Department of History, Carleton University, Ottawa, Ontario, Canada. Shawn.Graham@carleton.ca
Cite this as: Graham, S. (2016). Review of Evolving Planet [game], Internet Archaeology 42. https://doi.org/10.11141/ia.42.4
Platforms: IOS, Android. Cost: Free. Players: Single. Completion time: 3-6 hours. Languages: English, Spanish, Catalan. http://evolvingplanetgame.com/. Created by Murphy's Toast Games, the Supercomputer Centre in Barcelona, and the Simulpast Project
It's the year 3016, and you are in charge of an archaeological expedition to the planet Kepler-1138. Your aim is to know what happened to the Lovans, humanoid aliens that became extinct for unknown reasons. You will use artificial life to replicate the story of the mysterious species. Will you develop their technology, make them experts on warfare or strengthen their cultural influence? Choose your strategy carefully to reveal the past of the Lovans, and also their future.
Is this a good game? Yes. It is challenging, engaging, with beautiful artwork, a compelling backstory, and appropriate interactivity given its smartphone-based platform. It is in fact a very rare collaboration between computer scientists, game developers, and archaeologists that results in a casual game designed to convey important ideas in archaeology. Is it good archaeogaming?
The core of Evolving Planet is an agent-based model situated in the archaeology of hominin dispersal, wrapped in the trappings of a casual video game. The goal of the game's creators is to communicate to non-archaeologists something of that archaeology and the methods used to study it. Agent-based models (ABM) have been a part of the archaeologist's computational toolkit for quite some time (and have a rich vitality, cf. the CAA2016 papers here). Carefully formulated to describe a level of complexity below the targeted phenomenon of study, their value is manifold as the researcher studies what emerges across thousands of runs, sweeping through the landscape of possible outcomes. An ABM can be a useful tool to help think about past society, a kind of lens for refracting our understanding (Sinclair and Rockwell 2016, on the general use of computation in the humanities; Lake 2014; Wurzer et al. 2014 on agent-based modeling's histories and trajectories). That is to say, it's not just their results that matter in an ABM, but also the exercise and discipline of translating our stories about the past into code. Unless we are explicit about what we think about the past, we cannot encode it. Thus a carefully wrought ABM should make clear the model builder's assumptions about the past and her theoretical positions (on validating ABM, see the useful bibliography at Tesfatsion 2016). If the code is made open source, ABM allow others' scrutiny of those assumptions, that encoded representation of the world. To make an argument in the code itself, then, is a valuable scholarly exercise. Video games on the other hand, themselves quite complex or complicated simulations, do not (often) allow others to peruse their code, or discern the game's rules except by virtue of play. To play a game well means that the player has become proficient at absorbing and integrating the hidden rules (the representation of the world) through the performance of play. This is what gives video games their power to communicate ideas about how, for instance, society, culture, x, worked in the past. To be good at such a game is to necessarily internalise others' ideas about how the world works. On the other hand, it is often not clear (nor can it be clarified) how the rules actually work. One must play over and over again to determine the balance of probabilities that govern the game's world rules.
An agent-based model can be thought of as a game that plays itself. The investigator sets it up to simulate a particular starting condition (state of the world) and presses the 'go' button. Then she carefully studies what emerges from the non-linear and unpredictable intersections of rules (that describe our ideas about the past) with the environment (frequently, in archaeology, from GIS). Why not then merge the powerful methodology of agent simulation with the immersive learning potential of games, as a work of public archaeology, of public pedagogy? This is what Evolving Planet sets out to do with regard to the evolution of hominins, albeit with mixed success.
In Evolving Planet, a terrestrial world has been discovered, replete with artefacts and sites from a now-extinct intelligent species, dubbed the 'Lovans'. You (the player) are the distinguished archaeologist sent from Earth to oversee a project trying to work out why the 'Lovans' became extinct. You are briefed on the project by the archaeologist on site. The research program involves creating a swarm of bio-robots (Figure 1), artificially intelligent forms to whom you give various abilities, depending on the 'mission'. The missions are framed around a range of research questions: how did the peopling of this planet unfurl? How did the eventual dominant species, the 'Lovans', interact with other intelligent species (who became extinct before the Lovans)? For instance:
Our bots will need to interact with other groups during the archaeological expedition. The planet was not populated only by Lovans, just like the Earth was not a Homo sapiens-only club. We shared Eurasia with Neanderthals for 5,000 years. They were also hominins quite similar to us in many respects, including the use of technology, culture and even language. The Neanderthals disappeared as a species over 30,000 years ago. However, our DNA says that each of us has a little bit of Neanderthal inside. And trust me, DNA never lies!
