Game Structure

From CS160 User Interfaces Sp09

Jump to: navigation, search

Lecture on January 28th, 2009

Slides

Contents

Readings

Discussion Questions

1. Chris Crawford's taxonomy of "play activities" (in the Fullerton section on puzzles) distinguishes games as needing a means of "winning" the game. If you agree with this, discuss the role of victory/defeat in motivating the player. If not, pick a game without victory/defeat and discuss what other features of the game might satisfy players.

2. Use the structure of Fullerton's formal elements of games to analyze the three types of GWAP games discussed in Von Ahn & Dabbish. How are these formal elements similar/different across the three types of games? Does this analysis suggest other types of GWAPs that could be created?

3. Are there limits to the kinds of information we can extract using the GWAP-style game structure of Von Ahn and Dabbish?

Karl He - Jan 26, 2009 05:59:33 pm

1. The role of victory and defeat is in achievement. Games need something to achieve, in order to motivate people to play them, and in many cases this is winning. It can also be things such as getting as many points as possible, where the game always ends but with no notion of victory or defeat.

2. The games described by the GWAP article are mostly similar, being mainly cooperation between a pair of players, although there are competitive elements such as the top scores. There are slight variations in the procedures, rules, etc. but the goal of the players is more or less the same, coming to the same conclusion. There can be a large number of different types of GWAPs simply by introducing the idea of conflict, an example might be finding the most efficient walking route on a computer-generated map.

3. It is hard to judge if there are limits, at the time being at least it doesn't seem like similar techniques can be applied to, say, the early detection of leukemia. Yet, there are applications on Facebook such as SuperPoke which encourage the invitation of other people to the application, that show new ideas on what the limits of a game-like interface are.

Shoeb Omar - Jan 26, 2009 07:51:26 pm

1. I do agree with Chris Crawford's taxonomy of "play activities" such as "winning the game" as distinguishing games from puzzles, as a rule of thumb. Of course there are going to be exceptions, but generally the role of victory/defeat is a defining feature of a game. I believe its role in motivating the player is that it appeals to a natural sense of competitiveness within human beings. We're competitive in nature and get a sense of satisfaction in winning. Thus winning in a game is no different from winning in a sport or competition--it gives us a sense of satisfaction we wish to attain more of, thus causing us to play the game more.

2. Players, Objectives, Procedures, Rules, Resources, Conflict, Boundaries, and Outcome are the formal elements presented in Fullerton. Generally all types of games have the same kind of players, conflict (in that there isn't really any), and outcome (in that you do better if you match with your partner more). The objectives are generally the same too--to either have outputs or inputs match or to guess an input from your partner's output. The main differences lie in the procedures--the output matching has both players write descriptive output and matches them, whereas the inversion-problem game feeds the responses of one player to another, sort of flipping the roles of one of the players. Thus although most the general concepts are the same, the semantics of how the game is played have been changed. This analysis does suggest that other types of GWAPs can be created, because if one can find other procedures that cause the output of the game to remain the same, the main objectives and benefits of the game do not change.

3. I think there's clearly limits to the kinds of information we can extract using the GWAP-style game structure of Von Ahn and Dabbish. There is no way to really logically teach the computer how to adapt. Although a GWAP style game could in theory help a computer machine learn how to identify figures when processing images, the applications are limited. We can't, for example, pass in a picture of the Mona Lisa and ask a computer why it is so famous. Although it may learn from an input tagging game ideas about the picture, when questions start moving to ideas or concepts rather than basic facts, it is extremely difficult for the machine to take all of its input from its players and synthesize an accurate feasible answer. It perhaps is possible down the road, but for now, this approach can only be used for really basic identification techniques, not for more complex questions or reasoning.

Denise Ngai - Jan 27, 2009 11:43:05 am

I agree with Chris Crawford's taxonomy of "play activities" that distinguish games as needing a means of "winning." The possibility of victory/defeat motivates a player with regards to how he/she plays the game. Most people who play a game play it to win; they want to reach that final goal or score the highest score. It doesn't matter whether the final outcome of winning provides some type of prize. The feeling of victory itself is motivation enough to give the player a boost and up his/her game. The fear of defeat and losing the game motivates the player in the same way. If a game were to not have some time of victory goal, it could be seen as pointless. In this case, people would have no drive to continue playing the game. Whether it be a card game, educational game, or other game for entertainment, there is always an end to the game and the outcome of the end is either reaching some type of goal (which can be seen as a victory) -- for example, getting rid of all your cards, surviving a shooting game, etc. -- or the outcome is not reaching the goal (defeat). Educational games attempt to teach the player something. Even with these, there is that sense of victory/defeat. Certain educational games don't really have a scoring system. Their final goal is to teach the player what he/she seeks to learn from the game. The player's ability to learn the material can be seen as a victory. The player's inability to learn the material can be seen as a failure. Regardless of what type of game, there is always some type of victory/defeat associated with it.

Chang Su - Jan 27, 2009 03:15:15 pm

  1. Not all games need a means of winning, the children's play of "house" being a counterexample. However, in most games, especially those enjoyed by adults, winning or losing is a feature. The definition of victory or defeat could be absolute, wherein the player achieves an agreed objective or not, or relative, wherein the player tries for a score higher than that of his opponent. Regardless, the role of victory and defeat lies in the inherent competitive desire of humans.
  2. I think there is great similarity among the three GWAP games discussed. Firstly, they all involve two players at a time (though I do not see any reason beyond complexities of implementation why there cannot be more). Secondly, they all have two levels of objectives: for the players, the objectives are about anticipating and/or analyzing the response their partner; for the designers, the objectives are to use human intelligence to accomplish certain tasks unachievable by machines. Lastly, they do not introduce conflict between partnering players, but instead encourages cooperation by setting one common goal for both players. This definitely suggests more types of GWAPs that may be created.
  3. Currently, the limitation of GWAPs lies in the "recognition level" of the subject. That is to say, most of the players have to recognize the sound of a trumpet in Tag a Tune, or the meaning of the word "telephone" in Verbosity, or the image of Eiffel Tower in the ESP Game for these games to work. Non-common-sense subjects will be poorly handled. Also, it is worth thinking whether majority always wins. For example, if 80% of Verbosity players thinks that "tomato" is a fruit, does that mean they are right?

Chunwei Lai - Jan 27, 2009 08:08:52 pm

  1. Games have progressed from focused on winning to include more focus on cooperative and competitive play. Though winning is still the ultimate goal, there are now much more smaller goals such as trophy/collections/unlockables/etc that one can try to obtain without having to win. Winning gives a sense of satisfaction to the player and the feeling that the time and effort spent yielded some substantial result. Losing is a motivation for improving or at the very least change. There are some games where the goal may not be to win, some of those include shooters like Call of Duty - World at Wars which includes a game mode where the goal is to survive waves and waves of zombie. In such a case, players get satisfaction from not winning but obtaining a higher goal.
  2. Other GWAP-style games can be created focusing on the cooperative element between players. Games that were formerly meant for a single players can be played between two to gain a new experience. As goals in games grow complicated, how players interact will change greatly yielding new play-styles and goals.
  3. Due to current limitations of technology, the amount of information one can extract from GWAP-style game structure. To be able to do so would require a much more in depth understanding of how the human thinks since the computer is much more goal/task oriented, vague objectives are hard/impossible for computers to complete (at least from the human's POV).

Chris Thompson - Jan 27, 2009 07:28:49 pm

1) I don't think "winning" is a prerequisite for all games. A few things spring to mind, but the easiest to relate would be Tetris. Yes, there is a definite way to lose (your pieces go out the top of the screen), but there is no defined way to win. Some versions of Tetris end the game after a certain number of lines have been successfully cleared -- when one such version of Tetris surprised me mid-game with a victory, I was offended and enraged. Luckily, after "beating" it for the first time, I was able to turn that function off. I continued to play the game because the real appeal is, besides entering the stasis like trance that occurs once an attentive action becomes second nature, there is a mild feeling of accomplishment (I'm getting better reflexes by playing, supposedly) and I wish to test my endurance to maintain a difficult activity. There is also some element of self competition, though, as I like to see how many lines I can clear before dying. [But when I realized my most recent version of Tetris stopped counting line clears after 999 due to developer laziness, I was incised. I haven't touched that Tetris since.]

2) All of the GWAPs have players (unless of course the game is unsuccessful), objectives, procedures, rules, resources, boundaries, and an outcome, most of which are provided by the nature of their digital implementation. They also have conflict of the benign sort (trying to guess what somebody else is thinking), but they're lacking in the competitive style of conflict since both (and in these cases, there were only two) of the players are working toward the same goal. The high score boards allow you to indirectly compete against anonymous others, but you're always working in cahoots with your temporary partner; never competing against them.

3) All three styles of GWAPs in VA&D are limited to objective, concrete results. Each of the games tries to come to some mutual conclusion about an already existing entity, which is useful in identifying information about existing items, but does not allow for the creation of new innovations. None of the games allow users to create something that previously didn't exist, such as collaborative art (like that of www.drawball.com, which I would consider a GWAP with the purpose of creating art as well as offering surprising implicit insight into internet communities).

