Teaching through the use of online tools is an increasingly popular strategy for bringing scientific knowledge directly to students without the need for a classroom. The rapidly changing economics of university education are encouraging educators and institutions to shift some educational activities to less expensive online formats that aim to improve accessibility of college education for future generations of students. The status quo for education in the physical sciences involves a lecture format similar to those in non-science disciplines, which is supplemented with hands-on laboratory courses. Here, the students aim to carry out in practice what they have learned in principle. The fundamental theory behind this approach is that knowledge gained in class will become cemented as students apply their new-found understanding to solve real-world problems. In reality, this often results in excessive classroom time and high costs, to both the student and the institution.

Online education provides an opportunity to achieve these objectives, and it will as long as the tools take advantage of the inherently cooperative and interactive nature of emerging technologies. Active (and by some counts successful) examples of online course curricula range from online-only or online-supplemented courses with interactive assignments1,2,3, to in-class projects where students create new online resources4 — such as Wikipedia articles — as part of the learning process. Although remote education is often met with scepticism regarding its efficacy or benefits, there is strong evidence to suggest that this approach to teaching is equally as effective as traditional in-class lectures. Students who take the online-only classes — with no formal in-person contact from the professor — do just as well as those who participate in the exact same course offered in a brick-and-mortar setting1,5.

Holding the attention and focus of students, and providing information in a meaningful way, is a significant challenge for online education (and all education for that matter). Imagine having educational content that would not only command the complete focus of students, but even excite them so much that they would consume it without needing to be assigned it. As it turns out, video games have been exciting, entertaining and commanding the focus of millions of players for decades6. Humans have collectively spent more than 1.75 billion hours of their time playing the online world building game, Minecraft, which if measured linearly would predate the birth of Homo sapiens. Minecraft's popularity ultimately led to its purchase by Microsoft for US$2.5 billion dollars, a move generally described as purchasing not just a game, but access to the entire next generation of players7.

At its peak, the massively multiplayer online game World of Warcraft had over fourteen million active subscribers, and its community of players have populated multiple wikis with hundreds of thousands of pages of information, strategy, and theorycraft — far more than your average college textbook. Additionally, many of these players' discussions in forums and community sites can be considered scientific inquiry, suggesting that this kind of discourse could be used as a viable alternative to textbooks and labs8. Blockbuster game franchises such as Halo, Grand Theft Auto and Call of Duty have spawned numerous sequels whose release dates are treated like major events, to the point that the release of Halo 3, which garnered US$170 million in sales in its first 24 hours, was blamed by Hollywood executives for the poor performance of their summer blockbuster movies in 20089. The popularity of video games has resulted in the growth of an industry with multibillion dollar revenues (over US$22 billion in 201410) that now produces game content for whatever device or platform suits you, from mobile phones to high-end gaming computers.

Only a few years ago, games were played in arcades or at home with only a few players at a time, usually with friends or family. Now players can work cooperatively in massive online worlds with thousands of people that they have never met physically and may be on different sides of the Earth. The video game world, like the world of education, has gotten much more connected. Author and game designer Jane McGonigal writes in her bestselling book, Reality is Broken, that games — including those without apparent educational value — make us better, more productive people by allowing us to exercise our minds, connect with loved ones in different places and even recover from physical injuries11,12. Could they make us better teachers and students too?

Levelled-up lessons

There are many advantages to providing educational content in a computerized game format, one of which is that it permits learners to move at their own pace. A horizontal learning approach allows students who understand the material quickly to move on to more challenging material, whereas those that need extra time can take it, making it truly personalized6. Management of the pace of learning is extremely important when interacting with students remotely. Educational content must be provided in such a way that they do not become bored (the material is not challenging enough) or frustrated (the material is too challenging)13,14. A key advantage in our minds is the immersive nature of a well-constructed video game. While playing games, one must be constantly focused on any number of virtual priorities — for example hunger, health, resource management and external dangers — which combine to keep players hyperfocused on the job at hand. Game designers often describe this heightened concentration as being a state of flow15, and often design mechanics and content with the express intent of helping players achieve and maintain flow16. This focus is exactly what we want for students learning a new skill. Furthermore, the deep immersion also allows an immediate and elevated sense of feedback, which has been shown to improve learning outcomes17.

