They tested it out during the weekend beta tests.
The fact is, there was alot of people already signed up for guilds on this server, which was open since day one of early release. They didn't all get in on the first day, that number of people was obviously ignored. You can tell the idea of trying to fill the servers up was some pencil pusher in the company and others thought it was a good idea.
Now the problem is, their math is massively off. Blatantly didn't take into account of people signed up for guilds. Ignored the fact that people might not be able to play right away with their friends whom of which already chose a server. Someone who choose a server the first day but hasn't been able to log on much due to work. While making sure the servers were filled up to have people to play with looked good on paper, it was seen on the first day of pre-release, that it was a horrid idea to do it in the manner that they did.
The way to properly go about, is to have taken a save character population count, if they all logged on at the same time and it hit a high server load making it full, then that is when you add a new server. Cause when weekend comes and retail release comes, your going to have more people sign on as well as the weekend warriors.
Its all about looking for the invisible math that you wont know or cant know. In this case they ignored all of the unknown math. Anyone who runs a gaming server would understand this. Sure you can put 10, 30 slot counter strike software servers on a single piece of hardware. But what happens during prime time and some people find the server and like it, they will play in it and help fill it up(unknown factor), as well as someone joining cause of the number it has(Unknown factor), then your friends try to join(known factor) and it crashes the hardware server.