Stephen Wu - Jan 27, 2009 08:14:17 pm

  1. While many games do have some notion of victory and defeat, quite a few popular ones do not. The most prominent example would be sandbox games, where players simply spend time exploring the game world and what is capable within the game engine. The Sim City games are probably the most famous example of this class of games. There is no definite end to the game, and players can keep playing until they lose interest in their city. While there is a sense of positive reinforcement in terms of money and unlocked improvements, the game is really what the players make of it. Where some may have the goal of simply getting their city to a certain size, others try to get a certian look and function, and the very ambitious try to recreate real cities tile for tile, as seen here: http://www.simtropolis.com/forum/messageview.cfm?catid=36&threadid=37869 So while games do need to elicit a sense of accomplishment from the player to keep them playing, some games are implemented in such a way that the player can have fun exploring possibilities within a framework.
  2. GWAP games presented by Von Ahn and Dabbish, all have the same objective of using diffuse man power to solve complicated problems computers are not yet capable of solving autonomously. The GWAP games presented are for two players who cooperatively try to label objects. Output-agreement, inversion problem, and input-agreement are merely different rulesets towards that objective. Given that so far all GWAPs are cooperative in nature, there is no reason why broader classes cannot be created when the entire spectrum of competitive games have been ignored. For example, I've played a game created by university researchers where players are given a virtual army with a blank slate AI which they can program and train using some general move, retreat, and shoot commands and a reward/punishment system. As players compete to create the best personal army AI to fight with other player armies, they are also generating useful rules for AI progamming that can be used in other games. I would classify that game as a GWAP even though it does not match one of the provided three models.
  3. I think GWAPs are limited by a two things, the game designer's ability to frame the question and whether or not there exists a playerbase capable of solving that question. While it would be difficult to imagine a GWAP that gets strangers to do your taxes for you, one can imagine a hypothetical Tax Hero game which is so fun that the users learn the tax code to play it. However, using the 3 systems presented by Von Ahn and Dabbish, we are limited to problems to which two users can agree on an objective answer to. The system would never be able to reliably answer subjective questions such as which cat is the cutest.

Colin Downs-Razouk - Jan 27, 2009 08:47:24 pm

1. I disagree that games need a means of "winning". In some games success is quantitative, like a score or a time. One example is guitar hero, or rock band. In Guitar Hero, there is no clear cut situation in which you win. You can perform a certain song at a certain difficulty level, and you can receive a high score for performing that song. A similar situation is a time trial in a racing game. When racing against an opponent, there is a winner (1st place) and there are loser(s) (people not in 1st place). When racing a time trial, there may or may not be a time to beat. In these situations, players gains satisfaction in the fact that they can improve their scores or times, and that their improvement is recorded somehow.


2. In the three games described the players are the same, the resources are arbitrary, and the conflict is the same. The boundaries are different for the different games: in input agreement games you see the output, in output agreement games you see the input, and in inversion-problem games you may see the input or messages from the other player. The procedures differ with the three types of games, as necessarily do the rules (although I cant tell the difference between rules and boundaries.) The objectives have subtle differences between the games. Also the outcome of the games depends on how you implement them. Other GWAP's that could be created are those that vary (with the implementation) the outcome of playing the game. You could have people try to get the largest number of matching outputs/inputs or the shortest amount of time that it takes to match one output/input.


3. There are certainly limits to the kinds of information you can extract using GWAPs. You cant extract information that is beyond most people's comprehension. If you ask people to find the solution to some differential equation (which might even be difficult for a computer), most people will not be able to do that. You have to tailor your GWAPs toward the things that humans can do better than computers (like vision.)

Siddharth Shah - Jan 27, 2009 09:34:49 pm

1. I agree that in general, games are primarily focused with winning. This separates them from puzzles, which have little replay value; once you have mastered a puzzle, there is no need to play it again because you have already "beaten" it. However, games DO have replay value because beating one opponent is not the same as beating another; in each case, you have to prove that you can beat that particular opponent. An exception to this rule is the growing segment of games with a purpose. For example, typing games help the player learn to type faster and more accurately, although there is no real concept of "winning" anything. In such cases, the primary purpose of the game is to improve the player's skill; the concept of winning is secondary.

2. All three game types seemed to involve similar formal elements. They attempt to get different people to work together to produce an output that is either difficult or impossible for a computer to produce on its own. The conflict is somewhat ill-defined in that the players are always working together, rather than against each other, so there is no major conflict between them.

3. Yes, I think there are limits to the kinds of information we can get from the GWAP-style game structure. First off, we are limited by the amount of knowledge that the players possess; they cannot produce an answer that they don't individually have. Also, the answer selected by the majority is not always right. There are also issues regarding the backgrounds and beliefs of the players; people of background A might give different but equally accurate answers as people of background B, making it impossible to impartially choose which group is actually correct. Basically, it will be difficult to quantify results from any game that is not primarily fact- or skill-based.

Saung Li - Jan 27, 2009 10:58:27 pm

1. Having the objective of winning is a big factor in motivating the player to tackle the conflicts of the game, and I would say most good games use this factor. People don't want to lose to others so they are motivated to play well. However, some games can bring satisfaction to the player without winning. Games like Asteroid or other games in which the player needs to survive as long as possible to earn high scores can still entertain the player. Games like Sim City don't really have an objective but are still fun because they spark the player's imagination.

2. Pretty much all of Fullerton's formal elements are in GWAPs. Different objectives, conflicts, players, etc., can all be used to make different GWAPs successful. For example, GWAPs could make players either cooperate or compete with each other. Each game has different variations of the formal elements. These variations target different interests. For example, the GWAPs mentioned are mostly cooperative, and people who prefer competitive gaming may likely turn down these games.

3. We can't, of course, extract information that humans themselves don't know about. Sometimes only a relatively few people have knowledge on a particular subject matter, so the GWAP would have a harder time finding a pool of players. If there are no limits to the kinds of information we can extract, there certainly is a limit to the amount. Humans have so much knowledge and can do so many things that it may be impossible to extract everything due to a limit on human resources.

Adit Dalvi - Jan 27, 2009 10:52:16 pm

1. I do agree with Chris Crawford’s taxonomy of “play activities”. If we look at victory as a sense of achieving something and being satisfied with the achievement, then it is an essential component of any game. People are often driven to achieve goals in all their pursuits in life, whether these pursuits are personal or professional. When playing a game, there is no difference. This is why games must have some goals that players can achieve in order to give them a feeling of satisfaction when they achieve that goal. If games lack this important component, they are usually viewed as a waste of time and have no value at all.

2. The three types of GWAP games are very similar in the number of players (and the types of players) that can play i.e. they both choose random individuals from among all potential players. They are also very similar in their objectives in that there must be some form of coherence between the players whether it is with respect to the input or the output. Their outcomes are similar as well in that the people find enjoyment in playing and important information is obtained for implementing solutions to problems that are hard to solve for computers. The differences in the games are in their procedures, rules, resources and boundaries i.e. the main elements of game playing itself. This is clearly evident from the way the games. For example, in output-agreement and input-agreement games, the players must produce outputs based on inputs while in inversion-problem games, one player produces output based on input and the other player must produce the original input based on the first player’s output. This suggests that if the games differ in some aspects from one another, then creating more games is simply a matter of creating different forms for the differing aspects.

3. The GWAP-style game structure relies on people to supply information. If we were to only consider the collection of information, then it is theoretically possible to obtain limitless information from all the people in the world through the use of GWAP games. However, the information we would obtain would be facts known to people and we would not be able to obtain any information outside of this. Also, we would not be able to answer questions which don’t have a definite answer simply due to the inability of computers to think on the levels of humans.

Derek Liu - Jan 27, 2009 09:15:57 pm

1. The concept of victory and defeat is relevant in almost any game we play. Victory provides a clear goal for the player to work toward as an ultimate reward while defeat is the penalty for the player's failure. There are, however, games that do not have any real concept of victory and are simply wave based, Chunwei has already pointed out the Nazi Zombie mode in Call of Duty: World at War as an example. There is also the question of what constitutes a game. We can use many different criteria to evaluate if something is or is not a game. A game could be something as simple as playing on the monkey bars or on a swing set depending on our definition of a game, these playground games in particular do not have any real concept of victory or defeat, they are simply there for fun.

2. The three GWAPs discussed were output agreement games, inversion problem games, and input agreement games. Each of these three types of games have players, objectives, procedures, rules, resources, conflicts, boundaries, and outcomes. Objectives are similar in that the main objective across the three types of games are to generate some sort of matching between the two players. Procedures and rules differ across the three because of the layout of the game. Output agreement games require both players to generate descriptions of a picture until they match whereas the inversion problem requires one player to create a description for a player to guess the object similar to Pictionary. The input agreement game works in a similar fashion to the output agreement game but also differs in procedure. The output agreement game requires players to match on outputs whereas the input agreement game requires players to determine whether they have been given the same input. This analysis does suggest that other types of GWAPs could be created simply by altering any of the formal elements of games to a certain degree.

3. There are definitely limits to the kinds of information we can extract using the GWAP-style game structure. The GWAP-style game structure only allows for simple applications such as how to improve the results we get from an image search engine. The GWAP-style game does, however, allow psychological insight into the every day internet user's mind.

Stephanie Shih - Jan 27, 2009 10:17:56 pm

I think, given the way the article defined a game - as an interactive activity bound by a set of rules - of course the end goal of a game would be about winning in some definition or another. However, in some games, people may not necessarily play with that end goal in mind - take MMORPGs, for example, where while the end goal may be to become stronger and finish all the quests and reach the end of the storyline, the social aspect of an MMO can also mean that people will use an MMO as little more than a platform for meeting others. Victory/defeat can come in the form of PvP, getting the best equipment, etc., but it is not the only aspect of an MMO.