Although classroom environments might give feedback as often as once a week in the form of graded assignments, current online educational platforms far exceed that by giving a right or wrong status as soon as a learner inputs an answer. We think video games can improve on this again: they are constantly giving feedback at > 60 Hz on everything the learner does, from moving the avatar (low-level feedback that gives students a sense of control) to solving problems in the game (complex feedback that may tap into reward circuitry in the brain). This also leads to focus and determination on the part of the student. Most video games already have mechanisms in place to help find and maintain a balance of learning that feels challenging but achievable. One such mechanism is the challenge structure of most digital games, which constantly and continually assesses the player's skill in real time through entertaining gameplay18. Another common method is the use of asynchronous aperiodic rewards (long known among psychologists as a strong motivating factor in human behaviour): finding unexpected rewards gives a huge boost to a player's desire to keep playing a game, even when the game seems very challenging and the prospect of failure is high. This leads to another important advantage for learning — most students are already comfortable with the idea of failing in a video game. Many (if not most) digital games encourage trial and error in a fun, relatively consequence-free space, and failure itself has generally been considered to be one of the staples of both good game design and good gameplay19. Allowing students to learn in an environment where failing can not only be acceptable but maybe even fun allows students to try new things without being intimidated20. This in turn is another inducement to keep trying through failure, to keep trying to learn something new even when it is very complex and very challenging.

If video games are so great at doing what traditional educational methods are not, then why don't we make an analogue of Grand Theft Auto V to teach organic chemistry? Games such as World of Warcraft, Eve Online, Halo 3, Civilization and Minecraft present ecosystems where players collaboratively tackle applied problems and voluntarily develop, eagerly populate and addictively consume content. Why doesn't higher education adopt these practices? Games like Grand Theft Auto V generate huge buzz, sales and user bases prepared to pay big money for a fun game experience, but little more once the game is over. Developing really successful games from scratch, as it turns out, is pretty hard — and costly. Rockstar Games spent an astounding US$265 million on marketing and development for Grand Theft Auto V, which was wildly successful and earned its enormous budget back many times over. On the other hand, the reported fiscal year 2015 budget for the National Science Foundation's entire chemistry research division was US$237 million. For approximately 100 hours of gameplay, Rockstar Games spent US$2.65 million per hour to engage players, whereas we faculty staff have far fewer resources per hour of classroom lecture. From a financial standpoint, higher education simply cannot compete.

Funding is not the only barrier for educational game development. Developing an engaging, well-designed game for art or entertainment purposes is difficult enough without the additional constraint of creating something educational. Many commercial game developers are simply not interested in educational game development or do not have the resources to pursue it. Game developers generally do not have a great deal of experience in the specifics of teaching an educational field, much as educators and academics generally do not have experience in game design and development. And publication routes for educational games can be complex, particularly when dealing with the structures of primary, secondary and higher education.

Fortunately, there are still many examples of excellent educational games, many of which do not require the massive budget of Grand Theft Auto V. Successful educational games include classics such as Oregon Trail and Kerbal Space Program (the latter of which is commonly cited as “orbital mechanics: the game”, and is now in partnership with NASA for educational purposes: http://go.nature.com/2foEe4L). Meanwhile mobile device apps exist that allow you to learn multiplication tables or practice organic chemistry problems. These efforts have been successful at creating games that do what they set out to do — create interactive educational content. Some games such as FoldIt have even been designed to both teach and solve problems in research simultaneously: it is a game where players learn the concepts of how proteins fold by twisting computer models of unfolded peptides into the correct shape (http://go.nature.com/2gKOzhc). The players are scored based on how well the molecular interactions are optimized. Ultimately, the designers of this game want to 'crowdsource' this gameplay to learn more about the folding behaviour of proteins whose structures are not yet known. Other games use a simple design that takes higher-level content and allows players to access it through a convenient platform, such as a mobile device. Chiral, a game developed at the University of California, Irvine, allows students to practice identifying stereochemical relationships between organic molecules on their phone or tablet. Students and researchers at the University of Hull have used Minecraft to build a publicly accessible world of giant chemical structures that allow students and players to explore them in 3D through Minecraft's interface (http://go.nature.com/2h5IFXf). Despite these valiant efforts, nobody has yet been able to develop a video game that is able to completely replace an entire classroom-based course. We believe that this should be the next challenge.

