Turing Designer’s Diary Part 2 – Image and language

So, the central idea for Turing is that a series of abstract images will be presented to players. I’d been knocking the concept around for a while having come up with a prototype a few years ago that worked just fine, the issue was that Turing was always going to be a little pocket game, but it was a little pocket game that needed an absolute minimum of 40 or 50 pieces of unique full colour artwork, which was never going to come close to making financial sense. I’d been kicking around ideas for dealing with this and for a while was thinking of using the artwork I’ve built up over time from other games. The issue is that striking game artwork is generally not that abstract, and game artwork is also too strongly themed to allow for much interpretation of patterns which is pretty central to the concept of Turing.

The solution seems to have come from a site called Pixabay that I’ve been using for print and plays and prototyping for a while now. This is a fantastic site where artists can post work available free for commercial usage. The trade off is that you don’t get to ask for what you want so it’s been a long process of picking out suitable images and going back and forth on thinning them down or building them up. I’ve arrived at a list of images that I’m happy with and a set of artists to credit for their fantastic work, while the images have been offered free with a request for links back, we’ll be donating payments to the artists based on the success of the Kickstarter.

The process of picking images has been interesting in and of itself. Turing is based around players simply trying to communicate the presence of intelligent communication to other players, to signal that they are human through a series of abstract images while the game offers a randomly selected set of images. That rules out anything with directly interpretable symbols such as words or numbers, it also makes images with human figures a double-edged sword since people naturally gravitate to them, particularly when trying to signal their own humanity, but that does mean over-reliance on those images offers an easy win for the game when they happen to come out in the wrong place for the players. More interesting is to offer images that allow for patterns of communication to be built up by players, so much of the process has been looking for abstract themes in images, windows, moons, flowers. It has certainly been the majority of the practical work with this game and it has been a process that has been almost impossible to directly quantify and, unfortunately for this blog, put into words.

Possibly the most interesting part of the game, and in turn the process of creating the game, is the creation of a private language and how that can be taken advantage of and subverted in play. For example, I’m a big fan of the game Mysterium, within my play group it has been decided that a particular card in that game always means a particular suspect (the glowing Bear/Owl always means the doctor, if you’re curious). However, I’ve played the game with members of that group and outsiders on occasion and oddly the members of the group who “know” the meaning of that card sometimes forget that the card doesn’t mean that to anyone outside the group. Interestingly, the building up of those sorts of private language elements is what eventually limits the re-playability of Mysterium among the same group over time, but it’s a strength of Turing. Language is structured and as such a little more brittle than direct interpretation, meaning that if players examine the images offered to them every turn to make direct interpretations of the images the game can’t take advantage of them. But if they build up a private language where a particular image is always chosen by the human, or even a particular human, the game is capable of stealing that image away from the player, at which point the private language becomes a huge disadvantage.

This is one of those emergent moments in design that I had a vague and undefined idea of being a strength at the start of development but has come more and more clearly into sight with each additional play through and test. I was pretty sure that the basic idea would work as a simple filler game once I worked through a few scoring and presentation mechanics (which I’ll go into on the next diary) and I had a suspicion it would do something interesting if players sat down and played it through repeatedly, so it has been rewarding to see the game pull out odd little moments where through random chance it manages to perform a direct impression of a human player.

The downside at the moment is that image interpretation games are often seen as only being interesting until all the images have established meanings, hence why more and more image cards are generally offered to them over time. Turing becomes more interesting when the images have established meanings, which puts the number of images to include in the game as being a tricky choice. Too many and the odds of the game hitting a convincing sequence goes down, plus players are less likely to develop an internal language that the game can subvert. Too few and players will likely become bored of the images before they naturally develop a private language. Unfortunately, this is one of those number choices that traditional statistic methods struggle to interpret, or at least that my statistical capabilities aren’t up to figuring out. Fortunately, I’m quite of a gut designer anyway, so at the moment its 90 cards. That sounds like a good number and we can always pull some out over time.