Jump to content

  • Log In with Google      Sign In   
  • Create Account

[Theoretical] Games that evolve from player input

What is the Human GM Emulation System?

Posted by , in Development, Introduction 18 December 2013 - - - - - - · 738 views
human emulation, theoretical and 3 more...
What is the human GM emulation system?

In short, it's a system like no other that procures ethereal cloud games.

Although it is only theoretical at the time of my writing this. Here's a high abstraction of the technical requirements of the system, followed by some of my own expectations for it.

note: A complete documentation is beyond the scope of my current goal of explaining my idea.

GM emulation is human emulation.

This is a very important component of the system. The system must learn. In order to grow and behave like a human, it will need human emulation.

Human emulation is not a new idea. According to NextITCorp [1] in their video which references customer service software, it can be defined by:
  • conversational interface
  • personal
  • provides a single, correct answer
  • goal-based
  • contextual awareness
I didn't see the exact meaning given for each term. All of these things are certainly important for a meaningful human interaction. It is a good starting point to understand what the system needs to work with to appear human, especially while emulating a GM "read about The GM below to understand why."

If you view the video you'll see it indicates that human emulation is very important and why.
During my brief search I found this passage referring to human intelligence emulation: "It makes decisions by itself. If you teach it your strategy and tell it what you have to achieve that strategy, then it will tell YOU first, what's wrong with your strategy and how to fix it, and then how to use the resources you identified to achieve what you desire; and tell you if the resources could be improved. It will accept your feedback while it learns. And it can change real-time as conditions change, just like a human." - Russell S. CEO at Scientific Human Emulation, represented at Johannesburg - RSA, New York - USA, and London - UK [2]

The above seems too good to be true (if he was referring to a perfectly functioning program) or maybe I took the claims slightly out of context. This is exactly the functionality that the system needs.

I cannot go deeper to determine if the requirements are already met, or this would be a very complicated document.

A database containing the accumulated content is necessary. Content can be considered 'ideas in the form of multimedia' for the purpose of my explanation. Metadata will allow connections such as who created the content and where it is being used.

The beginnings of metadata and content on this database will be from numerous sources. Content can be unintentional, generated by response. Content can be intentional, uploaded.

The players will eventually determine how and what content is used through their input. They will generate new metadata referring to every bit of content that appeared in any situation.

GM emulation is human emulation in a game environment. So I think it's about time to define what a GM is.

The GM is considered the highest authority in any one game which requires a GM. They make the game happen for the rest of the players, and they can enjoy their role as well. This places the GM of such a game one rank lower than the original game developer in terms of defining it.

A GM often takes on multiple roles - GM, in-game characters, storyteller, planner
  • in-game char. - characters who behave and make GM-like decisions, other characters that allow the GM to take part in the game as a player so they may experience it, and npc-type characters that perform a specific task.
  • story teller - a narrator who talks about a scene as it unfolds to give it context
  • planner - Someone who provides additional rules, story, and objectives to create a unique player experience.
A GM is able to understand needs that different players have. They interact with multiple users on a personal basis.

They are a mediator and disseminator of information between players and how their actions affect a game. A GM understands the game in ways they may affect it, often they make decisions that dramatically affect the game for a single play session. Suggestions to the players to keep a game flowing, or bending and breaking rules are examples of influence the GM has.

Sometimes they are mistaken for god. If something clever or new is suggested, and the GM says it's good, then it will work. (There is nobody above god. The developer is above a GM. The GM is not god. The developer is human. I just thought someone might like me explaining that.)

They enforce rules and track a game's progress to prevent 'accidental' cheating.

A GM does not want an unpleasant player experience (for others or himself), so he can be persuaded with due effort.



The system will need to perform according to the user's connection speed.

This almost went without being written. Anyone playing a game will need to connect to the system to take advantage of it. However, connections relying on a slow internet will have realtime limitations. There is no reason for the system not to understand this.

Common sense is relative. A person must learn what is acceptable in as many situations possible in order to meet expectations.

