We don't ask randomly, we ask everyone, so everyone is free to offer up a rating.
CORRECTION: Just reviewed the stats again and it looks like the response rate is in the mid 30s.
The issue, instead, is whether the people who choose to respond--to take another step after their support encounter and rate the quality of service--are representative of all users, including the majority who do not rate the experience. That's a tough hypothesis to test, and I suspect it could cut either way: the happiest customers may choose to respond since they so enjoyed the experience, or the angriest may choose to respond since they want you to be aware of their disappointment. Finding out which (if either) of these is occurring is tricky, though.
That said, I certainly don't mean to impugn what you're doing here. I love the transparency and I think you probably are getting a fairly good snapshot of overall customer satisfaction. Also, even if there is a response bias, you can safely use these numbers to benchmark over time, which I suspect would be very useful.
My guess would be that the biggest benefit from this is going to be when people hear about how great the customer service was at 37s, stop by to check it out, and see a chart full of smiley faces, each one potentially representing an echo of the great experience they heard about.
PS: I would like to add a frowny face for never getting a response to my application :(