Thus primed, the robots are turned loose on the planet's surface, and you the archaeologist watch their progress via a map (Figure 2). You can direct their movement by placing markers on the map; you can watch their population numbers grow or fall (and take actions to alter that trajectory). The critical task is to make sure the robots reach their targets within a particular time window (which is based on the archaeological termini found via excavation on the planet's surface, which is part of the game's backstory). Complete the tasks in a(n archaeologically-grounded) timely manner, and you are rewarded with badges that at the end of the game can be examined and which describe various archaeological issues from Earth's history. It is not clear (at least to me, during my playthrough) how these badges are earned or if they alter game play. Robots have 'evolution points' that tally up the longer the bots live; you get a set amount when you start playing. They can be spent on mobility, reproduction, and technology abilities. There are also power-ups and boosts, and presumably the combination of how these are used has an impact on which badges are earned when.
In short, you are God. And therein lies the problem.
The game is meant to convey to the player the research problems surrounding the study of hominins here on Earth. The various missions replicate real questions that motivate study of Neanderthals and other hominin species. It is also meant to communicate something of how palaeoanthropologists approach some of these problems, through simulation and modelling. At the recent Interactive Pasts Conference at Leiden University organised by the Value Project (4-5 April 2016) Xavier Rubio-Campillo, one of the project leads, described the project and its archaeological aims. The session is available on Youtube, and is well worth watching.
Rubio-Campillo describes both what does and doesn't work. The 'God problem' is perhaps the most significant (as Rubio-Campillo himself discusses), as it emerges out of the awkward marriage of agent model with player agency, and inadvertently promotes so-called 'intelligent design'. In the remainder of this review, I will discuss what works, and what doesn't work, from my own perspective, and offer some thoughts on how the 'God Problem' may be resolved.
The initial back-story of the game is an interesting exercise in speculative fiction (see the opening hook for the game, described above). How might we, as explorers of other planets, understand the archaeological evidence of non-humans? How might xenoarchaeology actually work? (Another way of approaching this problem here on Earth, of understanding non-human material culture, is similarly being approached by Andrew Reinhard's 'No Man's Sky Archaeological Survey Project', see archaeogaming.com). The solution used by our future archaeologists is a combination of artificial intelligence and petri dish (that is, an agent-modelling approach that in archaeology might most famously be associated with the work of Dean et al. 2000). Biological robots are created, and we attempt to relive the past through the lives of these created beings. Over and over we send them out, trying to determine the optimal balance of paths, of abilities, of procreation, so that our outcome matches the archaeological evidence (presumably, the problem of equifinality has been solved in the future; cf. Graham and Weingart 2015).
The levels are challenging, if a bit repetitive. I certainly failed many times trying to shepherd my Lovans across the steppe, putting too much emphasis on reproductive fitness versus mobility. It was never clear to me exactly how all the variables and evolution points and boosts worked together (or at cross-purposes), even after 20 missions, and so one wonders if the larger archaeological point connecting the model to the source (human evolution on Earth) might be lost for non-archaeological casual gamers. The way-finding algorithm seems to run into trouble often, and I had to make sure to break my routes into many small sections (by repeatedly tapping the screen) in order not to strand my Lovans in arid deserts or unproductive rocky terrain.
In the interstitial scenes (which are gorgeously painted) it becomes apparent that the Lovan robots have achieved some sort of self-awareness; as the game progresses, they seem to be aware also of the archaeologists/Gods and begin to wonder why they are doing what they are doing. Religion is seen to emerge in the story (but not in the dynamics of the game's ABM itself). While interesting from a science-fiction point of view, in the context of the desired goals of this game to communicate archaeological method and theory, this is distracting. The artwork of the interstitial scenes is curious on gender grounds too, for instance, where the 'male Lovans' appear to do all the hunting, and the 'female Lovans' are passive (Figure 3).
These interstitial scenes are meant to move the game's story forward (that is, the story of the artificial Lovans' growing sense of the numinous). It perhaps would have been more effective for the game's archaeological ambitions if these scenes had focused on the game's archaeologists working through the implications of what the on-the-ground simulation had just achieved. Instead, they reinforce the 'God' problem here in a couple of ways. Firstly, they change the focus of the game from understanding the evolution of the species and its culture to the player's agency in giving life and direction to these biological robots. Secondly, it reinforces an unfortunate by-product of the merging of ABM with playful engagement.
By this I mean it seems to me that the player's agency in the game is directed at the wrong level. In ABM, the investigator sets up an environment, specifies a set of initial conditions, and then leaves the simulation to run on its own without further interference (if there is religion implied in that, it is of the kind where the Gods make the universe and then leave it alone). By allowing the player to interfere with the robots (the agents in an ABM) while the simulation is taking place, the results are not emergent; we have not learned anything new given this encoded description of past phenomena. The result here is intelligent design, where a deus ex machina intervenes in the affairs of Lovan-kind.
