I'm researching for an undergrad thesis, and I'm seeking input on 1) whether people think this is viable and 2) what kinds of criticisms/concerns people have about it. For a long time I've been toying with the idea that someone can learn about another player from their own personal gameplay experience. A player makes social decisions altering the status of the game world and subsequently exports this modified world upon completion. The exported data is imported to a new players game whenever they begin a new game (at which point they make decisions creating another export file, etc.) The theory is that the new player can learn personality details about the first player as a result of their own gameplay (either by working with or against with the social environment of the game world).
Example: I play the game and encounter a ruler who kills all traitors. During gameplay, I succeed this ruler and designate that traitors to my city are not to be killed or sent to jail, but added to a specialized squadron of the army for frontline combat. I save my game and it is sent to a database. My friend picks up the same game and begins a "New Game" at which point the program grabs a saved game from the database and uses it as a foundation for that world. When my friend plays, he finds a world in which the NPC ruler sends people to this other squadron, etc. At the end, my friend sees in the credits that the game world was derived from [my account profile name]. He then might begin to think that he can know a bit about me based on the game interaction (as the theory goes).
Problems I see off the bat:
1) In order for people to be able to learn about the player, it requires that the players must have choices, ergo, a large variety of choices must be possible for any given decision, ergo, a lot of content must be created. The best way of handling this seems to me to be to develop game social environments based on procedural generation, but I'm not 100% sure on how, in concept, that should be done.
2) In order for people to learn about the player, they must know what the player chose in comparison to what they DIDN'T choose. This means that I am likely assuming that a) the player knows all possible choices for a decision without being informed, OR b) the player will HAVE to make choices based on a comprehensive "list" telling them what options they can select (thereby increasing the probability they assume that all players met the same list of choices in their own playthroughs - that way they will be aware of what wasn't selected). This is a major problem if I want a natural decision-making model in which the player makes choices via actions as opposed to menu selections where their options are inferred based on the story context. Additionally, this is a problem if I employ a branching model in which previous choices alter their available options later down the line.
Any comments/suggestions are welcome.
Cross-Gameplay Communication
Yeah, I agree that those are problems, and it's tough to get around them in a "story-choice" game.
My thought is that some of this is mitigated in a more "builder" type game, where most player choices, and their results, are instantiated by physical changes to the "kingdom". The communication problem (how does the player know what options were open to past players and what past players ended up choosing) goes away; much of it is right there to be seen. And you don't have to plan out every choice this way. (For example, draconian laws will mean more convicts, and more convicts require more prisons to be built, and future players seeing those prisons will be able to understand, "Oh, that must have been a draconian ruler." Even if they don't know the game yet; "That's a hell of a lot of prisons" -> "That was a draconian ruler" is a natural real-world inference.)
The design challenge, then, would be to make a builder that maximizes "expressiveness"; one where your personality can have a big effect on the world to the extent that future players can infer it by observing what you built and what happened to it.
The second design challenge would be to give the player a reason to care, some reward for paying attention to previous players. (For example, picking through the ruins of past players' kingdoms might give resources, technological advances, etc., and the greater extent to which you can work out the player's personality the better you can work out what will be there and where to "dig" for what you need.)
My friend picks up the same game and begins a "New Game" at which point the program grabs a saved game from the database and uses it as a foundation for that world. When my friend plays, he finds a world in which the NPC ruler sends people to this other squadron, etc. At the end, my friend sees in the credits that the game world was derived from [my account profile name]. He then might begin to think that he can know a bit about me based on the game interaction (as the theory goes).
The first issue which comes to mind with this approach is, that your approach heavily leans on 'the player needs to play the game the designer imagined' and 'the player needs to follow the rules'. Both has been proved to be not valid in many cases.
One of the saddest example was Ultima Online. They created an fantastic AI/eco-system which would evolve to a living, dynamic world... but after the game has been released, all players just killed everything as fast as possible. There was no room for any evoluation of the system under this circumstances. The eco-system collapsed and had been removed afterwards.
Your system is similar, it will only works if your players will follow the rules. If they dont follow the rules, the next player might encounter a boring, dead world.
For a long time I've been toying with the idea that someone can learn about another player from their own personal gameplay experience.
This on the other hand is possible. Eg. instead of directly manipulate the game experience (in worst case this would be the destruction of it), your game experience could be enhanced/guided by other players. I remember a game where you saw red ghosts trying to accomplish a challenge, but failed (dont remember the game name, dark souls ?) These red ghost were attempts of other players who did not master the challenge and the player could just watch this ghost to avoid similar failures. This could be expanded to RPGs, where you hear a rumor about a den, where many warriors tried to free the pricess but got killed in the attempt (just analyse the statistics, which class would be more likely to fail at a certain quest).
