Postmortem: Guerilla Tea’s Play to Cure: Genes in Space

When we founded Guerilla Tea in 2011, one of our main interests as a studio was to find ways of using game design and technology in other areas. I felt like it was our unique angle, but wasn’t something we’d had a chance to explore too much in reality. I suppose this changed when, through industry connections, we were invited to pitch to Cancer Research UK for their latest project.

Using Video Games in New Ways

Cancer Research UK had plans for a citizen science project which stemmed from a massive backlog of genetic micro-array data. This data was collected from breast cancer patients over years and needed to be analysed. Successful analysis of this data could reveal specific genetic patterns, ultimately allowing for potential new treatments to be discovered.

The above graph is a sample of micro-array data. The pink areas are composed of thousands of individual dots. The analysis involves identifying the areas where the dots are at their most dense. In the above case it is (more or less) a strip moving along the X-axis of the graph, although each sample does vary in terms of general pattern and density. Notice above that there are highlighted sections where the dense area deviates sharply from the main strip. These are the interesting areas, which are considered anomalous, and could provide key clues to cancer treatment.

The overall notion of identifying the most dense areas, and in turn the anomalous areas is the ultimate goal of analysing each sample. There were two outstanding up front issues with this. The first being that the data really needed a human eye to analyse. While a computer could certainly do the task to a high degree of accuracy, the key was in the discrepancies, and certain lack of ‘black and white’ accuracy that would only come about from a human interpretation. Additionally, while computers are capable of identifying patterns, people are considerably better at identifying incomplete or partial patterns.

The second issue was purely down to man hours. It would take scientists alone years to get through all the data samples.

It was here that Cancer Research UK hit upon the ideal solution, to outsource the analysis of the data to the public. Something made possible by the fact that the analysis itself didn’t require any specialised knowledge of genetics; it’s simply basic pattern recognition.

Their new approach, expanding on successful crowd-sourcing projects such as Cell Slider, was to use the video game medium. We were invited to pitch for the project, where we were challenged to come up with a game concept, which could turn this data analysis into an entertaining mobile game.

Due to the amount of data the game would have to pull data samples from a server, allow players to analyse it by playing the game, and then submit the completed analysis back to the server, for collection and examination at a later date.

What followed was one intense week of research into the data and data analysis, where we experimented with a number of different game concepts, looking into the most effective ways of allowing players to analyse the data, along with a backstory to the game. We looked into the backend tech to support this, and also created a development plan. This culminated in a tough 24 hour shift in the office putting on the finishing touches, and a few days later we travelled to London to pitch directly to Cancer Research UK.

It was a tense week I’ll admit, but an understanding of the genetic data from Alex, our lead programmer, really gave us a terrific boost. Nevertheless the competition would be tough…

We waited on and a week later received a phone call. It was good news. Guerilla Tea was the chosen developer. Time to get started.

What Went Right

1. Understanding the client’s needs and a focus on the research data

From the very start of the project, we knew that it was going to involve some unorthodox game design. Cancer Research UK was very clear in stressing that the absolute most important aspect of the project was to provide the best possible data analysis.

Of course creating a game that was as fun as possible was essential, but it was inevitable early on that we would need to make some compromises in terms of the overall gameplay in order to keep that data analysis valid.

Play to Cure: Genes in Space is a space themed action adventure game, where the player guides a spaceship through an area of space collecting a valuable, fictional substance dubbed Element Alpha (inspired by Neutronium).

We have taken the genetic data graph (shown above) and imagined it as the Element Alpha deposits, distributed around an area of outer space. Players control a spacecraft, guiding it over the Element Alpha to collect it. In a sense, the spacecraft is flying ‘over’ a graph that has been rotated 90 degrees, and we are recording the position of the spacecraft in order to analyse the data.

This hands-on gameplay is one way in which the player takes part in the analysis, but there is more to it. To ensure the best possible analysis we decided to go for a type of modular gameplay design. ie. Separating the gameplay into distinct sections. Prior to the player flying through space to collect element alpha, they would have to plan their route. This was stylised as a scan of the element alpha deposits in space, but is in fact the genetic micro-array data. The scan mimics the pattern of element alpha which will then appear in the flying section.

The player taps to place the green markers onto the graph at the locations which they see as the most dense. The action of placing markers is almost identical to the process scientists would undertake in analysing the data in a lab, but in the case of the game it had a knock on effect. The player must then try to fly through space matching their flight path to the plotted route. In turn, this would yield more element alpha allowing for more credits used to purchase upgrades, and a quicker increase in rank and score, giving us our gameplay motivator.

What we are doing underneath the surface is combining three different types of analysis:

Combining the three different methods gave us the most effective data analysis, bearing in mind it is the human error and discrepancies that matter.

