PO(r)TA(l)T(w)O

Valve Software advertised their release of Portal 2 using an Alternate Reality Game, or ARG. A series of puzzles led to a game that encouraged players to play a set of indie games in order to release the game early. The players participated, and Portal 2 was released 10 hours early.

A lot of people are upset about this.

At first I was really confused about how angry people were acting, even accounting for the Greater Internet Fuckwad Theory. Valve had put together a cool set of puzzles, offered a bunch of indie games for cheap, and then actually gave players a real-world reward for playing. However, I’ve realized that the displeasure the ARG created is due to a classic problem in game design: miscommunication leading to false expectations.

What Happened

My main source for this post is the Investigation History on the ARG players’ wiki. On the morning of April 15th, players were given a page which implied a potential early release for Portal 2. Implicitly, Valve (in the role of Game Master) was telling ARG players, “Do well at this game, and we will give you a reward of an early release.” Valve clearly had a system set up for this, and a set of rules that governed when release would happen based on the number of players and their behavior. This set of rules was soon reverse-engineered by the players, who did their best to play optimally.

In the end, their efforts resulted in a release 10 hours early. The game was scheduled to be released on the 19th at 7 AM Seattle time. This was about 94 hours after the “early release” phase began. In the end, the ARG players shaved about 10% off of the remaining time until release. This wasn’t at all what many people expected. When presented with a screen promising early release, what they assumed was that they would get to play Portal 2 within a day or two.

What Went Wrong

In the end, players got to play a day early, assuming a normal work schedule (EDIT: and assuming they lived in the US). But they expected it much earlier, it seems. There were two variables which Valve didn’t manage properly: How early people considered “early,” and how effectively people contributed to the ARG by playing the associated indie games. It’s clear to me that Valve underestimated the first, and overestimated the second. When presented on Friday with the offer of an early release, the players envisioned getting it before the end of the weekend. At the same time, they did not contribute to the ARG’s metagame fast enough to make that happen.

By Saturday, the people at Valve behind the ARG had probably realized their mistake. They’d calibrated the game to a higher level of participation than they expected, and if they changed the rules to be more generous people would notice and likely call foul. Note that the late-game “potato countdown” clock acceleration began at 10:30 AM Seattle time on Monday: just around the time that a hurried Monday-morning code fix would be finished and deployed.

What Might Have Been

How could people’s grumpiness have been better avoided? Valve could have begun the ARG earlier, which would have resulted in an earlier release. They could have more clearly communicated the range of possible release dates: “Release Portal 2 up to a day early!” rather than the vaguer “Release Portal 2 early!” Finally, they could have had less confidence in their fans, and adjusted the initial game rules so that less contribution was necessary to accelerate the game launch.

Still, the game effectively got released a day early. People shouldn’t complain too much.

bookmark bookmark bookmark bookmark bookmark bookmark

17 thoughts on “PO(r)TA(l)T(w)O

  1. The first hint was the bathysphere puzzle sent by Gabe to various news sites:

    http://valvearg.com/wiki/Emails_from_Gabe_Newell

    The hidden message was “4/19/2011_7AM=4/15/2011_9AM”. What would you derive from -that-?

    Also, when each game got finished during the Glados@Home event, the clock got pulled back a bit. But that amount was not fixed or based on level of effort – when Killing Floor was concluded, the clock was pulled back -five- minutes. Audiosurf had concluded just a few minutes before.

    My theory for the underlying code is this: For each hour the 96-hour countdown ran, 5 minutes were added to a buffer. Whenever a task got completed, the buffer was emptied into the countdown. This ensured everyone “felt” they were doing an effort, but the countdown would -never- be pulled back more than 10 hours, even if all tasks were hacked to complete immediately.

    1. The bathysphere message appears to list the scheduled (non-early) release date and the time that the final-phase metagame was scheduled to go live.

      To give Valve the benefit of the doubt regarding the reward for game completion, it could be that the rate of “calculation” was solely dependent on the number of concurrent users weighted by game sales, as implied by Valve and discussed here. I’m guessing that you’re right that completing a game simply triggered a timer recalculation, especially since the projected time only changed when a game was completed. However, it could be that that recalculation just dumped all “distributed computation” work into the main timer.

      Note that in the 16 hours from TWEOTW completion to 123KI completion, the timer advanced 40.5 minutes, while in the 9 hours from 123KI to AAAA, it advanced 59 minutes. Clearly there was something more complicated than just the faked counter you describe.

  2. Do you think any of the backlash was also because participation in the ‘release acceleration’ required a financial investment, followed by a considerable time investment?

    I agree that the internet has much too much in the way of entitlement, however the game was only released ten hours early, which isn’t much at all, even at their most modest estimation.

    Out of interest, how fair do you think countdown schemes like this are? There is accusation that the results of these are rigged irrespective of participation. Is it important for them to be entirely forthright?

    1. There was certainly backlash due to money being involved, but that’s tangential to what I discuss above. I’m sensing more betrayal than indignance; people are hurt that they chose to play and didn’t get a proper reward, not that they had to pay to play. People who didn’t want to buy the indie games and didn’t already own them didn’t play the final phase at all; their annoyance at the ARG is understandable, but no game design changes would have satisfied them save a fundamental change to the premise of the ARG.

      I think whenever you are providing a real-world, appreciable reward like the early release of a game, you need to be very, very careful about communicating properly with your players. The results are always going to be somewhat rigged, governed by real-world business constraints and the inevitable victory of the Law of Large Numbers over individual efforts. It makes it worse if you are so vague with your players that they suspect you of fraud.

