Firstly, I'll say that I've met and worked with several people from Microsoft and there's no doubt in my mind that they're putting 120% into their QA mission. I don't think any company in the world invests more into and does more heavy-duty patch release testing than Microsoft.
That said, the general question "where's the value in hurting conversions and pissing of a lot of folks, just for a chance to get a $40K fee?" does have an answer.
Consider a company where departments are fighting for budgets and managers are looking after their own neck far more than the common good of the organization (believe it or not, this happens sometimes). Organizations can easily devolve into a state of loving those business units that are perceived as "revenue centers" and hating those perceived as "cost centers". Microsoft used to be famous for derisively referring to teams (and even individuals) as "overhead".
Obviously it's stupid and short-signted, but sometimes things like QA on patch releases for 3rd-party distributed software end up being several degrees removed from the revenue inflows, at least to the bean counters. In these situations, managers may be pressured into downsizing their quality efforts or trying to raise as much raw revenue as possible even if it really represents negative value on the whole.
When the pressure comes down, the perverse incentives pop up. This "stack ranking" scheme sounds like a brilliant example.
My guess is this policy comes from one or two^Hthree motivations:
* Users don't like patching.
* A desire to raise the cost of patching to hopefully encourage higher quality in the 1.0 releases. True or not, the perception among many in the industry is that Apple's app store requirements to jump through hoops to release versions has raised quality among those apps.
* But the simplest explanation goes like this: "Why did you have to hire these extra staff at 40K each?" "Well, because we had more patch releases than last year." "Hmm...."