a series of interesting choices thoughts on game design from paul sottosanti


why i’m skeptical about active reload in SpyParty

Tonight I'm heading over to Chris Hecker's house to playtest a new feature in SpyParty, active reload (in this case, "Action Testing"). SpyParty is an incredible game that I've had the pleasure of playtesting for many hours at this point. If you haven't heard about it, here's a quote from the official site: "SpyParty is an asymmetric multiplayer espionage game, dealing with the subtlety of human behavior, character, personality, and social mores, instead of the usual spy game explosions and car chases."

The easiest way to explain how it plays is to compare it to a reverse Turing test, where instead of an AI pretending to be human, the Spy is pretending to be an AI. The job of the Sniper is to watch for discrepancies in behavior between the AIs who inhabit the party and the Spy who is walking in their midst. This could come through simple mistakes (walking erratically, bumping into things), more subtle behaviors (that guy is spending a lot of time at bookshelves), or seeing mission completions (that guy just reached into a book and removed some microfilm!). These can be categorized as "soft tells" that make you more suspicious but don't confirm anything, and "hard tells" that give the Spy away completely.

Chris has already written a blog on the new feature. Here is his primary reason for wanting to try out the mechanic:

The flaw in the current game is if a merely good Sniper is playing an elite Spy, the Spy player can’t accomplish any missions if the Sniper is looking directly at the Spy, even though he or she is much more skilled. The Spy will still win most of the time, because the merely good Sniper won’t be able to tell which partygoer to watch if the Spy is good enough, but I still consider this a flaw because an elite Spy should be able to perform missions right in front of the less skilled Sniper due to the skill differential.

The source of my skepticism starts here, because I actually disagree with this premise. There are essentially four major levels of suspicion between the Sniper and Spy in the game (with more gradations in between):

  1. The Sniper has been completely thrown off track and has either marked the Spy as "not suspicious" or is focusing on another suspect closely.
  2. The Sniper has no idea who the Spy is and is either surveying the whole party or watching a particular objective.
  3. The Sniper is suspicious of the Spy and is keeping a very close eye on him, either from one or more soft tells, or process of elimination, or a gut feeling.
  4. The Sniper is "sure" about the Spy, likely from having seen a hard tell.

Once the Sniper reaches the fourth level, the game is essentially lost for the Spy. He is about to shoot. (Although I have won a couple close ones by performing the final mission in full view of the Sniper and then doing my best to hide behind and weave through other partygoers for the final ten seconds, but that won't work against strong players.)

So let's move on to the third level, which is what this mechanic is intended to address. Chris's point is that elite Spies in this situation should be able to perform missions right under the nose of a merely good Sniper, but I disagree. The interesting interplay of the Spy and Sniper is about two things: not letting it get to this point in the first place, and then recognizing when it does and going into damage control mode, blending back into the flow of the party.

There's nothing worse as the Sniper than having a hunch and following it, watching a certain person closely for a good 30 seconds, and nothing happening. A growing panic forms as you wonder if you were correct in the first place, and you start to worry about what missions might have been completed when you weren't paying attention to the others. You have to decide whether or not to cut your losses and abort your close observation or just continue following and hope for the best.

(As an aside, if you're wondering why a Sniper can't just watch the entire party at all times, one integral piece of the design is that there's simply too much going on at any one time to focus on all of it. At least, I certainly can't. If there are Snipers who can I think that'll create even bigger problems, but luckily there's an easy tuning knob, the total number of people at the party, that you can turn to increase the information density.)

Unfortunately, the Action Testing undermines these interactions. Now you really don't have anything to go on, because you might be watching the actual Spy and he's finishing missions right under your nose! My theory is that this will make losing a lot less fun, because it introduces that feeling of "there was nothing I could do." You could be watching the actual Spy for the entire party and still lose. Yeah, even the awesome results are still theoretically noticeable, but remember that the stated goal is to allow the Spy to win while being watched, so I can't imagine it'll be easy. There's a question I've been thinking about a lot recently: "Is your game fun to lose?" I think adding Action Testing risks hurting SpyParty a lot here.

You'll also have the games where the Spy flubs an Action Test for whatever reason. At that point, there should be a good chance that you notice him, and win the game on the spot. Here's another relevant question: "Is your game satisfying to win?" Sure, it's always fun to win, but the brilliance of current SpyParty comes in forcing me to interact competitively in novel ways. When I win it's because I picked up a subtle clue, or outthought my opponent. Winning because my opponent or I were good or bad at a reaction/timing mini-game is something I can find in countless other games.

And finally, there are two aspects of human nature that concern me here:

  1. Humans aren't that good at properly evaluating risk, especially when randomness is involved.
  2. Humans tend to overvalue their own abilities.

Both of these point to people trying the Action Tests far more often than they should. The design intent might be to provide a tool for someone under close scrutiny that shouldn't be used otherwise, but my guess is that people are going to use it far more frequently than that. The more people use it, the more the game becomes about the Action Testing and the less it's about its other novel aspects.

With all that said, I love Chris's specific implementation of the active reload system. Randomness in multiple axes should go a long way towards preventing people from mastering it, which was another initial concern. And I'm certainly hoping that I'll be pleasantly surprised tonight. We'll see!

edit: My post-playtest writeup is now up: everything went better than expected.

Comments (0) Trackbacks (1)
  1. Really love this, please keep writing more! I was (am) skeptical about the active reload as well, but as Chris said it’s not the major focus of this game, and I can see how it helps the interaction with the game.
    Anyways, he said testing went well, so interesting you hear your thoughts on it.
    And please keep writing! Chris sometimes forget I’m starving for blogposts xd (don’t tell him).

  2. I think (and hope) the design intention is that a perfect action test does not make an action undetectable, but just less obvious. If you’re already at the point of only watching one partygoer then you should be able to spot even the perfect actions.

  3. Yeah, that seems to be correct. In our playtest the Sniper wasn’t able to spot all of the perfect actions (even when watching the Spy somewhat closely) but I’m sure that will change over time.

    It’s interesting, in the playtest we were seeing that an “elite” sniper (Ian, who honed his skills over three days at PAX) almost couldn’t lose, especially when he saw something early and marked you as highly suspicious. That’s actually what I was arguing for here, but in practice, as the Spy it feels rather frustrating. You could blend in perfectly for a full minute and a half and then attempt a single mission and be instantly shot. The AIs don’t do enough to arouse suspicion on their own to ever take it away from you. So it did feel like there might need to be a way for the Spy to deflect some attention in that case, and while the Action Tests didn’t quite do that, they did force the Sniper to pay closer attention to a single character if he wanted to be sure to spot everything. And they made the Sniper focus more on watching behavior than on looking for animation tells, which we liked.

Leave a comment