This separation also allowed us to in a sense experiment and track which type of analysis was most effective, and gave us some leeway should players not fly effectively through space.

Essentially, these distinct but related gameplay modules were planned early and certainly worked from the research angle.

 2. The Overall Gameplay Loop

While I was doing some research for the project, I came across a fascinating talk from Torsten Reil of NaturalMotion. It was a talk on how to create an iPhone game which works within that market and ideally become a hit title.

The main item I took from the talk was certainly the fact that players play mobile games differently to console or PC. The gameplay needs to be delivered in very quick bursts, letting players ‘snack’ on the game. Basically after a few minutes, a section of gameplay needs to come to a meaningful conclusion.

I then played NaturalMotion’s and Boss Alien hit title, CSR Racing, where the majority of the time is spent in a passive car customisation area. This isn’t forcing you to do anything, and when you do need to engage in the game’s drag races, it gives you these quick, intense bursts.

I tried to apply this idea as best as I could to Play to Cure: Genes in Space. The fact that we’d built the game up individual gameplay sections helped a lot. The player would spend the majority of their time in the spacecraft upgrading and customisation area. They could view the vessel, change the colour, buy new parts, etc. When they are ready they are able to jump to the route mapping followed by space flight, with the important point being that these engaging activities are not dragged out for too long. Afterwards the player is returned to the upgrading and customisation area having successfully completed one data analysis, with no pressure to instantly repeat or follow up on it. This way I felt we would ultimately achieve better results.

Notice that there is an asteroid field included in the diagram above. This was again the result of our modular based approach. We felt that the route mapping and flight sections alone weren’t interesting enough from a gameplay angle, but we were hit by the problem of being unable to tamper with them design-wise to not skew the research data.

Instead we came upon the solution of adding a separate section directly after completing element alpha collection, where the spacecraft enters and must survive an asteroid field. It takes the form of a basic shooter, and only appears a certain percentage of rounds played. This ultimately goes towards a risk reward mechanism, whereby the player can collect greater quantities of element alpha over multiple flights, but also risks losing it should their spacecraft be destroyed in the asteroid field.

3. Project Scope

Although we drew on NaturalMotion titles for design inspiration, it was very clear that we were never going be able to create a game of their level (or anything close) in the 5 months development. This isn’t something we were new to really. A poor sense of scope is something that burned us while at University, and it’s generally the main failing of so many student teams. And since university is a time still relatively fresh in our memories, we were particularly conscious of it.

However, another aspect of Torsten Reil’s talk discussed the need for a high level of polish on games. It’s been drilled into us to set a modest overall vision, and then absolutely over deliver. Create a small, but slick and smooth game.

For example we worked hard to get the spacecraft flight working to the point where it just felt right. We found a sweet spot with the spacecraft speed which was the halfway house between being too slow becoming mundane, and too quick which could damage any analysis. Combining these statistical modifications with the look and feel; we wanted to hit upon that slightly mysterious ‘lost’ feeling of being in an anonymous part of the galaxy light years from any civilisation. We found that small changes like the position of the planets in the background (slowly moving past the spacecraft) helped to give the space flight some context.

The focus on this was more important, and in a sense attainable for us, rather than trying to compete with other titles in terms of sheer volume of content, such as truck loads of customisation items.

4. Audio

An aspect to the project which was the smoothest it could have been was the audio development. Euphonious Ltd are fantastic. Simple as that.

It was something that we were worried about before the project began, and we had planned to outsource the audio. Euphonious worked closely alongside the development of the game, working quickly and providing a lot of feedback and advice to us throughout. Exactly what we were looking for, since we were reliant on their expertise.

The spaceflight sections come to life with the electronic soundtrack and the tension of the fast paced music perfectly fits our intense asteroid fields.

The end product has been praised for its audio and to say we are grateful would be an understatement.

5. The Publicity

This project had garnered a lot of interest from the media even from the initial announcement which went out in summer 2013, so our hopes were high for a popular launch, but it surpassed that.

We attended the launch event in London on the 4^th February, where Dara O’Briain helped to coordinate things. It was actually quite moving with talks from cancer survivors, and it genuinely felt incredibly worthwhile, and that we really were helping to take the industry in new directions.

Twitter came alive and ‘Genes In Space’ was trending that day. Our own social media picked up a lot largely thanks to a retweet from Dara.

The most significant publicity piece we got was about a month after release. It was a full feature on The One Show (a major topical magazine-style show) where some of the stats were revealed, including the fact that in just over a month, the game has already saved researchers around 6 months of work, getting through 1.5 million samples. Afterwards the infographic below was shared extensively on social media, and a second major marketing push for the game began.