In terms of GWAPs discussed in Von Ahn & Dabbish, though, it very much fits into the definition Fullerton gives of a game, where the players driven towards victory/defeat, and it is played between two people who are interacting with something for their own enjoyment. The two people are encouraged to work together for a common victory, but there is still a victory to be gained. There are purposeful constraints that make it challenging enough for someone to keep at the GWAPs, over and over. It does suggest that other types of GWAPs could be made, by breaking out of this common mold.

Sum Sum Wong - Jan 27, 2009 10:57:01 pm

1. For most games there will be winners and losers but not all. Tetris is a good example. Another example would be the ESP game introduced in the first lecture. If we follow Chris Crawford's idea, Tetris and ESP game can only be "puzzle" but obviously they are games. Doubtlessly, pursuit of winning attracts people to a game, however, there are also other ways to satisfy players. People playing Tetris can see how themselves getting better and better after long time of playing which is another way of satisfaction. ESP game on the other hand attracts people by providing them feeling of cooperation.

2. I think the 3 GWAPs are generally similar to each other. They both aim to similar group of players and have same number of players involved each time. Also, the players cooperate with each other rather than play against in all three games above. The objective and procedure of the games are very much similar too.

3. I believe there will be limitations. Information extracted would come from only 2 people, anything that is out of either one of them's knowledge(professional) would not be recognized. Also, what the 2 people come up with might not be the correct and meaningful answer. There are also photoes which can be easily misunderstood, the keywords provided by the players could be wrong. As a result when people see the tag, they'll follow what the players provided and misunderstand the photo.

Timofey Titov - Jan 28, 2009 12:12:51 am

1) I agree that games need a means to "winning". This motivates the player in both victory and defeat. In the case of defeat the player gets a sense of satisfaction and achievement. This can also be reinforced through high-score lists as mentioned by von Ahn & Dabbish. Defeat on the other hand does not completely discourage the player, because the game is after all a small world. It makes the player try again to succeed, thus resulting in more play.

2) The similarities are in Objective, Conflict and Outcome. Two players compete against each other with the objective of solving a small task. The outcome is uncertain which creates suspense. The procedures or the means to victory are different. There seems to be a lack of Resources among GWAP games. GWAPs seem to be very trivial and almost too narrowly focused. Maybe one could integrate a GWAP as a subgame inside a bigger, more worthwhile game.

3) The "bite-size nature" together with a small time-limit is a problem. GWAPs can be seen metaphorically as parallel execution of a bigger task, where people work on primitive pieces of data. Tasks that can't be broken up easily into finitely small pieces can't benefit from GWAPs.

Sean Hansen - Jan 27, 2009 11:26:32 pm

1. Games do need victory/defeat conditions, but they need not be inherent to the game. Where none are provided, players will make their own goals. Depending on the game and the player, these will range from achieving a certain score to collecting in-game items to simply doing something better than someone else playing the game to accumulating real-life wealth via the game. Of course, new game concepts are always being tried, and goal-defying games may well be in the future (referring to the upcoming ps3 game Flower)

2. The GWAP models described are largely identical. They all have players randomly matched by the game attempting a description of a provided input and being scored on their ability to agree. Such a new type of game being limited in diversity is to be expected, though; innovation will come with time and interest.

3. The only limit is the boundaries of human knowledge. Just look at wikipedia: an interface with no positive feedback or provided incentives, but it now has information on practically every subject known to man. If a charity service can be so successful in accumulating data, games can do better.

Carol Chen - Jan 28, 2009 12:16:52 am

There are limits to the kinds of information that can be extracted using the GWAP-style game structure of Van Ahn and Dabbish. For one thing, as they mention, the "bite-size" structure of these games appeals to casual gamers; this constrains the content to simpler inputs and outputs. However, attracting participants who will enjoy and continue playing a game is important, so this is not to say that the games should be designed to be too complicated. GWAP-style games also tend to be more about identifying the 'what' of ideas, concepts, things, and images or videos of things, or about capturing related common-sense ideas. They don't work well to identify the 'why' or the history of inputs or outputs that is beyond the lowest common denominator of knowledge. For example, it would be difficult to capture the cultural knowledge and history of why Japanese people working in Japan feel obligated to be the last one to leave the office, even if they are not being productive. This leads me to my last point, which is that these games, much like Google's PageRank algorithm, rely on the majority's point of view. Not only does this quelch creative and unique viewpoints in favor of 'correctness', but it allows for the propagation of certain misinformation.

Rohan Dhaimade - Jan 28, 2009 12:36:01 am

1) The most obvious games that have no "winning/losing" are MMORPG. The benefits of MMORPG are mostly the social aspect of the game. You go and accomplish goals along with other individuals who are playing. You also compete and attempt to become the "strongest" if you so desire. You can also obtain many things such as fame, power, money, and other things which is more akin to real life. The overall "game" of a MMORPG is really more of a structure that has within it a set of complex and smaller games that might have victory. Defeating an opponent in player vs Player combat or finishing quests can be considered victories. There also types of "boss battles" such as the Lich King in World of Warcraft, that could be considered victories. These aspects though are less substantial than the social aspects though. I don't think winning is the only motivation and there might not be a clear "win" defined in the game.

2) All the games have two players, chosen randomly by the game itself to compete or cooperate in the game. The objective is to somehow get coherence between the two player's choices. The two player games lack a physical conflict, the conflict in the games is the ability of the two players to be able to obtain coherence in their answers. Procedure, rules, and boundaries are different in the games. The main differences between the games are who is guessing input, output, and how much transparency there exists between the two. In addition, there might be time limits, scores, and other mechanisms in order to determine "victory" that might differ across all the games. The outcome of the game might not in turn be a clearly defined victory, there might not be a way to "win" the game. The outcome of the game might differ based on the mechanism for rewarding good behavior would be in the individual game. Though all the games produce some beneficial output.

3) You obviously are limited by the fact GWAP is only useful for problems that are easily solvable by humans. Protein Folding is incredibly complex problem for computers but it even harder for humans to solve

Mark Dhillon - Jan 28, 2009 12:26:53 am

1) The role of victory/defeat is extremely crucial in motivating the player. Without a sense of this, most games would end whenever the players lose interest and decide to do something else; most games would go unplayed as a result. Furthermore, the experience of loss can also motivate a player to prepare better for the next match up, and propel him or her towards winning. I believe that winning and losing are an integral part of the human experience, and by dealing with those roles we push ourselves to achieve.

2) The GWAPs discussed fit nicely within Fullerton's formal elements of a game. There are clear objectives, rules, procedures, etc. I agree with Karl He that the element of conflict being introduced to a GWAP could help distinguish more gametypes within the genre.

3) I believe so, mostly because you have to establish a consensus between players. I just played the google image game, and I thought that some of the words I suggested were really good descriptions of the image. However, you only progress when both players agree on a word. Perhaps a bit more creative freedom with accepted answers and overall objective would help.

PS Tomatoes are definitely fruit, Chang.

Henry Sanjaya - Jan 28, 2009 12:51:53 am

1. I agree that games need to have a winning/losing element in it. Winning a game will then be a very rewarding objective to the players and it will motivate the players more. In a puzzle, for example, the player wins if the player is able to solve it. The sense of satisfaction of winning the game motivates the player to play in the first place. Furthermore, games that have winning/losing element are generally games that have the uncertainty element in them and also difficulty. As is described in the text, uncertainty adds to the thrill of playing the game while the difficulty creates satisfaction if the player can solve them or temporary frustration if the player has to think more to solve it. All this contributes to motivating the player to complete the game and for the players to see the game as a meaningful and fun game, not a pointless one.

2. The GWAP games have very similar objective, which is mainly to let the players have fun while labeling items that computers cannot yet identify using current technology. The only difference between the three GWAP games are the rules and procedures, where they are output agreement, inversion problem, and input agreement respectively. Other types of GWAPs can be created by simply modifying one of the formal elements of the game. For example, conflict can be introduced into the game so that the person with the least amount of time that suggest the most guesses have more points and the top scorers are winners.

3. Yes, there are limits. One limit is that GWAPs only works for the simplest of games, where the 'answers' or solutions of the game is objective rather than subjective. Imagine trying to give 2 images of different pair of shoes and asking which one is cooler than the other. Extracted information from this kind of GWAP can be unreliable and mostly subjective, and convergence of answers from players is very unlikely. The second limitation is human knowledge. GWAP can only be useful if humans can actually do the game or it is within the scope of their minds. Otherwise the GWAP may not result in being an entertaining activity for humans, let alone be advantageous for gathering data.

Yin-Zen "Johnny" Hwang - Jan 27, 2009 11:32:49 pm

1. Chris Crawford's taxonomy of "play activities" (in the Fullerton section on puzzles) distinguishes games as needing a means of "winning" the game. If you agree with this, discuss the role of victory/defeat in motivating the player. If not, pick a game without victory/defeat and discuss what other features of the game might satisfy players. However, a game can have states where one state is superior to another, in stacks of infinite states such that while one can advance one's states, one doesn't really reach a "victory" state.

I don't agree with it. The one game without a role of victory/defeat would be SimCity and other empire-building games. What satisfies players is that the city evolves through time so that it looks different, even though all you're doing is the same few combinations of moves. Ooh Fullerton mentions WoW! awesome!