One possible way out of this problem is to embody the player within an individual Lovan, to have the player experience the consequences of the world set-up first-hand. It is the Avatar solution. One of the most common agent-based modelling environments, Netlogo (Wilensky 1999) uses just such a solution in its curricular extension, Hubnet (Wilensky and Stroup 1999). This extension allows students armed with scientific calculators to assume the role of a single agent within an agent-based model. The student thus becomes subject to the understanding of the world encoded in the model's procedures. The student is not able to effect change on a global level, but rather must experience the world first-hand. This doesn't mean that they have a first-person point of view familiar in video games such as Doom. It's not about the graphics, but about the potential interactions. It's a participatory simulation (cf. for instance Guyot and Honiden 2006). Evolving Planet frequently features missions that involve the interaction of different species. A scenario that allowed the player to experience that first-hand could be very powerful indeed.
One could imagine the tension between the archaeologists and the game designers over this issue, but it's a critically important issue to address, and this team is to be commended for tackling it. Copplestone and Reinhard debate across several blog posts the (ideal) relationship of game developers with archaeologists (see Copplestone's first and second posts, followed by Reinhard's response; then enjoy Copplestone's prototype game that puts this conversation into action here). Indeed, if I were to incorporate Evolving Planet into my teaching, I would frame the exercise as a kind of debate on doing digital public archaeology through gaming. I would have the students explore the Artificial Anasazi paper and reimplementation as a Netlogo model by Janssen (2009). I would ask the students to craft on paper a mechanism for making that model game-like or playable. This would be in conjunction with reading the Copplestone-Reinhard debate. At that point, the students ought to play Evolving Planet in teams, where the player talks aloud about what they are doing or thinking, and the other students keep a critical game diary of that play through. (It would also be valuable to have the students repeat that exercise with non-archaeologists.) The summative assessment would be for the students to craft a response to the game that places the events from the game within the larger archaeological literature: in what ways does the game foster deeper archaeological learning? Does it do a good job, and how is good defined in this context? Is (should?) a playable ABM the route forward for writing scholarship about the past in this immersive medium? What are the strengths, where are the dangers? How would they fix it? A subsequent exercise might be to put those suggestions into practice via Netlogo.
That a casual game meant for a smartphone can generate this kind of discussion in the first place is a testament to the ambition, the vision, and the skill of the team assembled. Games as a medium offer great potential for communicating scholarship surrounding the past (cf. Tara Copplestone's ongoing research, and Reinhard's 2015 review of Never Alone), and for folding potential audiences into the processes of research. Few examples exist, and the tightrope between scientific integrity and the conventions of the medium is a difficult and dangerous one to walk. That Evolving Planet does as well as it does makes this game a milestone in this evolution.
Copplestone, T. Gamingarchaeo: The study of games about archaeology. The creation of games for archaeology. http://www.taracopplestone.co.uk/
Dean, J.S., Gumerman G.J., Epstein J.M., Axtell, R.L., Swedlund, A.C., Parker, M.T. and McCarroll, S. 2000 'Understanding Anasazi culture change through agent-based modeling' in T. Kohler and G. Gumerman (eds) Dynamics in Human and Primate Societies: Agent-Based Modeling of Social and Spatial Processes, Santa Fe Institute, New York & London: Oxford University Press. 179â€“205.
Graham, S. and Weingart, S. 2015 'The equifinality of archaeological networks: an agent-based exploratory lab approach', Journal of Archaeological Method and Theory 22(1), 248-74. http://dx.doi.org/10.1007/s10816-014-9230-y
Guyot, P. and Honiden, S. 2006 'Agent-based participatory simulations: merging multi-agent systems and role-playing games', Journal of Artificial Societies and Social Simulation 9.4 http://jasss.soc.surrey.ac.uk/9/4/8.html
Janssen, M.A. 2009 'Understanding artificial Anasazi', Journal of Artificial Societies and Social Simulation 12.4.13 http://jasss.soc.surrey.ac.uk/12/4/13.html
Lake, M., 2014 'Trends in archaeological simulation', Journal of Archaeological Method and Theory 21(2), 258-87. http://dx.doi.org/10.1007/s10816-013-9188-1
Reinhard, A. Archaeogaming http://archaeogaming.com
Reinhard, A. 2015 'Review of Never Alone [game]', Internet Archaeology 38. http://dx.doi.org/10.11141/ia.38.6
Sinclair, S. and Rockwell, G. 2016 Hermeneutica: Computer Assisted Interpretation in the Humanities, Cambridge MA: MIT Press.
Tesfatsion, L. 2016 Empirical Validation and Verfication of Agent-Based Models. http://www2.econ.iastate.edu/tesfatsi/empvalid.htm
Wilensky, U. 1999 NetLogo. http://ccl.northwestern.edu/netlogo/ Evanston, IL: Center for Connected Learning and Computer-Based Modeling, Northwestern University.
Wilensky, U. and Stroup, W. 1999 HubNet. http://ccl.northwestern.edu/netlogo/hubnet.html Evanston, IL: Center for Connected Learning and Computer-Based Modeling, Northwestern University.
Wurzer, G., Kowarik, K. and Reschreiter, H. (eds), 2014. Agent-Based Modeling and Simulation in Archaeology (Advances in Geographic Information Science). Berlin: Springer-Verlag.