Because of this need for common sense in order to meet expectations, it is not possible for the system to behave as desired until it has matured enough through extensive player interaction. At which point it will avoid very obvious mistakes, most of the time.

The system is defined by many players, and over time, many games. If the entire population interacting with it changed suddenly, it would require time to adjust to their changes.

The system itself may be considered a GM while it performs the duties of a GM. It must grow to understand every role it provides. Part of the learning process is practice and observation. It would be expected that the system itself is able to practice and observe.

For the same reason a GM does not want an unpleasant experience, it would allow 'exceptional' players a pass. Any undesired impact would be mitigated by a lack of exceptional players.
Exceptional players.
Such a player may complete a course in a racing game in 0 seconds.
Another may beg the GM for frivolous things.
Some players are simply being treated unfairly by the community.

In any case the GM would need to make an exception or avoid the player altogether. These examples would be hard to resolve fairly in a structured game environment. Usually these players are deleted, ignored, or they succeed in affecting everyone else.

Over time the system would recognize all three of these players and have suitable responses for them so everyone can enjoy it.

The system is a knowledge bank (database and other software) able to emulate a GM with access to everything required to replicate and create content between connected games.

There are many games, and many ideas. Games are often independent from one-another, but game sessions are not always independent from other game sessions. Many ideas connect in unusual ways. The goal would be to connect as much of this content possible and grow from how players respond to it, progressively improving and predicting what content goes where.

THE SYSTEM when connected

The system is allowed to monitor non-GM games, and it also provides a GM session to a game that explicitly requires one. By extension, the system understands countless other games and has a level of direct content control, new ideas in, new ideas out.

The system won't necessarily communicate through words, but through the content. When a user responds to content, the response may be interpreted as a rating. If a user response is also creating new content, it is shared with other users. This will allow the system to grow outside the confines of a word of mouth game architecture.

By extension, players indirectly communicate their gameplay through content even when games aren't multiplayer, so any games allowing this would be asynchronous.

The system is not "always on" unless a game needs a GM's decisive contribution. Some games could run without being monitored, and may even be offline by player choice.

With a GM session...
In a textual interface, the GM will use his words for communication. Both the story and the GM are expressed through words.

In a graphical interface, the GM 'may' also have a text interface, or a special content interface, but also access to spawning, difficulty, saving extra game states, and -if applicable- overall content management during that session. All the power which a real GM is expected to have is invested in the system.

Without a GM session...
The games would be very similar while running on this system, but the system would not attempt to interact with any player, and any influence would be delivered with subtlety. Possibly no change takes place until after a game session ends.

The system itself has needs to be met before it will achieve proper functionality and mature, much like a human. At which point it will be able to perform GM tasks, silently observe, and deliver new content based on the context that the game provides.

Now it should be clearer what the system really is. Both as a GM that grows from interaction, and as an asynchronous content delivery system for players.

Developers who want to remain in control of their initial game may have the option to do so. Games will have to be programmed to explicitly take advantage of the system, with or without a GM dependency. Also for the first time games will have a nearly human level of intelligence in their design, allowing an emulated GM.

The system could appear to display creativity in the bounds of any connected game, depending somewhat on connection speed.

What can we expect happens in ethereal cloud games? I would really like to talk about this more. There's a lot more to say about the system, and I can't think of a better name yet. I will write about whatever comes to my mind first.

1. http://youtu.be/c-lZe84Nb5E, NextITCorp, July 20, 2012
2. http://www.linkedin.com/groups/Human-Intelligence-Emulation-Call-discussion-127447.S.243756822, Russell S. CEO, June 2013

What is new, what is similar? Comparisons.

Posted by , in Development, Introduction 14 December 2013 - - - - - - · 718 views
crowdsourcing, empirical, concept and 3 more...
I'd like to coin "empirical cloud games," and "ethereal cloud games" as synonyms for "games that evolve from player input." The reason I choose these two phrases is self-evident, plain English.