2. Use the structure of Fullerton's formal elements of games to analyze the three types of GWAP games discussed in Von Ahn & Dabbish. How are these formal elements similar/different across the three types of games? Does this analysis suggest other types of GWAPs that could be created?

for the picture game, the constraints are time limits, have to enter the same output as a partner, and cannot use taboo words. winning condition is based on number of matches. for the pictionary game, the constraints are time limits, one partner must make it easy for other partner to guess, other partner must be good at guessing, and there are also taboo words. winning condition is based on number of matches. for the music game, the constraints are time limits, have to be able to guess if the song is same or different through tagging. winning condition is based on number of matches.

these are all similar in having time limits, a way to compute a score, and a focus on collaboration in order to win. with these formulas, one can insert other game rules, come up with other matching medium, to come up with other GWAPS, like one that tags short video clips.

3. Are there limits to the kinds of information we can extract using the GWAP-style game structure of Von Ahn and Dabbish?

Depends on the kind of information. If it's explicit information such as learning facts, like learning explicit arithmetic, etc, then there's only so much variation in the database and eventually gets predictable and boring. If what's being learned is more subtle or more unpredictable, like how to work in teams or tagging photos, then the gaming limit is much less visible.

Cuong Ngo - Jan 28, 2009 01:34:06 am

1. I couldn't agree more on the important role of victory and defeat in games. Psychologically, it is the mission-accomplished feeling that motivates us the most. The more challenging, the more games can engage the player. However, this doesn't mean that games have to be too hard to beat, which is because "puzzles that are too hard are discouraging." as Fullerton pointed out.

2. Fullerton's formal elements (e.g. players, objectives, procedures, rules, resources, conflict, boundaries, outcome) are fully incorporated into the three types of GWAP games. For instance, the outcome of the output-agreement game is an output based on an input given to players, and the objective is to come up with the same output. In general, the formal elements are similar across the three types of GWAPs. Fullerton's analysis suggests that other types of GWAPs could be created with the elements.

3. I don't think there are limits to the kinds of information that can be extracted using the GWAP structure. An example is wikipedia. The amount of information is growing at a rapid rate thanks to the contribution of collaborators from around the world. We can easily find information on everything.

Anatol Tsang - Jan 28, 2009 01:37:25 am

1. A game must have a "victory" of some sort. If it doesn't have a "victory", then it may have a sense of "improvement" (e.g. leveling in RPG games). The point of the game is victory- a player plays either to attempt to reach this climax point or just to improve their skill that would allow the player to reach the victory point easier. A game is intensely more satisfying with a sense of victory or a sense of improving oneself as a way to reach victory easier.

2. The three types of GWAP games are very similar in many aspects. The incentive is similar: match inputs, outputs, or input-output pairs. The players are random people online. The challenge is to do the task better than other people on the web. And the rules prevent the players from simply collaborating (in fact, the rules are to make the game useful). These three types of games seem to differ only in the details.

3. Obviously, the only type of information we can extract is with the help of humans. This, however, is the only limit I see. I think that almost any information can be extracted given a clever implementation. Asking if a problem could be solved using games is like asking if a faster algorithm exists. It depends on the situation and the ingenuity of the computer scientist.

Eric Hernandez - Jan 27, 2009 11:53:57 pm

1. As was discussed in the article, the act of traditional act of a player winning to end the game need not be part of the definition of a game. The counter-example to this claim was also given in the article; World of Warcraft does not have an actual ending, but players still receive moments of accomplishment through either leveling up or obtaining a new item.

2. The objectives, or "Winning Conditions" change for each game, the procedures are all different based on what data needs to be obtained from the players, the resources change in that sometimes players can share information but not in other game types, and clearly, by the definition of the "Rules" given in the VA&D article, the rules of the games change.

3. Excluding the limitations in human knowledge and ingenuity, and coming from the perspective of an engineer trying to obtain training data, it would seem as though a game could be adapted to fit into one of the three types of games that VA&G describe. In other words, it would seem possible to obtain pretty much any sort of human knowledge through one of these three means. For example, in the worst case, you could have two people take a question as input and answer it until they both generate a consistent answer, which is a sort of "complete" game in that it could be adapted to collect any sort of training data, and also falls into the "Output Agreement" game type.

Ian Hildreth - Jan 28, 2009 12:11:07 am

1. I agree that in general games need some notion of winning, or determining a victor for people to play and be interested in them. Most people are competitive by nature and find pure enjoyment in winning, whether accomplishing a mutual goal or defeating other competitors, winning is generally the sole motivation. I believe there are some people who would play games without a clear cut winner, maybe to benefit others, but in general people lust for victory.

2. In general, the three types of GWAP games are similar in many aspects. They all consist of players competing online, where the objective to to match in pairs the output or input of the other user playing. The incentive to play comes from completing the assigned tasks faster than the other players on the internet. Fullerton does suggest that there are other types of GWAPs that can arise, if there was mixing and matching of the elements of the game type, ie. conflict, boundaries, objectives, outcomes, etc.

3. There are limits to the things we extract, in specific settings, like the amount of time it takes, or the format of which the information is presented. In general, however, I believe the amount of things are limitless, as long as it is set up in the right manor, just some games are harder to set up than others. If given a motivation, such as defeating online opponents, people will play any game.

Andrew Shu - Jan 28, 2009 02:06:16 am

  1. There seems to agreement in one class of games that are not bounded by having a strict winning condition. Whether playing with dolls, mmorpg's, second life (which is officially NOT a game) or sim city, the goal of the game is one of exploration. Even puzzles such as sudoku, crosswords, rubix cube, and jigsaw puzzles have a can-you-complete-this win condition. One that comes very close to not having a win condition is the flash game where you're given 7 letters, and you try to make as many 3+ letters words as possible. For most people, getting all the words is so hard, it may as well be unreachable. At best, it's a distant goal. It was hugely popular because the player tries to beat their previous best. And lastly, I would argue that GTA series don't have a clearly specified goal for winning. There's a story mode, but it's just not as fun.
  2. GWAPs are constrained to needing redundancy to check the validity. Thus, players, objectives, conflict, boundaries, and outcome are similar. It is the implementation--the procedures and the resources that are most different. The varying aspects are: the types of multimedia to be used, and who gives what information when.
  3. Boring and difficult information will be have difficulty appealing to the masses. Basically, I'd say that the bottleneck appears at how well the information appeals to the masses. After all, the power of these games is the mass crowdsourcing effect. Difficulty of information may not be too difficult. You might be able to entice mathematicians to play a game to solve random math catastrophes for the benefit of mankind.

Kevin Huey - Jan 28, 2009 02:09:17 am

There are always going to be exceptions every now and then, but generally games have the overall result of declaring winners and losers. Just the idea of a challenge against a computer or other humans is enough motivation for many people to engage in these types of games. The thought of winning satisfies and placates the mind, additionally acknowledging that you are indeed better at something for that one instance.

In general GWAPs differ from the games described in Fullerton in the conflict aspect. There is a lack of intense competition among the participants of a GWAP. But this only enhances the GWAP's effectiveness because everyone is trying to learn something. It does no good to discourage another person from "winning."

There may be limits, or there may not be. It all depends on the technology we have now and in the future. There are always some things that a computer cannot do, but due to the state of the computer system between 10 years ago and now, it isn't right to rule out anything just yet. These games are tapping into the psychology of a human mind right now, so who's to say we can't go far and beyond the endless possibilities?

Ling Chen - Jan 28, 2009 02:25:42 am

1. Winning is a big part of a game and it does bring a certain level of satisfaction. However, I do not believe that a mean of winning is a must to distinguish games. Some games are meant to be simply entertaining or engaging. Games that one can play on his/her own, for example like Minesweeper, seeks more for a sense of satisfaction rather than winning anybody. In games like Ragnarok Online, players also seek to better their skills and to accomplish task to advance levels. There is no real sense of "winning" in such games.

2. There are similarities among the three GWAP games mentioned: output agreement games, inversion-problem games, and input-agreement games. We definitely see cooperation between the players in trying to accomplish certain objectives they have. They all follow some kind of procedures and rules. They also set some kind of boundaries and resources.

3. I think there are limits to the kind of information we can extract using the GWAP-style game structure. The limits may be stretched as our technology advances, but I just think you can only do so much with machines. Games might be good for extracting objectives factual information, but when it comes subjective feelings and thoughts, it's not as easy (maybe it will take a clever designer, who knows?).

Jeffrey Patzer - Jan 28, 2009 02:25:18 am

1. I whole-heartedly agree with the idea that winning a game motivates a player. Unless one is trying to deliberately lose a game to make their opponent feel better, then they are trying to win. When one plays a game they take part in a form of entertainment that in order to derive satisfaction from they need to do well. Doing well means winning the game or coming close to doing so.

2. The three games mentioned all draw on the same premise to keep their player engaged. They have two players and use similar objectives to entertain the player. Their resources, rules, and procedures are generally the same but with modified content. Since the three games in question are so similar at the moment this means that there is a large realm for creating new GWAPs. I think that finding a way to integrate more popular game types into GWAP's would prove very successful. Games like Super Smash Bros. or Halo have been extremely successful due to the party-game nature and I think that finding a way for GWAP's to harness this type of game-play would be spectacular.

3. Definitely. The limits however are only limited by one's imagination. This is both the designer of the game and the player. If both are creative and can make good use of their imagination then the type of information that could be extracted by GWAPs could be almost limitless. The type of information that can be extracted is limited to how the game is designed and what information the game is looking to extract. Therefore, the limit is going to be most heavily influenced the designer's choice of information to extract.

Jason Lo - Jan 28, 2009 12:42:37 am

1. I disagree that you must have a winning state to be a game. Some games are based on the challenge of getting a high score, like winterbells or Tetris. I think people enjoy these games to test their own skills and improve them. Comparing with friends may also be another challenge, but it isn't one I have find especially motivating.