Modding

To try and meet this challenge, and avoid the major hurdles of developing a game from scratch, we have taken an approach that relies upon the development of a modified version of an existing video game. We chose the popular video game Minecraft, because of its extensible nature and broad accessibility. A Minecraft license costs less than US$30, can run on nearly every common operating system, requires no advanced computer graphics upgrades and can be played entirely with a keyboard and mouse — in other words, nearly anyone with a computer can play it. Minecraft is a popular world-builder game in which each voxel (or volume element, a 3D analogue of pixel) in an artificial world represents a block of material such as wood, iron, gold, water, glass and so on. Players are immersed in a 3D world where they collect resources to build better tools, collect rare resources, and begin to build civilizations on massively multiplayer servers. Players must ward off threats such as starvation and enemies that emerge in the night to attack them (including their structures and stockpiles). There are no levels to reach and no set missions that must be accomplished. This allows ample space for each player to customize the game to their desires, with only their own imaginations to satisfy. In short: unmodified or 'vanilla' Minecraft has everything we want in a video game for teaching — except for the content.

Polycraft World. Our modification of Minecraft (called a 'mod' in the Minecraft community) integrates concepts of chemistry and engineering into the gameplay. Named Polycraft World, the development of the mod maintained two main goals — the science we add must be accurate, and it must add something fun to the game. The Polycraft World expansion is offered for free (http://polycraft.utdallas.edu), and can be used with any valid Minecraft license. The vanilla Minecraft interface is designed so that players can combine raw materials in certain ratios and patterns to generate new items — for example, a specific arrangement of wood sticks and iron ingots will result in a sword. A slightly different arrangement could result in a pickaxe or a shovel. The Polycraft World mod furthers this capability by adding chemical ingredients. If you combine the right reactants (in a chemically balanced way) you will get a new material such as Kevlar. This is a pretty simple concept from a game design perspective.

Of course, these chemicals do not come from thin air. Figure 1a shows the pseudo-realistic refinement process, or as it is called in the gaming world, a tech tree. It outlines all of the successive refinements you would need to carry out in order to obtain the refined petroleum necessary to make fuels and commodity polymers. Polymers such as polyisoprene and polyethylene can be made into new items such as pogo sticks or plastic chests, which improve the player's capabilities within the game. Kevlar armour can be made to protect the player from enemies, and polyether ether ketone polymers can be used to make more efficient tools. By coupling practical concepts such as distillation, extrusion and chemical processing to attaining highly desired items in the game such as enhanced tools, jetpacks and flamethrowers, we hope to strike a balance between being engaging and deeply educational. By further coupling the player's motivation to better manipulate and terraform the virtual world to real-world processes, then learning can take place in a way that does not feel like learning. The real question is why would anyone want to do this? If the game plays like a classroom assignment, there is no incentive to play through the educational parts. By linking more advanced concepts such as polymer synthesis to highly desired items in the game such as overpowered tools and jetpacks, the motivation to learn will be augmented by the student's desire to obtain these items. In short, learning more science makes the game more fun and the player more powerful in the game.

Figure 1: Anatomy of a jetpack.
figure 1

a, The pathway a Polycraft player must navigate to obtain a jetpack. Players must convert natural resources (crude oil) to manufactured products using distillation, chemical synthesis and manufacturing. b, Crude oil can be obtained by building an oil derrick and placing it on special oil-containing blocks that the player must search out in the game. c, The crude oil collected must be refined and purified before it can be used as a fuel or a reagent for further chemical reactions. d, New inventories, similar to chemical processors, enable the players to use chemically accurate organic transformations to synthesize the fine chemicals needed to make the advanced polymers and rubbers required for a properly functioning jetpack. e, After working through the chemistry, the jetpack rewards players with new in-game capabilities such as the ability to observe the landscape from above and travel much more quickly. See http://go.nature.com/2hjb2l1 for the 'How to make a jetpack' video tutorial.

Full size image

Integrated learning. Polycraft World is not designed such that all the learning takes place inside the game itself. Vanilla Minecraft uses an online wiki to teach new players basics about how to make items in the game, and we adopted that approach, creating a user manual with more than 2,000 online wiki-style help pages (Fig. 2). Players obtain the recipe for new items from an online wiki. Rather than teaching fundamentals from the ground up, players are enticed by the promise of enhanced gameplay. With clickable icons, players learn what is needed to make new items by starting with the desired item and working backwards. In synthetic organic chemistry, this process is known as retrosynthesis, and has been the go-to method for devising synthetic strategies for complex organic molecules for decades. The interactive nature of a wiki page gives players the ability to explore each concept at their own pace and interest level. This deeper learning is, of course, not compulsory, but if using an 'injection molder' in Polycraft World sparks an interest in someone who otherwise would not have been exposed to it, more information is available to them — encouraging curiosity-driven learning.