What Went Wrong

1. In-House Processes Not Optimized

Whenever I read articles regarding game development I’ll invariably find anecdotes from the developers about the stresses and strains of getting the game done including accounts of how long the hours are, living off takeaway pizza, etc. We’re going to be no different with the development of Play to Cure: Genes in Space which caused more than our fair share of grey hairs/hair loss/weight gain.

I suppose it’s par for the course and anything worth doing isn’t going to be easy, but we became frustrated because in the end a lot of technical problems could have been avoided.

Our version control unintentionally re-introduced bugs which were previously fixed. In turn this lead to a tumultuous internal QA process, where it became very difficult to keep track of bugs in the game. We found ourselves finding issues constantly re-appearing, and fixing one area of the game would tear a hole in another part.

We got there in the end, but there was a lot more worry regarding the stability of the game than we would have liked.

As a result of that, we’ve spent considerable resources since the project completed to optimize our asset pipeline, version control methods and build processes to avoid time consuming mistakes like this in our future projects.

2. Not Enough External Focus Testing

During the development process, the thing we shied away from was too much external focus testing. We did bring in testers to provide feedback, but in hindsight we would give this much, much more attention.

I again feel like we got there in the end, and in some ways the restraints we had imposed by data analysis prevented very valuable feedback from being properly considered. Nevertheless, the game was created and re-iterated mostly from pure ‘feel’ and opinion from within the company and Cancer Research UK.

I found myself getting too close to the game at some point, and my judgement as the designer became cloudy to say the least. This came to the fore in designing the Asteroid Field sections, where too much time was spent without a solid direction, and worse was the fact that the section itself was simply too difficult. I had played so much that it really did seem too easy to me. After the game had been out for a week, the amount of people complaining that it was impossible was overwhelming. A bit of a game breaker to say the least.

We were at least made aware of the issue and fixed it with an update.

3. Polishing Took Longer Than Expected

The development of the game actually included a short extension, with the initial scheduled released being in October 2013.

We simply underestimated the layers of polish that we’d always planned on including in the initial vision. This misjudgement certainly wasn’t helped by some of the technical issues we’d had in terms of our in-house processes.

I remember Alex making an analogy at one point (probably closer to release than we’re comfortable with) stating that the game was like a house that was beautiful when standing admiring it from the street. But after a week living in it, you realised it was always cold, the washing machine leaked and the toilet just wouldn’t flush properly…

4. Our work was judged as a game, rather than as a citizen science project

This is a strange point, but something that did surface after the game was released. It was something that we almost joked about during the development.

The game treads the line between a video game, and a citizen science project. It was expectedly judged by gamers as a ‘pure’ video game, and unfortunately it wasn’t possible to completely ‘disguise’ the data analysis within the gameplay; it was always going to seem a little facile or dry in comparison to other games with a free reign on gameplay design.

5. We Weren’t Insistent Enough on Press Coverage…

We were hugely hopeful for all the media coverage we were going to be getting upon release considering the nature of the project and the prior interest it had picked up.

In general the coverage for the game was excellent and we got a good deal of promotion as I’ve mentioned previously. This doesn’t however change the fact that there were several painful mistakes and omissions out there, which could have been limited by ourselves. The mistake we made was to sit back a little too much and assume things would be reported accurately, particularly when we did have opportunities to control the game’s publicity to some degree.

In hindsight, we should have put a lot more effort into having Guerilla Tea promoted as the developer of the game.

Not all publications needed to cover the story of how the game came to be, instead focussing more on the nature of the end result. This is fair enough and only one angle, however if you Google the project and you’ll find stories of it being fully developed in the course of single weekend game jam, organised by Cancer Research UK. Our involvement was not mentioned at all.

I’m going to take the chance now to clear things up. Cancer Research UK held a game jam last summer where a number of prototypes were developed. Guerilla Tea took a look at these for reference, which did give us some great initial inspiration. We then built Play to Cure: Genes in Space in its entirety over the 5 months, from establishing the project vision, through developing the backend data analysis, network functionality and the complete gameplay.

Conclusion

All in all, it’s been a monumental project for us as a small independent studio. It was something we desperately wanted to be part of, and it fitted the bill for our company as we’ve been involved with several smaller altruistic projects in the past.

Image Courtesy of STV Dundee

In terms of the success of the project as a whole, it’s just the beginning. Since the release our day-to-day work has gotten a whole lot busier. Work on the project continues and with some channelled effort we hope to highlight the potential of the video game medium for use in not only research but possibly other areas of study.

Play to Cure: Genes in Space was ultimately designed to speed up some seriously time consuming work, and over the coming months we believe the full effects of this will become apparent.

Data Box:

Developer: Guerilla Tea

Publisher: Cancer Research UK

Release Date: 4th February 2014

Platforms: iOS, Android

Number of Developers: 7

Length of Development: 5 months

Development Tools: Unity, Mercurial, Maya, Twisted