2. The three types of games, input, output, and inversion are all fairly similar in their setup of 2 players who are unfamiliar with each other. The conflict arises from the challenge of 2 players trying to achieve the object of matching an unknown variable. The procedures define whether the guessing format and input follows the method of input, output, or inversion. Inversion is slightly different in setup, but the general conflict and rules are similar. All these games are based on two strangers meeting and trying to align their thinking, but approach from different angles.

3. The information that can be collected from these games only makes sense for tasks that rely on humans like gathering opinions or thoughts and tasks that humans naturally do, but are difficult to describe with an algorithm.

Nalditya Kusuma - Jan 28, 2009 02:48:20 am

[1] I agree with Crawford; a game needs an end-state in which either the player wins or loses. The role of the need to winning a game is to create tensions between players so that they are motivated to do their best to beat their opponent(s), and also to create a clear/satisfying end of game to both players (either win or lose).

[2] Mostly the 3 types of GWAPs are similar except the way the objectives, procedures , rules , resources , conflict , boundaries are defined. Similarity examples: Players are chosen randomly and the outcome is uncertain. Difference example in rules/objectives: 

*Output-Agreement: all players work together to produce the same output.
*Inversion-Problem: all players work together to guess the input and to win the game.
*Input-Agreement: all players work together to determine whether the inputs are same to win the game.

More types of GWAPs can be created since the 3 types above are collaborative games; there may be competitive games that have not been thought of by the gamemakers coming up in the future.


[3] Assuming the users know the answer to the tasks asked by the game and there is a way to verify user's correctness, I believe the only limit is the gamemakers' imaginations/creativities. I believe the reasons why it's difficult to develop GWAPs right now is the lack of technology development and Artificial Intelligence. These 2 factors limit the kinds of information Games w/ purposes can extract. For example , with awesome AI algorithms and super-clear microphone , there could be GWAPs that can learn by itself how to pronounce words based on user's voice-input given some text.

Szu-Chun Mao - Jan 28, 2009 02:37:49 am

1. I would agree with how Chris Crawford distinguishes games from puzzles and toys. It is not perfect how he defines games, however it gives a clear differentiation between games and puzzles and this concept is suitable to the most games. The role of victory/defeat triggers players to seek out for a sense of satisfaction and achievements which comes from “winning” the game. It also helps you understand how your current level of skills stacks up to the difficulty of the game.

2. The eight formal elements defined by Fullerton are closely relative to three types of GWAP games discussed in Von Ahn and Dabbish. Most elements are similar in those three types, the main difference lies in how the procedures are performed in those games. Therefore, this implies that other types of GWAPs could be created if new procedure ideas are found and can be implemented.

3. Using the GWAP-style game structure of Von Ahn and Dabbish definitely has its limitation. A correct answer don’t always come up from majority of the players’ opinions and human beings are not perfect, we make all kind of mistakes even when we know the solution. Designers also have limitation on what and how the information can be extracted.

Sean Kim - Jan 28, 2009 03:10:14 am

  • There are lots of games without achieving a victory in game industry. the presentative examples are the games related with music such as DanceDanceRevolution and GuitarHero. The games give score as the result to players, but they generally enjoy the play itself. The players satisfy the fact that they can enjoy the music in the different way.
  • The three types of GWAP games by Von Ahn & Dabbish; output agreement games, inversion problem games, and input agreement games are similar to each other related with players, objectives, conflict, and boundaries. However, since they followed different rules, their procedures and outcomes are opposite to others, according to the type of the game.
  • Von Ahn and Dabbish focused on inputs and outputs related with players more than the procedures. I think that the satisfaction from the procedure itself can be considered more significant. As I already mentioned in question 1, players can get the most satisfaction from the game while playing it, not by achieving something.

Anjana Dasu - Jan 28, 2009 03:17:06 am

1. Crawford suggests that games must involve "opposing players' who acknowledge and respond to one another's actions" and each player's goal is "to win." Hence, the desire for victory fuels game play and becomes a defining factor of a game. However, I would argue that this feature applies to puzzles as well. The single player is trying to fulfill and, in some sense, defeat the constraints of the puzzle. On a side note, another issue I found with Crawford's taxonomy was the statement that "puzzles have little replay value." If this is true, why do people do the crossword daily or complete entire sudoku books? People who complete crossword books feel victorious as they move from the easy puzzles at the beginning of the book to the harder ones at the end. They are winning not only against the puzzles, but also outperforming their own selves.

2. All three games involve 2 randomly paired players (or a single player and an automated player). The actual objective of each of these GWAPs is to "teach" the computer-- to collect information; however, the players are presented with a different objective-- to provide some sort of general knowledge. The procedures may be slightly different for each of the three games (one is an input matching game, another is an association answer game, and the third is an output matching game), but they all have rules/boundaries that prevent cheating and generate a high enough level of difficulty to sustain interest. The resource at disposal for the three games is, naturally, the human brain. Interestingly, these games do not focus on conflict-- they are more time-based and outcomes depend heavily on the thought process.

3. As the authors suggest GWAPs are currently limited when it comes to more problems which require more creative solutions. Currently GWAPs are useful in doing "rote" work tasks (ex. the picture tagging) but I don't think a game could really help someone write an essay, for example.

Dwijgarg - Jan 28, 2009 03:19:02 am

1. I definitely agree with the taxonomy: Games do need to have a winning/losing element in them. Victory or defeat, in my opinion, have the highest factor of motivation for a player. It is true that any player is attracted to a game because of interest in the game. However, the player will only continue to play and have focus in the game if he/she can expect a certain feeling of closure when the game ends. This feeling, I believe, is best created by enjoying victory or suffering defeat.

2. The only differences between the games described by the GWAP article are mainly found in the rules and regulations of each game - how the game is played, what each player has to do in the game, and how the player wins. However, the end point of the game - either winning or loosing - is the same for all the games. Other types of GWAP games can be created by introducing differences in the rules and regulations of each game. One example would be to introduce some sort of conflict (as most other people of the discussion have already pointed out).

3. There are definitely limitations to the kinds of information we can extract from GWAPs. For instance, we can extract information about the game itself and how it is helping any player playing it. However, this can only be done if the player actually shows interest in the game, and if the player understands the structure, the point, and how the win the game. Also, there is no way for the game to change with human needs. Thus, information can only be extracted IF the game attends to human needs at that very moment.

Alan Young - Jan 28, 2009 01:21:24 am

1.

I agree with Crawford's taxonomy that distinguishes games as needing a means of winning because I feel the description of "winning" is broad enough to encompass all situations where someone can be considered to be playing a game. Players are motivated by victory, where victory can be the general definition of winning or the prospect of gaining more than what was invested. For example, someone could spend 5 hours on World of Warcraft, a game that does not have a definite end point, but if he/she stands the possibility to receive enough utility or satisfaction from playing (possibly through statistics or rankings), despite using resources like money and time, he/she would play. In this way, a game is anything that allows players to undergo an experience where they are challenged in some way to receive more utility or compensation compared to what was invested in the experience. Whether the player achieves the challenge determines how well the player played the game. I think that the notion of "winning", "victory", "defeat", and in general, "gaming" is all based on this psychological aspect that motivates people to play or not to play or whether they even consider the act to be a game because the objective with which a game requires in order to have structure can be subjective.

2.

For output-agreement games, we have a clear objective that is very specific and so it is easy to structure a game on the formal points mentioned by Fullerton. For inversion-problem games and input-agreement games, we can look at it as a modification or variation of output-agreement games and so the same elements mentioned by Fullerton apply. The elements all share the same basic structure and characteristic of the game that Crawford described and only differ by the mere specification in its objectives. Thus, I feel that there may be other types of GWAPs that can be created, with endless possibilities or variations that can be met using combinations of the 3 mentioned by Von Ahn & Dabbish.

3.

I think that as long as we can continue to investigate the utility of GWAP-style games, we will see how such data can be used in different fields and in a broader scope. As more and more people get plugged into the idea, things other than 2-D images or multimedia can be included and the data that can be gained could be applied to academic, law enforcement, military, or even entertainment purposes.

Adam Kauk - Jan 28, 2009 03:16:02 am

1. "Chris Crawford's taxonomy of "play activities" (in the Fullerton section on puzzles) distinguishes games as needing a means of "winning" the game. If you agree with this, discuss the role of victory/defeat in motivating the player. If not, pick a game without victory/defeat and discuss what other features of the game might satisfy players."

I agree that winning is a big part of something being a game. Even with Solitaire, you could say that you are fighting against luck. There is a great desire to beat other people, because people like to have an excuse to show off, which games give them. Defeat motivates them to practice and try harder next time.

2. "Use the structure of Fullerton's formal elements of games to analyze the three types of GWAP games discussed in Von Ahn & Dabbish. How are these formal elements similar/different across the three types of games? Does this analysis suggest other types of GWAPs that could be created?"

I thought that the two formal elements that were least represented in the three types of games with a purpose were resources and conflict--conflict because in all of these games the people are working together, and resources because the only thing that people really accumulate is points (which is a resource, but a fairly basic one). The lack of conflict brings to me the idea that the designers of the games could add some competitiveness by having teams of two people compete against each other. This would most likely motivate people to work faster. Resources could be incorporated as well; for example, in the tag a tune game, people could be able to collect tunes which they can say that the tune they are trying to tag is similar to or not. This could be a collectible good that people build up to.

