Sign up for news

If you set out to make a game that really uses choice to both immerse the player and make them come to a better understanding of themselves through those choices, there’s an underlying problem that we have barely hit our heads against.

Fallout 3 pushes on those boundaries, unrelenting in its empowerment of the player to change the storyworld. Now, it’s not like this has gone unnoticed, but it’s worth thinking about the design problem, and how to deal with it.

If you pay $60 for a game, you’re starting with a pretty decent urge to get your money’s worth. Games of this nature, for the longest time, have had traps of choice. Certain paths in the game, whether you had an idea of what the consequences would be or not, could ruin your play experience. Maybe those choices wouldn’t spoil the play experience for everyone, but they would ruin it for you by forcing you to play the game in a way you’d rather not.

Anybody that’s played games for a while has a story along these lines, a warning they remind themselves of every time they enter a world more expansive & with more agency than the last. I remember setting out in The Elder Scrolls: Arena, practically orgasmic with the back-of-the-box declaration that you could actually walk to each town in real time. I eventually made it out of that goddamn first dungeon, sorted out my business in town, and sallied forth, inventory-as-bindle in hand.

A week went by (and this was when I was 17 when a weak of play time meant much more than it does today). I hadn’t reached the next town yet. I was a little worried, but I soldiered on, exploring crypts, dungeons, and other buildings that are underground and populated with skeletons. After week 2, however, I came to the sad realization that I was never going to see another person in the game again. 

There was no way I could reconcile with the commitment it would take to restart the game and try it again. I was done. It served me as a warning for any open world or open ended game, narrative or no, watch out when they tell you that you can do anything. It’s usually a bad idea (no matter how cool it sounds).

As we all know, Bethesda eventually sorted out some of those problems. In Fallout 3, I am happy to be the wanderer. However, like most wanderers, I am an observer to the stories going on in the world.

In Arefu, I got the vampires to stop attacking the families, but I left the vampires be as well. I let the Antagonist go, once she gave up her uniform. In the Underworld, when Ahzrukhal offered his slave Charon in return for killing his competitor Greta, I jumped to dismiss the option. 

I refused to pass judgement on those before me, lest I be judged in turn. Who am I to leave my safe vault home, enter their world, and decide who is right and wrong? What if it were to come back to me?

There’s a tension between making the player connect strongly with characters, and giving them as much agency as possible (any frequent reader here will hopefully realize I am in no way talking any idiocy commonly spouted about conflict between games & story). This problem only presents itself once you’ve gotten past the design intricacies of actually making all the paths work, with different consequences for each. If you manage to do that and make it all tie together (most of the time), which is what Fallout 3 does better than most others before it, you’ll have painted yourself into an odd corner. Without forcing a choice, choosing nothing becomes the best way for the player to leave the world as they love it. By relying on forced choice, the player can resent the game for forcing its ruin upon them.

Most games of this genre falter at the start – you’re presented with character creation choices that will impact your entire game without any understanding of their real effects. D&D games get a free pass on this problem, because most people have acquired a general implication of how those systems work, what it means to play as a fighter or a mage, etc. Bethesda games usually give you a opportunity to change any selections once after a little bit of play, but is the player honestly yet in a position to judge the future entertainment value of being able to pick locks and sneak into places vs. talking their way past difficult situations?

During the game, obviously giving the player as much knowledge about the consequences of a choice helps avoid the problem. But it just avoids it, it doesn’t solve it. You want your world to have unforeseen consequences, because it brings the world alive and reveals character. If you kill Greta and Ahzrukhal gives you Charon, Charon will kill Ahzrukhal before he becomes your companion (oops, spoiler). You don’t want to get rid of that. You just want the player to be ok with it, and feel like a part of the world because of it.

You can manage and pace those effects, the perceived rewards or punishments. For an excellent analysis along those lines, check out Randy’s talk from MIGS/GDC 2007 on save/load compulsions. That thinking, while incredibly useful, still assumes these choices have consequences that will be predictably perceived as a reward or punishment. When it’s a change to a character in the world, you may not know if the player has any attachment to the character, or they might have some larger plans for that character in the future. That could turn a minor character change into a much larger pain.

