Foldit Critique
The game I’m reviewing this week is one I’ve heard a lot about; FoldIt. This game was developed back in 2008 as a joint effort between the University of Washington Center for Game Science and the Department of Chemistry. This is a puzzle game about folding proteins, and the goal for the players is to maximize their score by creating better folded structures. The game can be considered a crowdsourced game, and players’ solutions have been used by researchers to create novel protein structures. It’s available on PC, Mac and Linux.
While the game deals with advanced biological concepts, it is meant for the general public and novices to be able to play as well. This is the first game I’ve critiqued where I pretty much know nothing about the subject area, so in a way I was testing the game’s objectives on myself (spoiler alert: I still feel like I know very little about the subject matter :/).
Note that I played the publicly available standard version of the game, and played about 18 levels of the campaign. I noticed after my play session that there is an education mode that I was not aware of. As such, this critique is just about the standard version of the game.
Learning Objectives
The game is interesting in that its primary objectives are not just to educate players about how protein folding works, but also to leverage the player base to create novel protein structures for researchers to use. In fact, the bulk of the publications by the team behind the game have been about the crowdsourcing aspect of the game, as opposed to studies on its performance as an educational game. Protein folding is a hard computational problem, so by leveraging its player base, Foldit has become quite famous for the novel protein structures its players have come up with.
Despite its crowdsourcing goals, the game still aims to be widely accessible to people who are not familiar with protein folding. To this end, the game aims to do the following:
- Expose novices to protein folding, and inform them of the difficulties in executing the task using computers
- Teach players about the essential terminology of protein folding
- Teach players about the operations that are possible when folding proteins (e.g. hydrophobic side chains can be rotated inwards)
- Provide players with the opportunity to enhance their protein folding skills and derive more optimized folded structures
- Advanced goal: Allow players to design their own protein structures
In terms of prior knowledge required before playing, the game implies that most novices can get started without much trouble. In my opinion, I think previous exposure to Chemistry and Biology, either at high school or college level, is necessary to get the most out of the game and actually learn rather than just contribute to the crowdsourced protein folding data.
Game Elements
In Foldit, the goal is to manipulate the protein structure using the mouse and provided actions to reach the target score. Progress towards the score can be made by reducing clashes between side chains, eliminating voids between helices and backbones, burying the hydrophobic sidechains inwards so that they do not get exposed to water, and maximizing hydrogen bonds between sheets. There are 32 levels in the campaign, but the first 16 are quite tutorial heavy and don’t have a “final boss” puzzle like later stages. As the levels progress, additional capabilities and actions are opened up to the player, and later levels allow the player to design their own protein structures.
This game relies heavily on text-based tutorials, which is where most of the heavy lifting is done in order to get novices up and running with protein folding.
Using Zubeck’s updated MDA framework, the game elements can be broken down as follows:
Mechanics — The game screen has a 3-dimensional protein structure and various 2D interface elements around the screen (similar to an IDE or CAD software). Players can manipulate proteins by using the mouse and certain keys (e.g. ctrl, shift) to directly manipulate the protein. Across the bottom of the screen, there are various menus including the help menu, design mode, undo menu (which is more robust than I thought it would be, it’s not just a single undo button), and the action menu. The action menu is where players can use abilities that manipulate the protein for them, such as shake (which tries to eliminate clashes) and wiggle (which tries to eliminate voids). These two features in particular seem to be automating the same player manipulations that are possible via the mouse; they were respectively introduced after players learned how to eliminate clashes and voids using the mouse. The game is mostly single player, however the online mode allows players to share their scores online, which is how I assume the crowdsourcing is done.
Systems — There are a few systems at play in Foldit. There is a text-heavy help system that includes on-screen hints as well as text to educate about the content itself. For example, in the early levels, the on-screen hints explain how to manipulate the sidechains, but they also provide an option later on to learn more about how sidechains work. The game also has a score system where there is a minimum score the player must reach to target the level, however they can go beyond that score (and are encouraged to do so). The exact breakdown of what contributes the most to the score is still a little vague; there is some green text that pops up every once in a while if you bury the hydrophobic side chains that say something along the lines of “buried hydrophobic side chains, +300 bonus”, but this text does not always appear for other operations. There is also a robust protein manipulation system at play that includes mouse manipulations and many additional actions; on top of shake and wiggle, there is a design feature that allows players to change the protein structure. Finally, the undo system is also more substantial than anticipated in this game. Rather than just a single undo button, there are multiple undo options such as “return to best score”, and even a graph to show the moves that were made (although I’m still unsure what exactly is going on in this graph)
Dynamics — The game offers immediate feedback to the players if there is a clash or a void present in the structure using visuals and text. This encourages players to revisit their actions that led to the errors and to manipulate the 3D view of the protein to find better solutions. The game does encourage players to experiment, especially with the incentive to maximize the score. However the complexity of the undo menu, and how slow it is to access on this older interface, means that the explore-then-undo behavior is slower and thus players may be hesitant to experiment. The overuse of text also discourages players from learning more about how protein structures actually work (personally I was so tired of reading text by level that I just stopped processing the later tidbits the game provides). Finally, and this is definitely not ideal, the provided shake and wiggle features of the game, combined with the poor scaffolding and educational content design, means that players may resort to brute force methods to complete the puzzles.
Aesthetics — The game doesn’t have a clear cut identity or a unique visual aesthetic, it very much feels like 3D modeling software. There is music and sound effects present in the game though, which combined with the dark mode theme contribute to a sci-fi feel. The score system is meant to motivate players to replay levels and achieve better protein folds, which can result in feelings of accomplishment in players. Unfortunately however, the most dominant feeling I had from playing the game was confusion due to the overwhelming use of text to teach players about how the protein folding domain. This confusion, and occasional frustration from the lackluster interface helped contribute to the brute force behavior I had later on.
Learning Principles
FoldIt tries to employ scaffolding its design. The features are slowly added level by level so as not to overwhelm the player. It also tries to scaffold the educational text, which is mostly offered in small chunks of text, however the text can still result in some cognitive overload. The game does make good use of in-game feedback and visual cues with the clashes and voids. Levels are relatively self-contained, but the level map does emphasize what knowledge is necessary from previous levels to complete the upcoming one.
The game employs well is spatial contiguity, however I would have liked to see the modality principle being used more. The use of immediate feedback is also worth commending. I always knew when a clash, hydrogen bond, or void was introduced as I was moving the mouse.
Overall Critique
I know I said this last time too, but this was more of a negative game experience and I was more critical than I anticipated I would be. I had heard so much about this game, but in hindsight I realize that most of what I’ve heard has always been related to the crowdsourcing side of things. That being said, I became really interested in the subject area while playing, which frankly I didn’t know much about before. This interest is likely why I was disappointed with this game; I was so eager to learn more about how protein folding worked, yet the way Foldit supports this interest is by bombarding players with text. I think this game is clearly special in terms of its accomplishments, however there is clearly much room for improvement to make this game approachable to those who want to learn more about the area. I would recommend this game as a supplement to students who are already learning about protein folding, but not those who have not had any exposure. Perhaps the education mode does a better job supporting the learning outcomes of this game.