3. "Are there limits to the kinds of information we can extract using the GWAP-style game structure of Von Ahn and Dabbish?"

I would be tempted to say yes, but I wouldn't want to predict the future wrongly. I would be interested in seeing a game that gathers information on why people procrastinate. Of course, these games with a purpose are designed to combat procrastination (among other things), but if a game could gather info on that, it would be cool too.

David Burban - Jan 28, 2009 03:11:51 am

1. I don't believe that some games have any way to really "win". Take Tetris for instance, unless your opponent is the computer, the game just keeps on going until you lose. But why do so many people still play it? It is because Tetris has a top/high score. Every time you play the game, you try to beat that top score. So in essance, you are always competing against yourself. Some games for the Nokia N95 come with a couple of default scores with random names. So say you reach 5000 points in Bejewled, you have displaced Kenny from his #2 spot, which gives a sense of satisfaction.

2. The structure of the three types of GWAP games are very similar, except the procedures and rules differ. Outcomes are the same, as the games produce purposeful results. Conflict exists when the user wants to get ranked, or achieve a score greater than X amount. The number of players is usually two, unless the game is testing the validity of the input by pre-programmed scenarios. I'm sure that it would be possible to come up with new types of GWAPs by changing the procedures and rules.

3. Are there limits to the kinds of information we can extract using the GWAP-style game structure of Von Ahn and Dabbish? Yes, according to the article, it is difficult to extract diverse ideas or encourage creativity. The current general of GWAP strives towards conformity and description, without encouraging the aforementioned bases.

Alexei Baboulevitch - Jan 28, 2009 03:34:32 am

1, 2. Added late, don't have the super secret password. Will request during discussion section.

3. Yes, definitely. GWAPs take a while to gather data due to the fact that humans think much more slowly than computers, so only large, non-critical loads of data can be processed. The GWAPs described in the article work by analyzing consensus between random players, which creates generalized data that may not be appropriate in some situations. (For instance, tagging images using specific descriptors.) It's also possible to create a GAWP that requires another computationally impossible problem to solve - for example, tagging an image with quotations from literature. How would the quotations be compared between players?

Moonway Lin - Jan 28, 2009 04:07:51 am

1. I disagree. Many of the earliest games, from the arcade and NES generations, had no clearly-defined ending or definition for "winning." They simply looped indefinitely until the player died or retired. (See Pac-Man, Tetris, etc.) The player would be motivated to beat his personal best, or score reasonably high compared to his best if he has already reached his limit. For a game like Wii Fit, players are to define their own goals for weight loss, et cetera and are motivated to meet their goals for their own physical health. Regardless, it is still a game, not a "toy" by Fullerton's classification of SimCity.

2. GWAPs have the basic elements discussed in Fullerton's paper. Using Google's Image Labeler (a basic output-agreement GWAP) as an example, two players must cooperate to find as many common keywords as possible for a stream of images. This is the objective. The rules are that the two players cannot communicate with each other. The pair is given a score, and a global highscore list/leaderboard puts teams from all over the world against one another to compete for the top spot.

3. Yes. GWAPs can only extract information that is known or can be thought of within large groups of people. Even for seemingly trivial tasks like the Google Image Labeler, there have been instances where my partner and I were unable to come up with the same terms. In some cases, the images were ambiguous or lacked clear subjects. For example, if there were a GWAP for people to identify remixes of popular songs by their original names/artists, it would not work if half of the GWAP's players do not recognize the music.

Chao Michael Zhang - Jan 28, 2009 03:46:09 am

3. It seems as though GWAP-style games are primarly useful for "tagging" things with accurate keywords, rather than provide things that are just as descriptive, yet harder to obtain from a GWAP system. An example of this would be captions or descriptions for photographs and images. Though captions and descriptions would be informative and useful, and though it is a task that is impossible for computers to automatically generate, the GWAP-style game templates would not do a good job of collecting accurate captions or descriptions, since validating strings of words would be much harder than the simple comparing of single word "tags". Thus, one limit to the type of information obtainable by using the GWAP-style game structure of Von Ahn and Dabbish is the length and complexity of the answers obtained.

Timothy Yung - Jan 28, 2009 03:57:28 am

1. Games must have the "winning" aspect of the game, although the definition of "winning" may be very skewed. The role of victory/defeat (or satisfaction versus dissatisfaction) motivates players to "win" the game because all players (i.e. humans) strive to gain satisfaction when they play the game. Thus, they aim for the victory and try to avoid defeat (i.e. dissatisfaction).

2. All of the GWAP games discussed in Von Ahn & Dabbish require multiple (2) players, and there is generally no conflict because the players typically share a common objective and, thus, must cooperate. The procedures and rules of the games differ between the games because otherwise, they would simply be the same game. The outcome of the game for the players is the "winning" factor discussed in Fullerton, but it is also the result or data harvested from playing the game. From this analysis, we can conclude that the only thing changing between the games is the way in which the game is played, suggesting that other types of GWAPs could indeed be created.

3. The GWAP-style game structures are limited in the kinds of information that can be extracted because of the user interface available. People cannot provide much more information that can be helpful to computers beyond simple recognition. This is due in part to the lack of understanding of how the human mind solving problems can be translated into useful data for computer AI to decipher. However, there's no doubt that with some more ingenuity, we are still far from the limits of GWAP-style games.

Meiying Li - Jan 28, 2009 04:50:13 am

1. Everyone wants to win. Victory and defeat, as the possible outcomes of a game, can motivate the player by challenging him/her with the uncertainty of winning or losing. With a certain level of difficulty, the player needs to try his/her best (usually this means being more envolved in the game) in order to win. A victory of any form is like the prize of playing the game carefully. Even in games that do not have a specific objective (like World of Warcraft), there are both moments of resolutoin and measurable achievement which are just like "winning" partially of the game in some sense.

2.

  • Output-agreement games:

=>Players: Two strangers paired up by the game randomly =>Objective: Players need to produce the same output based on the same input =>Procedure: Each player sees the input and tries to produce the same outputs; if there is a match in what they have input, moves on; otherwise, they make new guesses. =>Rules: In each round the same input will be given to both players. Players cannot communicate with each other or see each other's answers. =>Resources: The players receive the same input for each round. =>Conflict: The players need to cooperate without communication. =>Boundaries: Not mentioned in the text, but probably in practice there might be maximum number of guesses each player can make for each input or time limit for their answers. =>Outcome: Whether there's a match in the player's answers or not.

  • Inversion-problem games:

=>Players: Two strangers paired up by the game randomly =>Objective: The describer needs to describe the input so that the guesser can guess the input. =>Procedure: Describer receives the input and tries to describe it to the guesser. The guesser make guesses based on the description about what the initial input is. =>Rules: The description provided by the describer should help the guesser in getting the right answer =>Resources: The input that the describer receives =>Conflict: The description provided by the describer is different from the input given by the game. Therefore in order for the guesser to guess the right input, the description needs to be correct and precise. =>Boundaries: Not mentioned in the text, but probably in practice there might be maximum number of guesses or time limit for answers. =>Outcome: Whether the guesser can guess the correct input

  • Input-agreement games:

=>Players: Two strangers paired up by the game randomly =>Objective: Players need to determined whether they have received the same input =>Procedure: Each player sees the input and tries to produce the same outputs; if there is a match in what they have input, moves on; otherwise, they make new guesses. =>Rules: In each round the same or different inputs will be given to the players. Players can only see each other's outputs. Based on these outputs they need to determined that if they have received the same input. =>Resources: The players receive the same or differnt inputs for each round. =>Conflict: The players need to determine if the inputs are the same based on each other's outputs. =>Boundaries: Not mentioned in the text, but probably in practice there might be time limit for their answers. =>Outcome: Whether the players have determined correctly.

The players part is the same for all three kinds of games. The objective, outcome, procedure, rules, boundaries and outcome are very similar and require cooperation of both players. Except the output-agreement game, the other two types of game requires interaction between the players. The resource is the inputs for all three types of game, but in practice one might better fit the purpose of a particular type of input than the other.

3. Using the GWAP-style game structure of Von Ahn and Dabbish, we lose diversity and creativity contribution from the players.

Daniel Kinder - Jan 27, 2009 11:50:10 pm

1. Victory and Defeat are not required for a game, for example World of Warcraft. What is a necessary part of a game is the reward of some sort of achievement. For games that existed before the internet, this mostly just meant victory, because it was the only thing attainable within the game space. Now, though, in separate worlds that do not simply cease to exist when we stop playing, achievement can be attained without actually beating anyone(though usually some small sort of victory or accomplishment is involved).

2. All three games follow rules that the players cannot break, they have clearly defined goals, and there are only certain game actions/objects available. One thing missing in the GWAP games is conflict, in that the players are not generally trying to beat each other as get general rewards. They all function very similarly in this too, mostly because they all have to do with each player trying to figure out the piece of information the other player has.

3. One limitation is that the information will be based on human thinking, which can be biased and opinionated. This makes for good information when trying to gain knowledge about what people will generally think about something(like what a general person will think is in a given picture). However, people will still get solid facts incorrect on occasion, which is probably why the GWAP game styles described are mostly about common sense information(like image observation).

Matthew Can - Jan 28, 2009 04:47:27 am

1. I disagree that games need a clear-cut notion of win or loss. They simply need to cater to the player's internal reward system to keep the player motivated. One of the best examples of this is the Grand Theft Auto games. The storyline and missions aside, players derive satisfaction in the game from exploring and interacting with the game's world: trying out new car stunts, provoking the cops then running away, etc. Notice that players explore the game world because it is intrinsically rewarding, not necessarily because they receive a concrete reward in the game. GTA was praised for being one of the first games to successfully execute this open-ended sandbox style of gameplay where play itself is more important than winning or losing.