There’s a larger overall barrier too – the simple fact that up until now, a player could rarely trust a game to do anything reasonable when it comes to this problem. It’s a bit of chicken vs. egg. If players don’t trust their games to not screw them over this way, they’ll never let the game surprise them by getting screwed over in a way that’s entertaining or meaningful.

Maybe if we work really hard, pacing those rewards and punishments as Randy suggests, we might slowly get over everyone’s impression that games are inevitably going to fuck you over when it comes to your choices. Only then will designers really be free to explore choice in settings like this, to present the player choices that will upset their preconceptions and make them learn something about themselves they weren’t expecting. (Well, we could make the damn things shorter, too.)

Maybe then we’ll come to grips with the choices that are our vault fault.

Sorry, couldn’t restraint myself on that one.

8 Responses to Your choice, and your fault.

  • Dave Mark says:

    Some of that reminded me of the infancy of Ultima Online. We were all faced with this enormous list of skills we could acquire and take with us into the “real world” of Britannia. Fishing, hunting, camping, tailoring, lockpicking, tinkering, cartography, tracking…

    And then, after working for hours to increase our “camping” skill so we could log out anywhere in the world, we realized it was probably better to simply cast Recall on a rune, teleport immediately back to the safety of town and log off there.

    Tailoring? Why bother… no one will buy your clothes. Hunting? If you truly need more material, Recall your way back to town, buy something, then instanly Recall back to where you were.

    Lockpicking? Heck… what level spell does it take to open that chest? Pthew… easy enough!

    The funny thing was, it took months for that realization to set in. We wanted the experience that was on the “back of the box.” We wanted to truly see the benefit of having 40 different skills. Imagine a world of specialists… each plying their own wares! Magnificent!

    A year later, everyone in the game was maxed out in 3 categories. And only about 5 of the 40 categories were the ones people chose to max out in. So instead of having a world full of individuals with individual constellations of skill mastery, the whole world boiled down to GrandMaster Swordsman, GrandMaster Mage, and a few others.

    So much for choice, eh?

    I think one ironic drawback of a game that changes a moderate amount, but not a significant amount based on your decisions is there is a possibility of a disproportionate balance of regret vs. replayability. I believe you touched on this above. If you realize that your decision changed the gameplay enough to disappoint you but it doesn’t change the gameplay enough that replaying it will be significantly different, you are going to be unhappy.

    An example would be NWN2. I realized eventually that the decisions I made in dealing with my party-members affected how they were dealing with me… even to the point where there were parts of the game that were blocked off. I lamented that and even regretted some of the decisions I made. However, there was not enough that was different that I was interested in going back through and playing in a completely different manner.

    On the other hand, games such as Black & White made it so that the decisions I made changed DID change the world enough. The experience could be completely different the second time through. That meant that there was replayability value in the choice/effect chain. I didn’t have to feel like I missed out on something.

    Anyway, great post.

  • Borut says:

    Thanks Dave!

    Yeah, the length vs. actually different content is a big limiter. Even if 10 hours out of a 20 hour game can be greatly different, I think it’s pretty rare for people to go back and redo the game. Not that it doesn’t happen, but, comparatively, if you just watched a season of a TV show on DVD, would you go back and watch it immediately again? Real fans might go back and listen to the commentary, but that’s about it. If you have the real posssiblity of replaying it you might not feel that regret – that’s only realistically going to happen when a larger percentage of it is not just the same game.

    So you end up with a situations like you described with NWN 2, the way you played would probably not have been as entertaining if you had made the decisions with the awareness of the depth of the party character storylines. When you just have no idea those decisions are going to lead to that, you’re inevitably going to regret it as a player, unless the developers have taken the time to balance for it (if you make a limiting choice that you weren’t aware of, there had better be something compelling going to make you go “well, I didn’t want to do that, but that was pretty neat so I’ll stick with it”).

