Behind the scenes of Launchball – an interview with Daniel Evans, Frankie Roberto, and Mike Ellis

There is a lot to learn from the Science Museum’s (London) recent success with their Launchball online game.

The project has been enormously successful and recently won ‘best of show’ at SxSW. I conducted an interview with Daniel Evans, Frankie Roberto, and Mike Ellis to explore some of the ideas and processes behind the project.

Launchball was built to support and extend the Launchpad gallery experience. Launchpad is a highly interactive physics exploration space specifically aimed at 8-14 year olds.

How did it start?

Frankie Roberto:

“The key idea came from a realisation that the mission of the gallery was to allow kids to explore and play with real physical phenomena. As it’s nigh-on-impossible to experience real physics through a computer screen, we decided to drop this element and use simulation instead, and to focus instead on the exploration, playing, open-ended nature of the gallery . . . it follows the ethos of the gallery completely, and covers much of the same topic areas too (electricity, light, energy transfer, etc).”

Daniel Evans:

“As Frankie says Launchball is quite closely tied in to the gallery in many ways: the same set of themes are explored through similar types of activity. This isn’t a groovy game vs. a worthy gallery – Launchpad is a riotously popular place itself. We were very lucky that the exhibition team had an extremely thorough understanding of their brand and its meaning and were able to strip it of its gallery-specific contingencies and give us a brief at quite an abstract level. We were therefore able to work with a useful set of ideas (“Launchpad is about learning by doing what you fancy and seeing what happens”) rather than a more facile and literal interpretation that would have left us doomed (“Launchpad has a machine for blowing bubbles therefore its presence on the web must too”).

… we had series of brainstorms over a year or so exploring what Launchpad was and what it could look like online, with the web and gallery teams (including, importantly, the people who manned the old gallery on a day-to-day basis and really knew the audience) and a few gurus from outside the museum (Ben Gammon, Joe Cutting). This was interspersed with some research into audience needs and expectations and resulted in an ideas harvest, followed by an ideas elimination and finessing process.”

How did you keep the focus on making the game engaging, fun and addictive versus a more straightforward approach to making it overtly ‘educational’?

Mike Ellis:

“I had a really interesting meeting with the Head of Content at the Science Museum very early on in the project. I specifically asked her whether she thought the concept that we had just presented was “too fun and not educational enough”. She absolutely backed up our approach, pointing out that people in the “real” Launchpad space were having a great time and that the learning followed on from that rather than the other way round.”

Frankie Roberto:

“I felt strongly that the game should avoid the trap of having a traditional arcade-style gameplay, with a light level of theming and text-based interstitial screens that carry the educational method. 90% of museum games fall into this category. Instead I felt strongly that the playing and experimenting with the game should be an educational learning experience in its own right, and a fun one and enjoyable one.. For me, this idea follows the constructivist theory and follows in the footsteps of things like Lego and the Logo programming environment (which I also played with at school).

That said, when producing any simulation game, from a football manager to a theme park game, there have to be some simplifications from reality in order to make the game playable, and just to be able to physically make the game. In Launchball, for example, there’s no real friction, light can’t be diffracted, batteries never run out, and so on. Adding these in might have added additional learning outcomes, but it’d have made the gameplay more complicated, and not to mention would have made the game more complex to produce.

One of the key decisions we made, which I really insisted on, was a ‘sandbox’ mode where people could just play with all the blocks, with no limits, no set task, no specific learning objective. This I think is critical, as it really allows people to truly play, in an open-ended fashion, and to learn through doing all the way. Additionally, the sandbox allows people to create their own levels, truly opening up the game.”

Daniel Evans:

“It’s an interestingly loaded word, “educational”. In a sense your question is based on the very model that we were determined to avoid – the idea of using a game as a sort of diversionary tactic to buy some goodwill while you cram some facts down the audience’s throats, with the challenge for the developers being getting the right balance between sugar and pill. It’s amazing how persistent this fundamentally rather pessimistic idea is – the number of commentators for example who have seen the factoids at the end of a launchball level and thought “ah, that’ll be the educational bit”, when in fact the educational bit is the bit they’ve just finished.

It was absolutely crucial that the game was the education and the education was the game: the activity and the ideas it was communicating were indivisible. The key assumption behind the whole of Launchpad is that there is no tension between an activity being fun and it being physics-based. Online this is amply evidenced by the fact that almost all commercial games, including those with no educational remit whatsoever, are built around … physics engines. In a sense Launchball is just a giant exercise in drawing attention to this fact.”