2. All three GWAP games discussed in the reading have two human players. In the input-agreement and output-agreement games, the players have the same objectives. In the inversion-problem game, each player has a different objective. The inversion-problem game has resources in that one player's output is the input to the other player. This is also true of input-agreement games because each player can see the other's output. None of the GWAPs have conflict between players. Instead, the conflict is between the player and the game's limited means of communication with the other player. And of course, all three types of GWAPs have well defined rules that give the games their structural differences. Given that all three of these GWAPs share many similar formal elements, it seems that there are possibilities for creating other types of GWAPs. For example, one could consider a GWAP with more than two players or perhaps a competitive game with conflict between players. The reading cited a potential loss of "computational efficiency", but perhaps there is a creative solution that has not yet been explored.

3. There likely are limits to the kinds of information we can extract using GWAPs, if only because there are limits to the kinds of information we can extract from the humans playing the games. Disregarding the human factor, I don't see why the GWAP structure has any limitations. All designers need to do is come up with an entertaining way for humans to give out information.

Salman Rahman - Jan 28, 2009 05:35:05 am

These readings were actually pretty interesting. I especially enjoyed the second one about GWAP.

1. Chris Crawford's taxonomy of play activities distinguishes games as needing a means of winning the game. I agree that victory and defeat is a strong motivation for players to continue playing the game. Victory and defeat can come in many forms. There is the obvious and standard definition where the player either wins or loses. However, there are other ways to create the sense of victory and defeat. Another way would be to create a game where the player is awarded points during the course of the game. If he or she does well, he will gain many points as opposed to not as many. The motivation here is that the player will be 'victorious' in gaining as many points as possible and setting a 'high score' benchmark for his or herself and continually trying to improve that to outscore other high scores. Victory / defeat give the player goals to aim for in playing the game and make the game fun to play more than once. This is in contradiction to the puzzle which doesn't have replay value once the goal is achieved the first time.


2. Fullerton's formal elements consist of players, objectives, procedures, rules, resources, conflict, boundaries, and outcome. All three games share similarities in that there are two players who are not in direct conflict. Each succeeds when both cooperate and match. The difference amongst them are in the procedures: whether the match is on inputs, outputs, or guessing input from output. This analysis suggests that other such GWAP games could be created with variations in the formal elements.

3. I believe there is a limit on the information that is to be extracted. Currently the information that these games are attempting to extract lie in the realm of facts. It would be harder to extract opinionated information since one set of outputs would not necessarily produce the same outputs - in fact, the outputs could wildly vary!

Hari Ananth - Jan 28, 2009 09:49:11 am

1. I agree with Crawford in that all games should involve victory/defeat as the means to motivate the player. Although not all games have a distinct winner or a winning situation, they all involve some sort of goal that can either be achieved or not (such as a high score, completion of a task, etc.). One of the main reasons we play games is to measure our own ability and see how it stacks up with others' abilities, and this sort of competition is fun and healthy. If games did not provide people with the ability to "win," I think that they would be significantly less interesting, as we would have no measure of our abilities--something every "interesting" activity in life gives.

Teddy Allen Quan - Jan 28, 2009 07:20:08 am

1) As with most things, it really depends. Although, I agree with a need of "winning" a game is important and should be considered a staple of any game, some games like Second Life have no such feature. Many educators and educational institutions use Second Life as a meeting place where they can educate one another, which is pretty remarkable since it's considered a game!

2) Von Ahn & Dabbish and Fullerton both recognize that GWAP all rely on human computer interaction to solve problems which would be otherwise unsolvable to the computer. The three types of GWAP games discussed in Von Ahn & Dabbish include all the basic elements of Fullerton's structure (players, objectives, procedures, etc). New GWAP games must incorporate both the elements described by Fullerton and something unique or intresting to spark players interest.

3) I believe GWAP are limited to how well the programmer/designer can incorporate the problem into the game which has the basic elements described by Fullerton. However, it is important to note that human knowledge is also a limiting factor; not everyone knows fluid mechanics!

Shendy Kurnia - Jan 28, 2009 09:49:48 am

Games indeed need a meaning of winning. It is the goal that a player needs to achieve. If ones ever argue that there are games without winning state, maybe the winning state itself is not obvious enough. For a game like World of Warcraft which seems does not have a winning state, ones should see defeating a boss, or completing a dungeon as the winning states; it is just after that, there are more to win. The meaning of winning is important in a game for it is a destination where a player should go; otherwise, player does not get the point of the game.

There is limit of information that we can extract from GWAP-style game structure of Von Ahn and Dabbish. Games whose solution can already be determined by computer or AI do not need human's help to serve the games' purpose. Those games are just regular games, or puzzles, and are not GWAP-style game explained by Von Ahn and Dabbish.

Michael Cohen - Jan 28, 2009 10:17:27 am

1. I am not sure if I agree with Fullertons assertion that a play activity must have a means of winning to be defined as a game, however I do feel that a means of winning is necessary to motivate players. Generally people like to feel as if they have accomplished something with each task. Games are fun because they allow a person to escape the world, focus on a simple small problem, solve it, and feel like they have accomplished a goal, no matter how small. An important part of winning ( accomplishing the goal) is setting up an easy to achieve goal ( the thing to accomplish. ) For example, when presented with the thankless task of tagging photos, most people balk, however if an easy to accomplish goal such as guess a word, and some minor positive feed back, like an alert saying "Match!", the task is basically unchanged, but you have created the popular esp game.

2. All of the 3 discussed GWAP types share many formal elements. They share players, in the sense that there is one or more human players, and a computer "player" watching or learning from the results. As expected, each specific game has varying rules and procedures, however looking at a quote from the GWAP article can clarify the similarities: 

" A GWAP, then,  is a game in which the players perform a useful computation as a side effect  of enjoyable game play."

This implies that all GWAPS will have some method of HCI as a resource, and they all share the same objective in a general sense( perform useful computation). This general requirement suggests that as new types of computation become useful, and more methods of HCI are developed more types of GWAP could be created.

3. Yes, an important limit is deciding on the relevance and correctness of results. As the point of a GWAP is performing useful computation that is difficult for computers, this means by definition it is hard to check the accuracy of results by standard computational methods. Thus, in any situation the GWAP designer must choose an acceptable margin of error, and use methods such as player history or community consensus to try to achieve the desired accuracy. Thus, for any absolutely critical information GWAPS are not appropriate.

Bernardo de Seabra - Jan 28, 2009 10:22:06 am

1. I agree with the author in the sense that a game without a notion of victory/defeat couldn't be classified as such. Most people are driven by goals at every aspect of life: professional goals, personal goals, etc. This goals are important for them to wake up every morning and go to school/work, improve their skill set among other efforts. Hence, if someone is going to spend time playing a game it better have some notion of goals similarly to what happens in life. It not only gives the person the motivation to try the best they can to achieve the goal but also the feeling of accomplishment once a goal is reached.

2. Fullerton presents a set of formal elements that can be used to analyze different kinds of games: Players, Objectives, Procedures, Rules, Resources, Conflict and Outcome. Using these elements to evaluate one can notice that the GWAP's described by Von Ahn and Dabbish are very similar in most aspects with the exception of the "Conflict" element. In these kind of games the "Conflict" element is transformed in one of "Cooperation". Since there is no way for a system to evaluate whether or not the information the user submitted is accurate (hence having the users generating this information) a common technique is to match it against other users' submitted information. The analysis certainly doesn't lock all the games down to follow these characteristics. There is certainly room for innovation and creative GWAP's that could be a truly competitive game rather than cooperative.

3. The limitations of the kinds of information we can extract are based on the miss-informed users. A user might believe that, for example that a country belongs in a certain continent when that information might be completely wrong. Usually the knowledge of the crowds tends to be correct but there are times that this is not true, especially if we start entering subjective fields.

Alexander Cho - Jan 28, 2009 10:18:05 am

1. I agree that games should have a means of winning. We were raised in a naturally competitive society, and we all have somewhat of a competitive drive that motivates us, so it is natural for games to harness this drive and convert it into a desire to play the game. Also many of us are achievement oriented, we need a way of knowing that we finished, and how well we did (which explains the incentives raised by keeping track of scores and grades).

2. The formal elements of a game (objective, procedures, rules, resources, conflict, boundaries, and outcome) are all very similar for the three types of games discussed (output agreement games, inversion problem games, and input agreement games). The objective for all three is to make an agreement about a certain item. The procedure is slightly different for each game with the different roles, different approaches (input versus output). The rules are similar as players are restricted to either guessing or giving hints (hopefully accurate to the item presented). The resources are the computer and the UI. The conflict is matching. The boundaries are limited to words and the UI and the outcome is either making a match or not.

3. I would say there are limits. Right now I don't think there are games that would make users want to give out their most private and personal information (like past regrets, hurtful and traumatic pasts, etc.) And I think users would be slightly offended if a game was used to extract such sensitive information.

Shimul Sachdeva - Jan 28, 2009 10:19:05 am

1. Although victory/defeat are a good means of engaging a player, it may not be an exhaustive condition. Some games are different where the purpose may be to have as many points as possible in a limited amount of time or finish a task in the best time possible. In this case a player is challenging his/her self to perform better and there is no clear meaning of "winning" per se.