    Setting off the bomb in Megaton in Fallout 3 is a little like this. Granted you know it’s going to have some impact ahead of time, but no idea what (aside from the obvious). Once you do it, they open up some different paths – I think some of the paths you could go down otherwise, but they take more effort to go down. It’s not just shutting down the missions in Megaton, but opening up others and showing you had a really big change on the world at the same time (which has some inherit coolness/ownership to take the edge of having made a limiting choice).

  • Dave Mark says:

    For comparison purposes, I believe the other side of the replayability coin is something like an RTS or a TBS. I can replay Empire Earth or Civ 4 many times without tiring of it. The reason for this is that the possibility space generated by my actions is near infinite. Right out of the gate, if I do something different, there is a significant probability that the outcome will be affected.

    The reason this is simpler to do in a strategy game, however, is that the content is self-organizing. The narrative is self-directing based entire on the chaos theory that is at its core. There doesn’t need to be a designer playing “guiding hand” to move the story along and make sure that I get there. Likewise, the assets are all the same. There is no extra animations, cutscenes, MoCap, or dialog that needs to be provided to account for the massive possibility space. The game arc simply uses the same units in a different way.

    Compare to the types of game we are discussing above where, due simply to the necessity of crafting the experience, there is a finite number of paths to take and cause/effect pairs that can be seen. FPS and RPG games can’t handle in minutia of contingencies.

    In your Megaton example, someone had to account for that particular “what if?” and pre-determine the effect. As developers, pre-determining responses to outlier cases gets expensive. We don’t have the luxury of pre-crafting a response to every possible action that the player may take… therefore we have to limit their actions. Selecting those limits is best done on a likelihood basis. “It is likely that the player may want to do X, Y, or Z, but not A, B, or C. Therefore, we will write the effects for X, Y, and Z into the game.” By accommodating the most likely decisions, we also tend to select the ones with the most obvious effects. Players can therefore infer what would have happened if they chose something different. Exploration of those obvious scenarios is not rewarded.

    On the other hand, if the player is aware that many wild and crazy possibilities are on the table, they are more likely to explore those possibilities. It also leads to replayability because the ability to infer the outcome is out the window.

    To use your movie metaphor, I think my last point is more like a TV show. Same characters, same world, but the plot of each episode is completely different. Even entire seasons of a show are worth watching because the plot of the season is going to be different. Think “24”… we know Jack is going to save us all in one day… we just don’t know how it will pan out. Therefore, it is worth watching each episode of each season.

    Until we can more adequately manage the content issue, however, we are going to be unlikely to put that much dynamic content in our games to account for that free exploration of ideas. After all, not all game studios have the $100 million sitting around that it took to put the content in to GTA4!

  • Reid says:


    “We don’t have the luxury of pre-crafting a response to every possible action that the player may take… therefore we have to limit their actions.”

    That’s why I wish games were both shorter and more focused. That way the possibility of player decisions has more focus instead of “they can do anything.” For devs, I think a shorter game will all for more depth in consequences from player choice. I’d much rather have a short, but in-depth game than a long and shallow game.

    I haven’t played Fallout 3 so my comments aren’t directed towards it.

  • Dave Mark says:


    While I agree with the logic your suggestion, there are other methods. There is a lot of exploration being done with more dynamic content and responses. Even going back to something like Dungeon Seige where the world continued to generate no matter where you moved was a start.

    Now obviously, generating a random open world without any depth of interaction is relatively simple from a simulation standpoint when compared to generating “deep” content such as quests and especially dialog. However, I believe this is a cusp which we are approaching from a technological standpoint.

  • Borut says:

    I think the depth of replayability of an RTS is possible in a character-based game, pushing more things into simulation. But this is pretty hard, i think the only way any game will actually start to innovate in this area is if the length is much more constrained.

    I don’t know that content generation will necessarily help – that is, until we can build something of a short length that has this amount of depth & consequence. Content generation would allow us to extend the length of the experience and maybe add options as well, but only once we figure out what we’re doing on a small scale.

    It’s maybe also worth noting that by simulation I don’t automatically mean realistic simulation, making things work more like the real world, that’s just one way to skin that proverbial cat.