Amazon should remember those with $200 MSFT monies and claim it ludicrous to
suppose that they'd compromise their editorial integrity for a mere $10K*.
Journalists like to keep a chinese wall between the editorial and
advertising sides of the house (you'll see the division in mastheads, often
on different pages). Publishers, however, aren't journalists, and the
easiest way to make sure the correct review gets written is for the vendor
to provide some appropriate copy along with the check.
I don't have any references (one could probably start with whoever asked
"quis custodiet ipsos custodes?") but I recall an argument that the Romans
didn't maintain a public police force because they figured that les flics
would almost certainly be paid off by somebody, so at least with private
outfits you'd know from where the funding came.
* If one wants to avoid extinction of newly-learned behavior, variable
reinforcement is the way to go. Even if your subjects are more
sophisticated than psychology sophomores, compromising integrity should
produce more lasting results; anyone who actually sells out for enough is
more likely to hold out for more primary reinforcement than habitually offer