Key takeaways
- The Experimental Gameplay Project ran at Carnegie Mellon’s Entertainment Technology Center in spring 2005, with four graduate students and Jesse Schell as advisor.
- Three rules: under seven days, exactly one person per game, and a common theme. Fifty prototypes came out of one semester.
- Tower of Goo was one of them. World of Goo, its commercial descendant, arrived in 2008.
Two things on this domain share a name and are constantly confused. The Experimental Gameplay Workshop is the GDC session. The Experimental Gameplay Project was a student research project at Carnegie Mellon. They are related in spirit and separate in fact.
The three rules
Matt Kucic, Shalin Shodhan, Kyle Gabler and Kyle Gray set themselves constraints that sound arbitrary and turn out to be load-bearing. Each game had to be finished in under seven days. Each had to be made by exactly one person. Each had to be built on a shared theme.
The seven day limit removes the possibility of polish, which forces the question of whether the core idea works. The one person rule removes coordination, which means nobody can hide behind the team. The shared theme makes the results comparable, which turns fifty separate experiments into a data set.
Fifty prototypes
The output of one semester was around fifty games. Most are footnotes. A handful – Tower of Goo, Gravity Head, On a Rainy Day and others – circulated widely enough to be remembered, and the group’s write-up of what they learned became one of the most cited pieces of practical design writing of the period.
The interesting statistic is not the fifty. It is that the group kept going after the semester ended and published the rules for anyone else to use, which is why the format outlived the project by twenty years.
From Tower of Goo to World of Goo
Tower of Goo is a physics toy: balls of goo, joined into a structure, stacked upward until the structure fails. It is playable in a minute and has no progression at all.
World of Goo, released in 2008 by 2D Boy, is that toy turned into a game – levels, goals, a world, a soundtrack and an ending. It became one of the defining independent successes of its era, and the distance between the prototype and the release is the whole argument for prototyping: the mechanic was proven in a week, and everything after that was production.
Why the rules still get copied
Every game jam since owes something to this format. The seven day constraint in particular has become a default, and the reasoning behind it is still sound – a prototype that takes a month is no longer testing the idea, it is testing your endurance.
Our own view is that the one person rule is the part people quietly drop and the part that matters most. A prototype made by four people is a negotiation, and the negotiation smooths out the weird decisions that make a prototype worth looking at. The Carnegie Mellon rules produced fifty strange games precisely because nobody had to agree with anyone.
The relationship to the workshop
The GDC session and the Carnegie Mellon project are separate, but they were part of the same moment – a period around the middle of the 2000s when the industry briefly decided that small unfinished experiments were worth showing in public. That period is what this domain’s workshop archive records.
The group’s own write-up of the method is on Game Developer. For the session that ran alongside it, see the workshop explainer.
What fifty prototypes taught them
The group published what they learned, and the findings have aged well. Polish is not what makes a prototype work – a game that is fun in a week is fun because of its core, and a game that is not will not be rescued by another month. Constraints generate ideas rather than limiting them, because a shared theme forces fifty people down fifty different paths from one starting point. And most ideas are bad, which is only a problem if you have invested in them before finding out.
That last point is the real output. The value of the seven-day rule is not the games it produces; it is the speed at which it kills the ones that do not work.
Why game jams inherited the format
Every weekend jam since is a descendant. The time limit, the theme, the enforced smallness – all three came from here, and the reason they spread is that they are the cheapest known method for generating design variance.
What jams generally drop is the one-person rule, and the effect is visible. A four-person jam team produces a more competent game and a less surprising one, because every strange decision has to survive three other people’s reasonable objections.
The distance between a toy and a game
Tower of Goo takes a minute to understand and has no ending. World of Goo has levels, a difficulty curve, a world, a soundtrack and a conclusion, and took two more years to build. Everything added in those two years is production rather than invention.
That ratio – one week of invention, two years of production – is the honest shape of the work, and it is why a prototype session is worth an hour of a conference. The invention is the part that can be judged quickly.
Images: 2D Boy, via the World of Goo Steam store page; conference photography via Wikimedia Commons (CC BY).