My thought is that some of this is mitigated in a more "builder" type game, where most player choices, and their results, are instantiated by physical changes to the "kingdom".
It would seem that the natural experiment would be to have a bunch of people play MInecraft for several hours (days) and then switch around the worlds people play in and see if anyone can tell anything about the previous player. But knowing that they might be observed, it's possible that people might change the way they play their game and affect the results.
I think people are perceiving the structure of this game slightly off. The game will not be open-ended in the sense that players will have complete freedom to do many sorts of activities and then other people see what they've done (which seems to be Ashaman's primary concern). Rather, it is more akin to playing a role-playing game such as The Witcher 2 or Mass Effect, and they actually witness the results of the decisions made via the social interactions they have in the world. So I think Ashaman's concern about the Ultima Online-like isn't as valid because the game I intend to build is already more like the type of interaction he described in Dark Souls (though slightly more indirect).
I don't plan on this being a "builder" type game because the game will likely end up as a text-adventure game and it is the story and adventure that will be the primary attraction.
The design challenge, then, would be to make a builder that maximizes "expressiveness"; one where your personality can have a big effect on the world to the extent that future players can infer it by observing what you built and what happened to it.
Maximizing "expressiveness" in player decisions is where the hyper-branching design challenge originates, so I have already accounted for this challenge (though not handled it adequately enough. Meeting with a mentor to discuss the details more tomorrow). I am told that procedural generation of story content could be a possible solution, though I'm not entirely sure of the details for implementing such a system at the high-level stage even.
The second design challenge would be to give the player a reason to care, some reward for paying attention to previous players. (For example, picking through the ruins of past players' kingdoms might give resources, technological advances, etc., and the greater extent to which you can work out the player's personality the better you can work out what will be there and where to "dig" for what you need.)
I actually -really- like the idea here of ensuring that the story and environment for the game world precludes a need to involve the player intimately with the past. This seems like a sure-fire way of generating user interest in the previous player.
But knowing that they might be observed, it's possible that people might change the way they play their game and affect the results.
I hadn't considered this...It's quite an issue for the end goal of the project which is simply to have people learn about others in a natural, fluid way, using a video game. I'll need to add this to the list of design challenges. "How best to encourage players, via design, not to bend their behavior due to other non-present players"
I actually -really- like the idea here of ensuring that the story and environment for the game world precludes a need to involve the player intimately with the past. This seems like a sure-fire way of generating user interest in the previous player.
Erm, just to check: do you mean "includes", rather than "precludes"? The latter means, roughly, "prevents the occurrence of", I believe.
However, my main question is: how does a given player interpret information about the previous player? If they arrive at a flourishing kingdom and hear of that past conquerer (presumably identified as a previous player, rather than an NPC or backstory character) who brought down the old tyrant and instituted an age of peace, how do they interpret that? Was that player an idealist, wanting to make the world better for everyone? Were they simply in it to conquer a city, but happened to be an effective and non-tyrannical monarch? Did they conquer it because it was a threat?
In short, how does one read player motivations, especially in a world of (presumably) limited interactions?
Hmm... Another thought, this one perhaps more positive: it seems to me that one can learn from what a person doesn't do as well from what they do; for example, when I played The Path, I almost never (knowingly) directed my character towards her "wolf". I guided her around the rest of the forest, and into the house at the end of the eponymous path, but steered her clear of the "wolf".
Finally, perhaps a partial answer to my question above might be to take multiple player actions together to form a single observation: a player who kills a single foe may have been doing it as an expedient, or because they saw that individual as a great enough threat to warrant immediate death, or a variety of other reasons; further, they may have decided afterwards to never do that again, or made some other (or no) decision. One may perhaps draw more accurate conclusions if one knows that the same player killed all of their foes, and did so brutally and publicly.
Maximizing "expressiveness" in player decisions is where the hyper-branching design challenge originates, so I have already accounted for this challenge (though not handled it adequately enough. Meeting with a mentor to discuss the details more tomorrow). I am told that procedural generation of story content could be a possible solution, though I'm not entirely sure of the details for implementing such a system at the high-level stage even.
Before even tackling the procedural story generation, set for yourself a task of writing 5 to 10 example stories that a player could experience, along with the post-game evidence (whether that's NPC testimony, lore books, obelisks, ruins, whatever) that would allow a second player to deduce the choices and personality of the first player. (Full-game stories, not just 5 to 10 one-choice vignettes.) You can't procedurally generate stories if you don't know what kind of story you're trying to generate. But once you have a set of stories and say, "OK, I need to make a system that can at least generate these", you've got the beginning of story generation.
If during this process you're having trouble writing that many stories that are interesting for the player, interesting for the next player, and clearly illustrate the first player's personality, then that's a sign that there may be a more fundamental problem than implementation.