The games are separate from the system required. In previous and future entries I will need to refer to them both quickly. I will refer to "concept" without specifying often in future and past entries and this may make it a little harder to read, I apologize. Please understand that empirical cloud games can't truly exist without the system, and the system won't function properly without the games, they are separate and reliant, there is a symbiotic relation forming this concept.

I sat down to start reading about human computation games for the time I usually would have typed earlier in the morning. I found a very interesting research paper.* I encourage you to read it if you have any interest in Human Computation games. I may have a cooling off period while I familiarize myself more, by playing them, and determining if any are actually fun (for me). I will be able to differentiate them much easier once I've played. After writing this entry, I will distinguish an important difference.

Here I shall define important differences between existing types of games or other programs and empirical cloud games, games that evolve from player input. I will also talk about the system that is required, and compare it with reality. I would attempt to compare the system with any other applied programs to establish a difference, but I'm lacking a reference, a robot comparison or any one collaborative knowledge database is not nearly close enough, other than as an example about how people interact with it and how complicated the algorithms would be.

Comparisons regarding an Empirical Cloud Game

Games with a purpose (GWAP), human-based computation games are very similar to empirical cloud games. They also serve my purpose as a proof of concept why players will participate playing, and the importance of letting them.

Human based computation games require players to input very specific data, new content must be tailored for the purpose. This heightens the speed of research gained, but it denies freeform from the gameplay. The purpose and game content is limited to the researcher's needs.

Empirical cloud games take advantage of users creating content while playing. In turn, the games are unpredictable and allow human-level evaluation of both actions and storytelling. the content is flexible, allowing player creativity. The games themselves would not all be programmed with a purpose other than entertainment.

Comparisons Regarding the System that will allow Emulation of a Human GM (game master)

Real world game design is my comparison to the system. Both the system and spoken word require storytelling. Games teach the players to perform and improve skills required to be better at the game, and these skills may apply to reality. Concepts evolve over time and new innovations emerge which improves the enjoyability and the effectiveness of the original games.

Someone writing a story about cats that live on the moon would get a very limited opinion about whether it works realistically or as an amusing concept from reading it to different people one at a time, or even to a crowd of people. Either way, only one opinion is available at a time. On a forum, a slightly better range of responses and comments is expected as each individual has a chance to voice as much as they like.

If the system is submitted a story or paragraph about cats that live on the moon, it delivers an educated response based on any references it has. Over time the story is reviewed by every user who encounters it while they play games, reading cat/moon stories that the system connected, users make comments and revisions to any passages they encountered. The story is rewritten to an extent, the rewrites that do not match old content are in turn evaluated. The submitter reads and uses the new information to suit their own needs.

In the real world opinions strongly depend on a greater range of senses other than logic, social standing easily holds as much influence as validity in an argument.
In either world, no one's guaranteed to recognize opinions objectively.
From the system's point of view everyone's submission is equal, and you don't necessarily need to know who you're reading from.

PROTECTION of GOOD and from the BAD
The outlying good (useful) and bad (confusing, enraging) elements are more likely to attract attention.

If a story about cats is representative of a good story many people like to hear, they are more likely to listen and give a response, which in turn may result in the story being retold to them again.
This allows the system to recognize the response as an indication of specific user interest in cats, and many different but similar stories are more likely to be suggested. The story will be rated higher and occur more often.

If a story about dog poo is representative of a bad story very few people like to hear, they are more likely to ignore it and anything very similar will not reach them. If a specific author is coercing the system somehow they'd quickly be blocked and unheard from by a majority, but they'll continue to be as creative as they like and other people who want to hear it will rate it highly.

If some content is not making sense, it will remain unrated, or it is rated low enough it is very rarely suggested or displayed until it gets revised.

In the real world, a person isn't guaranteed to go away and the legal and civil solution to disputes against human annoyances is blocking yourself in away from the source, a self-restricting policy.

