Mar 12

2010

Robots Will Pick Our Next Game

Posted by: Adam Strong-Morse | Comments (8)

As we’ve mentioned, we’re currently finishing up Choice of Broadsides.  That means that we’re also working on picking our next couple of games for development– whether that’s Choice of the Dragon II, Choice of the God, Choice of the Consort, or something else.  We thought it might be interesting to discuss our method for selecting projects.

The first thing we did was do some brainstorming for some ideas that we thought would be fun to write, fun to play, and popular.  We came up with a long list of ideas–really, any nifty genre with any nifty character type can be the basis for compelling choices.  We also had some suggestions from e-mails we’ve received, which basically provided more support for ideas we already had–various people have suggested werewolf games or vampire games or other ideas that we were vaguely talking about.

So, the question, is which game should we make?  That’s where the interesting choice is.  The first thing we did was to toss up a blog post asking people to vote for their favorite ideas.  We’ve gotten some great feedback, but that’s still pretty limited.  The key step is our next approach:  using Google AdWords to test a bunch of different games.  Google AdWords is a service that serves short text ads to lots of websites on the web.  The key is that AdWords ads can be cheap– we’re aiming to pay about 5 cents per click.  That means that $100 will get us 2000 clicks.  And AdWords can rotate among a bunch of different ads.  So we’re building ads for a bunch of different games, none of which we’ve written yet.  The AdWords robots will rotate our ads for us.  If one game idea gets 500 clicks over the course of the experiment, while another gets 10, we know that there’s more market demand for the 500 click game.  We can also try tweaking the ads as we go– if one ad is doing badly, we’ll pull it and replace it with a variant for the same game.  If it then does better, we understand what aspect of the game to emphasize.  If all ads for the same game idea do badly, it’s probably not the most popular idea.

So that’s how robots will pick our next game.  But they won’t really, because while we’ll take into account demand, we’re also going to focus on which games we most want to design/write.  (That’s how we chose Choice of the Dragon and Choice of Broadsides–those were both ideas that we were particularly excited to write.)

There are a couple of reasons why we’re going to take into account our preferences.  First, we’re writing these games at all because it’s fun.  (That’s also why we encourage you to try writing your own ChoiceScript games.)  Choice of Games isn’t anybody’s main job– it’s a hobby job for all of us, although it produces some real money, and we hope it will make a meaningful part of our incomes.  But more generally, we should be doing things that we like and that are fun–that keeps us motivated, productive, and satisfied.  That’s what we aim for in our day jobs, and it’s doubly true for Choice of Games.

Second, if we tried to write things we weren’t interested in, we would do it badly.  When you care and are excited, your work is better.  We would make worse games more slowly if we were just doing what we thought would get us the most downloads.

Third, if we’re excited about it, we think others will be, too.  The blog voting provides us with some useful information, and the AdWords experiment should provide even more, but it’s still all tea leaves.  There may be many thousands of people who will download our game or play it on the web, but would never click through an AdWords ad or vote in a blog poll.  And some people who click on an AdWords ad may be looking for something very different from our style of games.

So, robots will pick our next game.  But they won’t pick it on their own.  They’ll more identify particularly strong contenders, and then we’ll pick our favorites to do now.  We think that will be an effective strategy, but equally importantly, we think it will be lots of fun, both for us and for you.

Subscribe by E-mail