Figure 2: The design of the Polycraft World wiki page is such that the game instructions (recipes) are intertwined with real scientific information such as chemically accurate reactions and conditions and clickable links.
figure 2

a,b, Each wiki section (a) begins with chemical information and facts which leads to (b) instructions on how to make the item in the game. These reactions are representative of the real-life reaction (including balanced chemical equations). By starting with a desired object or molecule, and working backwards through the wiki to obtain recipes for each material needed to make it, the students and players are replicating the process of retrosynthetic analysis.

Full size image

Fabrication of each of the new items in Polycraft World requires multiple levels of chemistry knowledge. This ranges from knowing that rubber is soft and bouncy (thus suitable for a pogo stick), to understanding the substituent effects on aromatic ring substitution reactions. For example, using basic game skills from Minecraft, players can craft simple 'tree taps' to harvest natural rubber (polyisoprene), which can then be processed into items like pogo sticks and running shoes that in turn allow the player to run faster and jump higher (see the 'Crafting a pogo stick' video tutorial: http://go.nature.com/2gdIG6L). More complex and powerful items such as Kevlar armour, however, require more knowledge. Shown in Fig. 3a is an example of how basic undergraduate organic chemistry information can be weaved into the gameplay of Polycraft World. In order to make the monomers needed to synthesize Kevlar, one must understand the electronic effects of substituents on aromatic substitution reactions. If the wrong reaction sequence is used, the wrong polymer will be obtained and the player will not be able to craft stronger armour to protect them in the game. This example illustrates a mechanism not only for combining hardcore fundamental organic chemistry into a video game, it also provides a pathway for teachers to evaluate student performance — how many times was the reaction attempted before the correct product was obtained? Did the student use the Polycraft World wiki page for information or did they search the web while playing the game to solve the problem?

Figure 3: Examples of approaches to teach and assess basic organic chemistry using in-game content.
figure 3

a, The synthesis of p-phenylenediamine, a component of Kevlar, requires an understanding of how substituents affect the reactivity of aromatic rings. b, A Minecraft style interface for a test question from an organic chemistry exam at the University of Texas at Dallas.

Full size image

If games are to work as effective educational tools, an important design characteristic is the seamless integration of the educational content into the game. Players will interact most effectively with educational material if it has the same look and feel as the rest of the gameplay so that it is not viewed by the player as something separate from the normal game. Disruption of the interaction between the player and normal gameplay could discourage the voluntary use of these features within the game and thereby limit their effectiveness. Shown in Fig. 3b is a prototype testing interface that can be used to adapt typical organic chemistry quiz or exam questions into a Minecraft format. This interface uses the same gameplay concepts of Minecraft, namely dragging flashcard icons that represent items made or collected in the game, into a specific, predefined pattern that generates a new item or response. We believe that this type of interface, combined with chemical structure drawing capability, would clear a path between normal Minecraft gameplay and the ability to teach a conventional organic chemistry course using a Polycraft World-based textbook.

Although Polycraft World's gameplay is fundamentally rooted in the concepts of organic and polymer chemistry, the process of building new molecules, materials and machines will develop more general problem solving and teamwork skills that are applicable outside of the fields of chemistry, engineering and polymer science. As students are working in a cooperative online environment, the instructors can observe them from within the game, either silently or interactively, to witness their problem solving and reasoning processes in action. The Polycraft World mod contains a server-side analytics package that collects data in the form of mouse clicks, keystrokes and web browsing so instructors can observe behaviours such as where students begin a task, what mistakes (or changes to their approach) they make and how they interact with one another. The data obtained from the student's activity online would provide both the student and educator with a personalized outline of processes used to solve problems, and indicate where deficiencies can be addressed. The analytics engine can provide data as simple as how many times a student tried to make a certain chemical before they got it correct within the confines of a simple classroom assignment, to data as complex as long-term outcomes describing the pathways students may take before achieving proficiency in a subject.

