Learning from Portal
4th January, 2008I finished Portal last night. I haven’t enjoyed a game so much in ages: fantastic script, funny and chilling in equal measure, puzzles that made me think but not reach for a walkthrough, and the ability to bend space at whim. If your PC’s up to it, get hold of it and give it a go.
I don’t really mean to talk about how good the game is here, though I want to (”subject home town here”! companion cube! that final song! cake!). What’s really interesting is that once each section of the game is complete, you can play it again, only this time with commentary from the developers. And it’s this that should be interesting to anyone who’s been trying to work out how to make educational games that feel as well-designed as the top commercial games.
Although there’s some discussion of the programming difficulties involved in making a portal mechanic work, most of the commentary is on the thinking behind the way the levels are structured. It’s a real insight into the lessons good developers are prepared to learn from testing and play-testers: I came away with the impression that the way the final game turned out was almost as much due to the players who tested it in development as it was to the designers. Each level had a clear learning goal, and the developers were prepared to make whatever changes were necessary to the level design in order to make sure players had achieved it – because if they hadn’t got a clear picture of their character’s abilities, they wouldn’t make it through the later levels. High stakes assessment indeed: when you’re trying to shoot holes in space so your momentum can carry you over the talking machine-gun droids, you find out quickly whether or not you took away everything you might have from the early levels.
Such an approach was necessary, of course, because the gameplay introduces radically new concepts to players: it’s arguable that had they been designing a more familiar style of gameplay the tutorial approach would have been overkill, perhaps even frustrating for players used to that style. It’s possible, too, that the designers aproach to players’ learning is too structured to appeal to many educators. And so I’m not necessarily arguing that Portal is a great educational game for anything much except learning to play Portal, though I’d love to see some maps for physics teachers.
But as an example of just how much thought goes into level design in the best games, Portal is superb. It’s an illustration that high production values aren’t everything, an argument for thoughtful, player-centric design and a warning never to leave a supercomputer alone with only a cake recipe for company.
Futurelab is always looking for ideas. Find out how you can 
Bob Harrison
Ben Williamson
Dan Sutch
Richard Sandford
Leon Cych
Martin Owen
Sarah Grier
Tim Reader
January 4th, 2008 at 3:29 pm
We used to run games testing sessions with really hardcore players (some of them were 7 years old) in my last job for a range of developers. What fascinated me then was the way players of all ages wanted to experience a “fully-formed” world with believable rules (ie fantasy games still have to be “believable” and “realistic” even if they’re completely unreal in planet earth terms). That meant anything which seemed out of place to the genre rules (with the exception of things that stretched the genre logically) could completely switch players off. (On a personal note, I despaired when Spiderman turned up in Tony Hawk’s Skateboarding, and Dobermann-headed German soldiers with rifles in Medal of Honor 2.)
The smallest of oversights by the developers could also make testers really impatient. “Where are the footprints?” I remember one teenager asking while testing Operation Desert Storm, before adding, “They’re running through sand!”
Years later, what interests me about this is not so much players’ engagement with genre (or perhaps pickiness) that play-testing reveals, but what it might tell us about young people’s capacity to take a system they recognise and to give it a good shake. Can we see school students “play-testing” the 3D CAD designs for their new schools please? Can they test out all the materials and rule systems that are due to come into place with the curriculum review?
I’d say this was far more important than identifying the lack of footprints in pixellated sand, yet people who design sand pixels seem to care a lot more sometimes.
January 4th, 2008 at 4:36 pm
Absolutely. There was a great description in the commentary of the debate the designers had when certain players found ways to circumvent a whole level: in the end, they decided that as the “ninja” way was harder than the “official” route, they should let it alone, and reward players’ skill and exploration.
January 7th, 2008 at 8:31 am
It’ll certainly help with childhood obesity rates, what with the cake being a lie and all.
Seriously though, it is a fantastic game, although as you say, it doesn’t necessarily teach much more than how to succeed in that particular universe. Surely you’ve played the physics-based puzzles in Half-Life 2 as well though?
January 7th, 2008 at 10:36 am
No spoilers! It’s not a lie!
You’re right, HL2 has some great physics potential as well, although the whole picking-things-up-and-putting-them-where-you-want aspect could use some work (I spent much longer than I wanted to piling bricks onto that seesaw). With Portal I was particularly thinking of momentum, and possibly the way you can see round corners (maybe some sort of activity answering the question “where would you need to place the portal so you could see X?”).
But yeah, I think there’s more lessons for devs than pupils in Portal. We’re a long way off being able to teach <subject_name_here> with it…
January 7th, 2008 at 5:58 pm
I heard an interview with one of the Portal developers on the Game Theory podcast a few months ago discussing the difference working on their student prototype project (Narbacular Drop) and their commercial product Portal for Valve.
They said that Valve told them to get a working version of a level up and running as soon as possible and start playtesting immediately. They constantly iterated on the levels developed based on both player feedback, as well as a technology which let them analyse player actions as they happen in the level – where they get stuck, how they behave.
Now they can even get this kind of detail as players play the game through their Steam platform. To get a feel for what I mean, check out http://www.steampowered.com/status/ep2/ep2_stats.php .
January 7th, 2008 at 6:03 pm
You’ve neatly reminded me of this:
http://kotaku.com/gaming/clips/watch-3000-barrels-fall-down-in-crysis-333902.php
Now that’s physics in action!
January 8th, 2008 at 12:15 pm
I am developing a serious game for teaching science so I am interested in the comments about the importance to players of internal consistency in games. ie. it doesn’t matter what the rules are, as long as you stick to them.
I think this shows that we all have a natural desire for the world we’re exploring to obey laws – to follow patterns – and we like to discover what those patterns are. Very encouraging for the prospects for the teaching of scientific laws through simulation.
January 8th, 2008 at 12:59 pm
It’s amazing how quirks of systems, or ‘glitches’ as they’re generally known, quickly become accepted parts of the system itself, almost like in the Matrix movies:
http://en.wikipedia.org/wiki/Glitch
There are loads of YouTube videos dedicated to the demonstration of such glitches.
February 5th, 2008 at 6:18 pm
[...] How can playing an FPS become a informal learning experience, and lead to other learning experiences? Granted, there are those games such as Portal that provide a reasonably new experience (ever heard of Narbacular Drop?) that requires you to learn the rules, which are based at least in a physics that at least is partly grounded in this universe, in order to succeed, but what game doesn’t do that to some degree? [...]