WillJ: this isn't exactly about postcount. Turner's screenshot shows a thread with zero *replies* (not necessarily zero posts) and zero views. He is right that without further information (namely was the first post actually there) this could be slightly different than what was happening in the other thread. What I meant in the passage you quoted from me is the following:
1) Darius types a new message and hits "submit."
2) The system posts the thread, ups his PC and displays the "thank you" screen.
3) His browser processes the meta refresh on the thank you screen and requests to see the post.
4) Things get real busy.
5) Because the server load is over some threshold, instead of displaying the post (and upping the view count) he gets the "server busy" message.
6) Load drops a bit and Turner gets the index page before anyone else views the thread and sees the 0/0.
The result of that scenario, assuming counts are processed at the times I have guessed, is a new thread with an actual post but zero views and zero replies. The simple timing alternative would be that #6 above occurred in the very small amount of time between #2 and the server processing the view request of #3 and this was followed by Darius seeing his post just as one normally would and the view count becoming one at that time.
For what it's worth, my guess on PC is that it is increased when the system enters your message into the database following "submit" and not when you are taken back to it.