Results

We have some preliminary evidence that this approach works. Over the autumn 2015/spring 2016 academic year, Dr Christina Thompson taught a one hour general topics class entitled 'Video Games and Learning' to a small (thirteen students in each class) group of students of mixed subject background and age, through the Honors College at the University of Texas at Dallas. No in-class science instruction was given, but a non-graded part of the class was to play Polycraft World, with two checkpoints given during the semester to ensure that students had progressed in the game as far as creating their own distillation column. Crafting a distillation column in the game requires advanced polymers, so we knew that the students had at least been exposed to their synthesis. Eleven weeks into the class the students were given a pop quiz written in entirely scientific language asking them questions ranging from identifying the chemical components of polymers based on their acronyms, to drawing a distillation tree for crude oil given a blank piece of paper. Remarkably, in the autumn semester five of the thirteen students could correctly draw a crude oil distillation tree to three levels of distillation, with three more able to draw it to two levels of distillation. In the spring semester we saw very similar numbers, with four of the thirteen drawing three levels of distillation, and a further three students correctly drawing two levels of distillation.

There is also some correlation between the polymers that students could correctly identify from their acronyms with how advanced that polymer is in the game. Seventeen of the twenty six students could correctly identify LDPE as low density polyethylene (a low-level polymer that is relatively easy to make and useful in many contexts in the game), whereas only four of the twenty six could correctly identify EVA as ethyl vinyl acetate, a much more advanced polymer that they would not necessarily have made given their advancement in the game. What should explicitly be stated is that none of these students were asked to learn anything: no memorization was required as part of the class or their final grade. Students learned the real-world processes required to get benzene from crude oil because they wanted to make jetpacks in a video game. Now the number of participants thus far is of course quite small, offering only anecdotal information on the effectiveness of this mod in actually teaching students about science, but it is encouraging that the students appear to be learning some real science even when there is no traditional instruction, or grade-based incentive to do so.

Outlook

Just as not all gamers are drawn to the same video games21, not all students will be captivated by Polycraft World. This is another reason why the community must work together to develop a wide range of games that are both entertaining and educational, and decades of video game development has led to a variety of genres that appeal to different types of players. Civilization, a popular turn-based strategy series, spawns players in a world devoid of human ingenuity and asks them to build cities, units, buildings and wonders of the world in a quest toward culture, science, political prowess or military might. The game has evolved considerably over is various incarnations into a captivating experience with nearly infinite replayability. In Civilization V, for example, you can choose from among more than forty civilizations with exclusive traits, to build a variety of buildings and monuments that give each civilization unique bonuses and improve cities' abilities to make its citizens happy, well fed, more faithful, more productive, and culturally and financially richer. There is no doubt that players learn many things from Civilization despite not being designed as an educational tool: the third iteration, Civilization III, was studied extensively as a way to teach history to underrepresented, poorly served, or simply digitally invested populations of students22,23. How much more educationally valuable might the game be if the academics of history and engineering partnered with the developers for future releases of the game?

So where does this leave researchers and educators who aspire to create educational games that are just as fun as commercial games, and draw the same interest? Collaboration would be a good place to start. Symbiosis between the chemical industry and academic researchers has been greatly beneficial for the development of new pharmaceuticals, polymers, fabrics and many of the healthcare and consumer products that shape our modern life. Maybe researchers and educators should start fostering these same kinds of relationships with game designers and companies. Players of adventure games could learn to heal their characters by obtaining real molecules from the game's environment (quinine, antibiotics); using real chemistry rather than conjuring fictional cures from electronic alchemy. First-person shooter games could teach the concepts of brisance and ring strain to those seeking to build more powerful explosives to blow up their enemies (or friends as it often tends to be). Could researchers be encouraged to translate their research into such media? We would go so far as to encourage all video game developers to allow for academics to create mods for their games that could incorporate educational elements.

Although it may initially seem farfetched that video games could play such a large role in education, consider that the Fédération Internationale de l'Automobile — the worldwide governing body for auto racing, and responsible for licensing drivers in Formula One and rally racing — has recently announced a long-term partnership with the game studio Polyphony Digital (creators of the racing video game series Gran Turismo) to create accurate models of real-life racetracks. Eventually, players who master these virtual courses under the supervision of this programme will be able to use their in-game progress towards obtaining a real-world racing license (http://go.nature.com/2glFJD2). If video games have become sufficiently accurate to teach and evaluate a player's ability to drive a high-performance car at high speed, why shouldn't they be able to do the same for molecular orbital theory?