* Frontiers of a Paradigm - Exploring Human Computation with Digital Games by Krause et al,

Issues to Consider

Posted by , in Development, Introduction 13 December 2013 - - - - - - · 675 views
crowdsourcing, games, flaws and 1 more...
There are some issues that arise from the concept of "an evolving game" and I'll attempt to address what I think of and any that people suggest. Mostly I think of this as the first go to so I can look back and see if there was anything I write about the concept now that may have changed.

In descending order by fun [concept first, sales last]:

Issue: The system, what's that again, it sounds like a conspiracy theory
The system is just a way to refer to the games, the database, the software bridge between the two, and what players who interact with any game on the system may be connected to. Ok, it sounds like a conspiracy theory about as much as any huge unstoppable force has.
Issue: Inevitable flaws
Every generation of gaming has flaws, players try to cheat. It will happen, and it will be hilarious and inventive. The system will let them. The best discoveries are accidental.
Issue: How this is going to work. A brief explanation.
It all begins by letting users enter text. All it takes for this to function is a program that understands one written language at a time, instead of key words. This continues indefinitely with people reading the text and reevaluating it until the accuracy of the program's output resembles flawless storytelling.

Issue: That last part was way too short, more...
Take a look at Captcha, the developers are geniuses as far as I'm concerned, I don't like to laud anyone on a personal level too often. They already have such a language system budding in Duolingo (but maybe nobody's pointed it out).

Issue: This will never replace existing games. It is an attempt to create a new kind of game.
The system will initially be inaccessible to new developers. There's no way for this to be produced and shipped like a band aid either. If developers wish to maintain a personal artistic license they will keep away, we stil have retro games, some sell, some are still developed, they are never replaced.

Issue: I know fun games will evolve, what about the other games that become abandonware?
Have you ever looked around Wikipedia for things to edit? Competition and the demand to share opinions and facts is integral to being human. Badly written stories and poor graphics are reused every day because they compel some meaning. Even incomplete games can be viewed as such, they are incomplete stories. Any incomplete story remaining in your mind becomes a part of it, the system is no different.

Issue: This idea may not seem new to a reader, it uses crowdsourcing, and some other ideas
Show me the game that already exists and does exactly what I described and not just one of the things at a time. I'll thank you.

Issue: Artificial Intelligence Doomsday Goodbye World
The games evolve, not the core program, no true AI is possible from my comparatively simple plan here. If the core program would actually change it'd quickly stop working, this is fundamental to understanding how programs work. Programs can't learn, programmers tell them what to do and the program repeats it, with variations in input processing time and output.

Issue: User input devices
This isn't integral to the concept, but it's a beginning. The end users input alphanumeric text, they'll upload sound, images, multimedia, anything that can be parsed and reused by the program.

What's important is users input more data in the form of evaluations. This is every day stuff on the internet. But now it has importance.

Issue: Who makes money from the games, the users? Developers? The system?
That's actually a very abstract problem at the moment.

Developers deserve payment for their game, and they need to know exactly how much they'll get.

The game system will need to be funded to maintain expenses.

There are current online applications that allow consenting users to participate and earn money.

Issue: If we make it right now, this instant, without looking ahead. How does this make money?

Obviously it wouldn't, this isn't a get rich quick idea. I wouldn't let someone who's seriously thinking this touch the concept. The initial development would be costly, the crowdsourcing would generate free research, databases would grow. Money is a failing here and there's no estimated time before completion.

Some of the things users have already contributed to crowd sourced games has saved research time, which is equal to money. They never even had to know about it, and they never got a dime back. Examples of crowdsourcing games: Foldit, Galaxy Zoo, cerberusgame.

Issue: Nobody's going to make this right now without validation

I'm not able to sell it to myself.
1. It can be made
2. Someone will pay for it

Issue: Monetization and Customers
Monetization of hard work. The ideas I have for the system produces games. The system itself improves by playing games. How can this be priced correctly? I'll have to break this into a new blog entry anyway.

Monetization of data is better. After it's completed, other companies will have uses for human data, and unless they're into games they won't reproduce it the same way. It will have uses I can't really understand, and I wouldn't need to.