  3. I think there were two other factors to people’s expectations of the ARG: firstly, the launch date of 7:00 AM PDT on 4-19 wasn’t widely publicized, if at all. The date most commonly given out was 4-19, and I think most people assumed that, like pretty much all previous Valve games, this meant a midnight launch on Steam. Both Left 4 Dead games, for example, launched at 12 AM EST on the day of release. Portal 2, IIRC, came out around 11 PM EST, which means that everyone (who was willing to stay up very late on a work night) got one hour over what the common expectation was. So for a lot of people’s expectations, the game was very close to zero-sum – and to be honest I’m not sure it wasn’t planned that way from the beginning given the short tempers of retailers when it comes to perceived threats from digital distribution.

    I think the second thing that raised expectations was marketing that seemed to extend beyond the bounds of the ARG. I know that sounds ironic, since ARGs by definition go beyond the bounds of normal marketing, but it seemed to me at least that things like the ad banner which got placed on Portal 2’s store page with “Help release Portal 2 early” in big bold letters on it certainly didn’t help manage people’s expectations of something further away from the midnight launch window that they were used to on Steam. Nor does it help the justification that trolling makes sense in the fictional reality of the game they were promoting.

    Personally, my policy is to ignore ARGs altogether. The one that Valve ran got through to me because of the free content they gave out for games I already owned (which was an awesome move on their part, and one I don’t want to seem ungrateful for), but I think the end of it was a misstep and reinforces my desire to continue ignoring that sort of thing in the future.

    1. Anyone playing the metagame was able to very clearly see the originally stated release date. Maybe Valve could have designed their UI better to make it more obvious, but they were perfectly up-front about the initial release date that was being modified.

      I do agree that expanding the advertising was a mistake, since it gave people who weren’t following the ARG got incorrect expectations. I wonder if they didn’t increase advertising at the last minute in an attempt to get more ARG participation, hasten the release, and prevent exactly the situation we’re now discussing?

    2. It’s not just that Valve’s earlier releases were at midnight. Several retailers had midnight launch events, so on the East Coast, the people who bought direct from Steam only got the game an hour early.

  4. I agree with most of this well-written article, but the whole effectively getting released a day early thing only counts for some people (most notably the United States, of course). Here in the UK, for example, it ended up being released at around 7am. Obviously this meant that most people had to wait until they got back from work to play it, meaning they got to play it at most a few hours before they would otherwise, and in most cases (9am-5pm being common hours, with the scheduled release being around 5pm local time if 10 hours were saved) ended up making no difference at all.

    I agree regarding the ways to improve transparency, but you can’t help but think that if they HAD made it clear that it was a case of “help release Portal 2 hours early!” rather than the days that people had assumed from the ambiguous wording, it wouldn’t have drummed up anywhere near as much attention.

    1. Yes, the UK and other non-US people did get the short end of the stick. I meant to specify that I was talking about the US with regard to the day-early playtime, but that clause got lost in a revision. I’ve edited to not exclude non-US people.

      It’s definitely true that more clarity would have reduced the marketing effectiveness of the ARG. I suspect that was a debate that happened internally at Valve.

  5. Having missed out on the majority of the ARG (HA), I can’t complain. And since my computer couldn’t keep up with such a game, I’m still waiting for Amazon to send the game for my Xbox.

    Which means that I wouldn’t have won out, regardless. So, like previous ARGs, I am curious as to how many this spurned on to play. Did it garner more attention than than the Invest in Aperture Science videos? Or was it a very clever ploy to trick more people into playing a bunch of rather cool indie games?

  6. Valve basically held their own game for ransom. They pretty much said “we’re ready to release the game, but we won’t until you give us some money”. I don’t know what to think about that really.

    1. I’m not sure if it counts as holding it for ransom if they’re offering to do something ahead of schedule. A bunch of people preordered it, effectively giving $40 or so to get the game on the 19th. Valve then said, “You can do something extra to get it earlier!” No one was ever going to get the game later than promised.

      1. But, the only way they could have done that in the first place is by delaying the release date a day or so late, then resetting it back to normal with the potato deal.

        1. Or projecting a release date with a tolerance for error, knowing that software schedules can run late, that console certification can take time, and that Valve has a history of late releases. Then, when they actually get the game done a bit early, they realize they can release it early.

          Besides, no game on the scale of Portal 2 is released right when it’s ready. The release is always scheduled to fall on a day that’s strategic for the market. Things that can affect a release date in this way include other games being released around the same time, when weekends fall on the calendar, and when digital download services are expected to have the capacity to support a huge release.

          1. Good points, I hadn’t thought about those. I guess it’s still possible that they set a much-later-than-necessary release date in order to do what I suggested, but that seems less likely now. Still not sure what to think about the whole thing though.

          2. I would have to agree. Really, I was just putting the idea to voice. Also, I don’t doubt that that was what most people were thinking of, afterwards at least.

  7. I have to say, I didn’t think the last portion of the ARG was a betrayal, but it did sort of end what was fun on sort of a sour note for me. I had bought the potato sack moderately early- once I had caught wind of the thing going on- and had a good time looking for all of GLaDOS’s stuff.

    What bothered me about the end was that it felt sort of like a break in the ARG. I mean, I know its marketing, and they know its marketing, but part of the fun depends on both of us pretending it’s NOT marketing. If the game had released once all the intrigue and hidden puzzles had been sorted out, I think there would be much less complaining, even if that had theoretically taken till the exact same time. As is, it felt like they moved the bar from “Reboot GLaDOS to release Portal” to “When we meet our sales quota, we’ll release Portal” even if that isn’t entirely accurate. Suspension of disbelief got pierced.

Comments are closed.