many have reviews from Vine?

* Of those Vine reviews, what is the distribution of votes?

* Of non Vine reviews, what is the distribution?