From topic to blog

Posted by , in Introduction 12 December 2013 - - - - - - · 774 views

[Theoretical] Games that evolve from player input
See topic here:
Original title "The most scientific game."

I will write journal entries instead of posting to the topic now. I'll be rewriting some of the content or elaborating. It's mostly for fun, I need something entertaining to write about and avoid repeatedly bumping my own topic at the same time. The entries I make are not based on any intentional research, so I'll be citing few hard facts. They are all creative writing. If someone asks I can only say straight out "It just came to me. An hour of typing later I explained the idea by making some references and started to eat breakfast."

I started the topic partly as an experiment since I was reading copyright law recently. The idea is realistic, it's just not something I see myself programming right now, or even in 20 years. Like writing a new Google search engine, or the Cleverbot that fools people into believing it's human, a Wikipedia sized pool of objective opinions, and memes with the survival rate of a Cheezburger... many levels of software complexity combined with human influence. Only this is once it's finally applied to game development, which will attract essentially the same user base. Heck no, I'm not programming this.

What's the experiment?
First of all I wanted to see if anyone would respond to the topic. Nada. I received zero external input other than observing the idea is plausible (all those things I said last paragraph), and as far as research goes, the reason I started on this is because I heard a lot about what games CAN'T DO.

I'm adamant about this, games CAN DO ANYTHING, it just hasn't been implemented, simulated, or conceptualized correctly, or fast enough yet. So, yeah, even subconsciously I formed an argument before I woke up in the morning.

Some people talk about tabletop games, GM'ing role-play games like Dungeons and Dragons as the de facto limitation on our hard coded games. If you try to do something really clever the programs will not recognize this, it's just not possible and nothing happens. A human GM will think of an explanation why not, at the very least, they may kill you for it. But this limitation remains constant only as long as games are "hard coded". We have people in the game industry trying to reproduce predefined rules of old games, tabletop or retro arcade style, CONSTANTLY. They don't have enough time or experience to reproduce a human GM, nobody does.

Also I want to get this across now, this won't produce a true artificial intelligence. But it does fuel a paradigm, the software will be reprogrammable by non-programmers.

The premise was how games will emulate a GM (I never said that in the initial topic, but I felt this self-explanatory). It can accept just about any input and reuse it, evolving. Taking everyone's ideas and trying them at least once, it is the most scientific game. You begin by creating a database, games are programmed to interact with it and the user. The software that connects the games and software will parse input from the user and store it. As many people who interact with the system at any point in time will have the opportunity to input cause and effect by playing any connected game and disagreeing with what is or is not possible.

This is a system for games that emulates human creativity (the GM) by starting with preset rules and then crowd sourcing data until it appears to understand cause and effect, more than the average human would (it will emulate a GM afterall). The more users who have influenced the system the merrier, more interactions will make it more accurate.

More about what a game will be like, and less about how it does it. When someone who does not know rules to Dungeons and Dragons wants to play, they can load up Dungeons and Dragons from the database once it's programmed in, stay connected to the database that knows all the user input for a more human experience, and let the game system GM their game session. The player needs an input that translates to text at first, future input technology may change this ideal, any data suffices.

From the player's point of view: The game appears to be creative and organized (if desired it could be less organized and make a lot of assumptions and preset decisions to help new players), it accepts the player's information in any order and asks about info that's missing, it'll start the new player out in common situations to teach them the game, it will reuse situations anyone ever thought of if the opportunity arises, it'll know all the rules and then some, it will be able to answer like a human would to any OOC remarks, and it will punish and reward players for doing clever things that weren't even in the rules.

I wrote about the first and second generation of games that appear to logically progress from this concept so far. To develop the second generation you definitely have to have something similar to the first, and so on.

I could use a little help anyway:
Until I learn otherwise this idea's original and entirely mine. Please notify me if someone else thought of this before December 2013, or if you see anyone using this idea in the future.

January 2017 »

15161718 19 2021

Recent Entries

Recent Comments