What was the role of the external developers?

Daniel Evans:

“. . . it is impossible to overstate the role Preloaded had in the success of Launchball, both in terms of the craft skills of game design (hats off to Henry Cooke, Jon Mallinson and Phil Stuart in particular – three exceptionally talented individuals) and in terms of their endless willingness to rise to each new challenge (and we threw them quite a few) to deliver something truly excellent rather than just acceptably good.

User testing was the other major factor – there was a huge amount of testing by our in-house team on real members of the target audience at all stages of development. If you’re clear about your objectives, work with talented and creative sources of good ideas and then subject those ideas to really searching evidence-based quality control (and the developers engage positively with the findings, which Preloaded to their credit always did), then you stand a pretty good chance of getting a decent product.”

Digg played an early role in the viral promotion of the site, generating an enormous rush of visits to the site, putting strain on the servers.

Mike Ellis:

“I had actually left the museum at the point when I punted it to Digg :-) I got some fairly sharp words from the IT dept telling me I should have told them but to be honest what would they have done, gone and bought a server farm if I had let them know?”

Frankie Roberto:

“To be honest though, if it wasn’t Mike that had submitted it to Digg, someone else would have, and I think it would have had the same effect. Ultimately, it was the users of Digg who decided that the game was good enough for it to have made the Digg front page. Digg hasn’t actually been the biggest referrer to the game, that accolade falls to onemorelevel.com, followed by 2 Norwegian sites, then stumbleupon.com, then Digg. However Digg was certainly the most sudden.”

Daniel Evans:

“It wasn’t so much the traffic, it was when it hit: we hadn’t tested it under heavy load at that point. A few weeks later, bottlenecks identified and fixed, the game was coping comfortably with several times the traffic that Digg drove. In our final stress testing we simulated 100 times the load that brought the site down and although the servers were obviously straining they didn’t collapse. We were a bit naive and were caught slightly unawares by the instant attention. Although we were confident the game was pretty good, we underestimated the full extent and immediacy of its viral strength. Still, a nice problem to have, on balance.

… [now] we’ve had over 1.5 million visitors and about 100,000 people saving levels. 25% of Launchball visitors (most of whom are new and there for the game) go on to visit other parts of the Science Museum website … the figures for duration are a bit unreliable as it’s all on one page and a lot of people never go to any other page, but it’s obviously a lot longer than the website average.”

Frankie Roberto:

“It’s now the most popular page on our website!”

How much involvement has the museum had in engaging with the communities that have sprung up around the game? How is the education sector responding to it?

Daniel Evans:

“We’ve had ‘design a level’ competitions, there’s a facebook group and we’re continually getting feedback. There are over 25,000 pages on the web discussing Launchball though – so we can’t keep track of them all. Launchball discussions spring up in lots of surprising places too – there was a fansite for one of the other games at SXSW that had a thread complaining that Launchball won, that swiftly turned into a discussion about how you complete ‘The Volcano’. We’ve had lots of positive feedback from teachers too, including some unexpected stuff – for example some teachers are finding it very useful for teaching English as a Foreign Language.”

What are the lessons you have learnt from the experience?

Daniel Evans:

1. The quality of the brief is key: being clear what’s an end and not up for negotiation and what’s a means, with creative interpretation and challenge welcome, is vital in terms of both briefing and being briefed.
2. Argue your point strongly, and work with people who argue back just as strongly.
3. There are few challenges that can’t be overcome with a really first-rate Project Manager.
4. It’s fun. Enjoy it.
5. On the web you have no say in whether you’re doing things quietly or not.

Frankie Roberto:

1. Better to have one big, high-quality, immersive game than lots of smaller ones.
2. Trust in constructivism.
3. Shoot high.
4. Work with the best.
5. Build a multi-disciplinary team, and work together.

Mike Ellis:

1. Trusting the web team is absolutely vital in making a success of web products. The ‘traditional’ model of curator/content type coming up with web idea isn’t usually as well focussed.
2. Give freedom and time to ideas and let them flourish.
3. Use storyboards and wireframes at every stage of the process to help explain what exactly it is that you’re on about.
4. Be prepared for the Digg Effect (not even now entirely sure how?)
5. Go to creative agencies with a good idea of what you want to acheive. Brief it well and have lots of user testing research on board too.
6. Fight hard for realistic budgets.

Thank you to Mike Ellis, Frankie Roberto and Daniel Evans at the Science Museum for their time and openness in sharing more about the experience with Fresh & New.

Mike has blogged about the initial stages of the project with some early wireframes.