2. The three kinds of game structures have 2 players each, require players to guess labels/terms/answers for a given scenario and in order to win, the 2 players need to come to a consensus. The games differ mainly in their objectives. One game relies on similar outputs, another relies on similar inputs while the inversion problem requires one player to give information and the other player to guess the object of discussion.

3. Using the styles described by Von Ahn and Dabbish, GWAP have limited usage in terms of the data that is extracted from them. They are limited in their respective domains (labels for the ESP game). But there can surely be other styles of games that are useful can can provide useful information too.

David Jiang - Jan 28, 2009 10:08:52 am

1. I do agree that games need a means of "winning" or else people just wouldn't play. Though some games means of winning are different from others every game does indeed have it's own means. Some games goals are to kill your opponent, while others maybe to get a high score, complete the task in the shortest amount of time.

2. There 3 games are very similar. It was always two players playing at one time. The two players cooperate with one another to solve a common task of guessing what the other person is trying to communicate through standard means based on the respective game. Each player has their own objective. For Pictionary, one draws while the other tries to guess what their partner is drawing in the fastest amount of time. More GWAP games can definitely be created.

3. There are limits to the kinds of information we can extract using GWAP-style games. The games presented so far are all based on something concrete and it makes it hard to extract something subjective. Those seem to be what's limiting the games now, but maybe later, brilliant game designers will find a way to surpass those limitations.

Prahalika Reddy - Jan 28, 2009 09:50:05 am

1. I agree with Crawford in that games must have a means of "winning" the game. He indicates that the point of the game is to have players play until they reach an ultimate goal, at which point one player, or a team of players, have somehow won the game. Most games usually don't end until a winner has been discovered, and most people only play games so that they can possibly beat someone else or somehow "win" the game, defined clearly by the game's rules or by goals they set themselves.

2. The first element is that there is a "closed, formal system" in which the game is played. In the case of the three GWAP games described in Von Ahn and Dabbish's article, there is a closed system. It doesn't consist of an alternate world, but it is basically the web interface of the game in which you're given an input and have to produce an output. The next element is that "players are engaged in some sort of conflict." In the GWAP games, there is interaction with a partner, though it's not a competitive interaction. Partners have to work with each other so they both can achieve a similar goal. The major conflict in these games seems to be the race against the clock, seeing how many answers you can get correct before time runs out. The last element is that the games "resolve uncertainty in an unequal outcome." For most of the GWAP games, there is no clear winner and loser, and no real end to the game. Each time one plays the game, the game ends when the clock runs out of time, but there is never a time when the player can reach a certain level and reach the end of the game. The "uncertainty" that is resolved through playing the game is how many points one can accumulate before time runs out.

3. There are limits to the type of information extracted from GWAP games. I believe this is because most of the GWAP games use human abilities to replace those the computer doesn't have. And most of these involve labeling of images or sounds, or word associations, like describing other words. There needs to be clear input and output for these games to actually serve their purpose. For instance, I don't think it's likely that a game like World of Warcraft could be a GWAP game because even a portion of the game was devoted to getting answers for labeling images or sounds, it wouldn't be the main purpose of the game and would possibly make the other parts of the game less enjoyable and this part less important. I think the GWAP games have got to be a clear input-output scenario, because in any other case, the "purpose" part of the game would lose value.

Andrew Chen - Jan 28, 2009 10:47:26 am

The Structure of Games defines the concept of "play" as "free movement within a more rigid structure." I agree with this definition - after all, all forms of play exist within some structure, e.g. a child playing with a toy must accept the rigid structure that the toy has a fixed shape and size, but what he does with the toy is up to him - but after thinking for a bit, I think that definition could be augmented to say that play is "engaging free movement within a more rigid structure, without pressure." The reason is, play has no real life consequences (except for learning skills), and so is not driven by the pressure to accomplish or to survive. That is why I believe when people turn their work into play, they succeed more, since they are both enjoying it and also interpreting their work not only as a means to survival, but as a personal enjoyable experience. It must be engaging or fun to be enjoyable and to count as play, and the pressure must not be so great as to make it into a "grave and serious" experience with huge consequences.

Aaron Hong - Jan 28, 2009 10:10:44 am

1. I agree that winning is a distinguishing factor and nonetheless important. However, I think it is simplistic just to say victory/defeat is the strongest motivation to play. The nature of winning as used in terms of games involves people, which ultimately come with the competitive nature, ingenuity of people, and diversity (randomness).

2. The formal elements are: Players, Objectives, Procedures, Rules, Resources, Conflict, Boundaries, and Outcome. They are mostly the same, except the rules and procedures are sort of different. What is interesting to find, that in the Von Ahn and Dabbish article, the main objective of the game in “social interation.”

3. There seem to be limitations to the GWAP-style game structure of Von Ahn and Dabbish because they rely on "commonality" between players. So ultimately the players will strive for the most general answers, thus the very specific nature of the problem may not be addressed. For example say they have a picture of a dog, the picture will most likely be tagged as dog and not as German Shepherd of anything.

Raymond Young - Jan 28, 2009 09:13:22 am

1. I don't think all games must have a means to win the game, but those are the kind of games that are most successful. I would like to combine all the concepts Crawford distinguished and call that combination a "Universe" or "World", so as to say that each game presents a different universe or world with its own dimensionality (via rules). Then, by comparing the concept of a game world with our own world, it might be easy to see why winning is important to human players. Our own universe has rules, physical laws, and our human world has rules, human-made laws, but it's not clear what our objective really is, except maybe to be happy, or to love, or something like that. So we pick an objective, based on what we feel like doing, or what we feel will make us happy, like getting a high-paying job, or teaching others and watching them learn and grow, or becoming a con-artist. In other words, we can't really just be happy just by existing. We have to feel like we're doing something or accomplishing something, which we hope will make us happy or at least send us into pleasure or out of pain. So it makes sense that the concept of winning would be necessary for a successful game because it makes us feel good to have achieved the objective, which is to win. And since only one can win, the winner gets to feel extra special..

2. It seems to me that there are many more GWAPs that could be created. For instance, the concept of a character. The GWAPs analyzed in the VA&D article had no characters. A character, as VA&D explains, allows for an emotional connection to the game. I would think that this is because you are able to see yourself as someone else, that you are able to become someone else simply because their mind is now your mind since you control their every move. It becomes interesting for the mind because it gets to have a new body to inhabit, with new limitations, with a new look. I believe that the use of Avatars in GWAPs would be useful since the player gets to have a say in how their avatar looks. Since they went through the process of creating the look of the avatar, an even greater emotional connection may be developed.

3. There are major limits to the VA&D analysis. To describe only three types of GWAPs indicates that there are many more, very quickly accessible ideas for the design of GWAPs. I think that the games described in GWAP will hold one's attention for a far shorter time period than some of the possibilities, due to the repetitive quality of all their games.

Phiroath Chan - Jan 28, 2009 10:52:53 am

In Luis Von Ahn and Laura Dabbish's article on designing games with a purpose, they mentioned three game templates that were shown to be sucsessful in terms of games with a purpose (GWAP). My question is which of the three templates was the most successful? Output agreement games involve players doing performing the same role to acheive victory. Inverse problem games involve players performing different roles, but still in a way rely and support the other players. Lastly, input agreement games focus on again two players performing the same role. In my opinion i would have to say the inverse problem game template seems like the most likely to be successful. The concept of two roles allows a player to have choices in what the player wants to do in the game rather than the game deciding the role for them like in output or input-agreement games. Plus inverse problem games provides the option of switching roles rather than performing the same one throughout the entirety of the game which is also an attractive feature. Anyone have other thoughts?

32.152.228.177 - Jan 28, 2009 11:01:41 am

1. I disagree that games need to have a way to be completed or won in order to be fun. An example illustrating this would be the sims, in which there are no real game winning conditions.

2. The three are similar in that they all somehow match together two unacquainted parties, allowing them to collaborate to complete some task. They differ in their setup.

3. I believe these games are particularly useful for identification and classification of various entities.

Sean Ahrens - Jan 28, 2009 02:48:36 am

1. I agree that "elements" of victory as necessary to motivate the player -- but these elements do not necessarily have to comprise an actual "victory" whereby the game is "won". "Elements" of victory could include things such as character building, small goal achievement, etc. Take for example the Sims. Motivations here take place in small goal achievement -- like getting your character a job that will pay his/her bills, earning enough money to redecorate a room, or buy a new house.

2. Let's take ESP for example. The rules of ESP, like each of the other GWAPs, are enforced by the game itself. The procedures are to type in and submit a word, or to pass. I would say the game's only resource is the clock (which isn't replenishable). The conflict in this game arises out of the rules of the game itself -- that fact that you cannot talk to the other player and must somehow agree with him/her regardless. I'm not sure how to apply "boundaries" to this game. The outcome is to agree on a word as quickly as possible. Each of the three types of GWAP games seems to fit this same breakdown fairly closely. I think this surely opens the door for more types of GWAP game possibilities that differ from the game structure just described.

3. Additionally, I would say, yes there are definitely limits to what we can extract from the GWAP style game structure. It seems to only be able to apply to very basic identification techniques -- things like tagging photos, describing narrowly and concretely an object or idea. Information that has a diversity of opinion like "What is the meaning of life?" or differ in answers based on culture like "What is considered a breakfast meal?" are outside the bounds of the GWAP structure as described.


[add comment]
Retrieved from "http://vis.berkeley.edu/courses/cs160-sp09/wiki/index.php/Game_Structure"
Personal tools