NOTICE: I HAVE MOVED MY BLOG TO MY PERSONAL WEBSITE!Please check http://www.pmhsilva.com for more details. You can red this blog post here.
I've been reading a lot about Game Design as part of my effort to get back to the gaming industry and one of the subjects that I've been thinking about the most is the amount of testing you should do and, more important, when to start testing the Game Design.
I am a very big fan of iterative development so I believe that making the full game design document, then implement it and then test the game is just plain wrong. That is, unless you're happy with just fixing sowftware bugs and launch the game without considering the testing feedback.
I prefer to start off with an idea and put it down on paper. Just a simple synopsis and eventually some concept art. Then get together with the team of artists and developers (if YOU are the whole team, get together with all aspects of yourself as a game developer) and discuss the concept with them. The output of this should be at least one "level" or prototype of your game. Just like they do on TV series with the pilot. Put one "pilot" together and start testing immediatly!
Why? Well, the more feedback you get, the better! You may think this is an awesome idea and determine that your target will be a specific demographics and when you actually get people from those demographics to actually play your game, they may tell you it is indeed awesome or they may tell you it sucks big time! And you should listen to them! After all, you're doing this game for them, aren't you?
Recently many people resorted to places like Kickstarter to help fund their games. But to attract investment on your games, you have to convince people that your game is good AND give something back. Makes sense. One thing that occurred to me when I was reading this was that maybe one of the advantages Kickstarter investors might like to do is actually be involved in the development process, as testers. Why not? Who better to help you test your game and tune it than those who actually are willing to put money behind it?! But how would these tests work when your testers might be spread around the world? That is in itself a challenge, but totally doable.
Nevertheless, in response to the question that entitles this blog post, I would say that testing should start as soon as you have a prototype, a pilot. Iterate through that and you're game will be much more successful!