It looks as if WarTron will be the next incarnation of “The Game.” If you are not familiar with The Game – no, not that movie with Michael Douglas – it is, in brief, a weekend-long road trip in which you solve puzzles to reach particular destinations. “Race” is probably a little more apt than “trip” since it is a competition against other teams trying to do the same.
My description actually gives it no justice and I highly suggest you check out the Wired article about the WHO game, which was the most recent instance of The Game. Inspired by Dr. Horrible, it revolved around superheros and super-villains and, in fact, contained several videos starring Neil Patrick Harris along the way.
Last weekend, WarTron had an applicant competition in the form of a game. They only have twenty slots for WarTron and more than that many teams want in, so there is a puzzle-off. The first 20 teams to solve their way through the application are accepted. Our team was a cross-pollination of two different DASH PDX teams and an Alternate Reality Gamer friend in the Bay Area. The application puzzle trail was meant to last a few hours, but due to some technical issues was more like 6 hours. We ended in (d’oh) 24th place.
[Edit: we’re currently shown in 23rd place, but that is only because one of the top 20 dropped out after the fact. Technically, we were and are 24th.]
[Edit 2: it turns out that they are letting folks in based on headcount, not strictly by a number of teams. That 20 team limit was a soft limit and they invited us into the game because the first 20 had fewer participants than the theoretical team maximums. So we’re in!]
WarTron, as you might guess is a contraction of War Games and Tron. The backstory, qualifying puzzle trail, and general theme of everything pulls elements from these movies.
GotoVision is a game company, bringing a new game to market in August: WarTron. We’re applying for a “Weekend with WarTron” challenge, which sounds like a happy-fun-times party to beta test the new video game. I thought I remembered something about the mysterious disappearance of the founder, though I can’t find it now. Of course, the company had ties in with the DoD in the past. There’s a rogue military AI (BIGMAC, not WOPR) that was accidentally restored from backup and probably threatens the end of the world, and so on.
Cliché? Perhaps – but one with rich details. There is a short story with a little background info. You start by taking a test (“Really? This is it? That can’t be it. It’s too easy.”) and get a peek at the AI in the process. You sort of go off the rails at that point and start interacting with the AI. At one point, you actually log in to a text console and interact with with it directly.
I felt that one of the great moments was when we were all gathered around around one computer. It really felt like I was immersed in the move War Games, with all of us gathered around a single computer, blurting out suggestions. “Type in ‘list games’.” “Yeah, now try tic-tac-toe.” “No, run chess.” It’s the kind of thing that I, perhaps, tried to emulate as a kid writing Atari 400 programs in BASIC, but if I wrote the program I also know all of the back doors, so it was never as fun or realistic.
I do not think that I am going to get into the specifics of individual puzzles. It is actually a pretty good experience with some great eureka moments. I understand that they left everything running in case other teams that have not yet made it to the leaderboard still want to play through. (There’s no indication how many signed up vs how many completed.) I can only hope that they find a way to open it up for others to experience after the fact, so I’d rather not spoil anything specific.
Despite the warm fuzzies and fun times, there was one particularly aggravating part of the puzzle trail. The truly cool interaction of “hack into the computer” was also its Achilles heel. You see, puzzles typically have a self-reinforcing win state. You know you have solved a particular puzzle when you reach it. The online chunk had some good interactivity, but part of the interaction was exploration: discovering how to interact with the system and then probing that interaction to find and solve the actual puzzle. It turns out that this interaction was with an Eliza-like bot, but we didn’t actually know that. The point at which we launched the command to start “talking” to the bot was the same point at which everything crashed under server load. The folks at Game Control attempted to fall back to using email to accept and respond to commands. Having launched it but not yet interacted with it, we hadn’t quite reached the point where we knew what to expect. Because the first email seemed somewhat conversational, we initially tried some long-form communication in English, expecting to talk to “an AI” (or a human playing one), not an Eliza-bot. I feel that the folks that hit this point online had the benefit of interactivity and quick iteration to discover how to interact, then could use the slower email channel to refine the results, whereas we floundered around in email a bit. We really spun our wheels a lot on this interaction. When the system came back online an hour or two later, we got through the bot interaction fairly quickly.
My background is not in puzzle hunts, but more from the world of Alternate Reality Games (ARGs). I understand that live events, especially ones that utilize server resources beyond HTTP (which is typically well-refined and hardened against load issues) are difficult. As comparison, the Matrix-inspired ARG that came out a little after the first movie, tried similar “live hacking.” It involved an FTP server with hidden but readable (if you knew the correct name) files. Puzzles led to filenames, and you’d attempt to log in and retrieve a file by name. It was great stuff, but it collapsed horribly under the load of all the Matrix fanboys. (And as an aside, had an information disclosure vulnerability – I accidentally outed the puppetmasters behind that game when an “ls -lad” showed the directory itself plus its owner.) Live events, even online (and maybe even especially online, since you can only have a finite number of people at a physical event, but many, many orders of magnitude more online), are hard. Still, I do have to question the decision to use ShellInABox to provide the interactive “terminal.” This looks like a really cool app for personal use, but their issue tracker shows a longstanding bug that 10–15 users crash the server. WarTron is capped at 20, meaning they would probably expect at least 20, so I think some quick vetting by skimming the bug list should have turned this up.
Still, the interactive shell quasi-terminal was a highly compelling leap of puzzles from the fantasy world into the real world, despite any flaws and technical problems. I think they were breaking new ground here, given that ShellInABox is a real web-based terminal into a real command-line application running on a Linux server. Just off the top of my head, I’m not sure how I’d do it differently if I was constrained to a web browser. Though, expanding the parameters, I might just use a telnet server on a port above 1024, with limited permission in a chroot jail, to run the command-line app. But that’s me with 20/20 hindsight. What’s done is done.
I’m really grateful to the Game Control folks (I know for sure that Curtis is behind this, but this thank you includes all of you people for which I don’t have names). I found it fun, creative, and a great little rabbit-hole to pull you into the story and backstory of the game. Despite not getting in to WarTron, we had a great afternoon of coffee, friends, and immersive puzzling (And I got to taste cactus tacos for the first time and discovered who on my team hadn’t seen War Games and who hadn’t seen Tron.) It also spun off a few potential bits of research and side projects (more on those later — tomorrow I’ll share what I have learned about Unix dictionary files).
3 thoughts on “How I failed the WarTron application and the fun I had in so doing [Updated x 2]”