Results from Game Design Challenge: The ESP Game

By GameCareerGuide.com staff [07.17.08]  How can you design a game that tricks the player into thinking she or he has psychic abilities? That's the question we posed on a recent Game Design Challenge. Here, we present the three best solutions as chosen by special guest Mick West, a game programmer and founder of Neversoft Entertainment.

Observations
The focal point for judging in this challenge was positive reinforcement. The game didn't have to be overly sophisticated, but it did have to praise the player for doing well, thus focusing their attention on successess or wins rather than losses.

Matt Roberts (with help from fellow students), level design student at The Guildhall at SMU, Project(ion) (page 2)
Using Facebook social engineering, Matt Roberts came up with an idea that
The submission, said Mick West, was well detailed, showed a good sense for using manipulation, and had good positive reinforcement. It's also quite feasible as a concept.

Sina Jafarzadeh, media and computer science student at the Technical University of Dresden in Germany, Psychic! (page 3)
Sina Jafarzadeh's idea was "the best in terms of answering the challenge requirements," said West, though it could have used a bit more positive reinforcement. West also praised the idea for its "good use of probability and visual cues."

Nikhil Murthy, prospective student at Birla Institute of Technology and Science, Pilani, India, Sub-Clouded Mind [download: zipped, compressed rar file (please note that this file will only be hosted until 1 August 2008)]
Nikhil Murthy gets props for sending a runnable game. Although the idea is solid, "the gameplay is too slow to create an impression, " said West. "It does not specify what range the numbers are in, so it's hard to guess." The idea holds strong enough to take third place, though because of its subtle use of positive reinforcement, and a nice style that promotes magical thinking. [Note from Murthy: The numpad does not work with the game. Sorry for the oversight.]

Honorable Mentions
Louis Fontaine, MSc AI student at the University of Amsterdam, Precog Outlaw (see page 4)
Louis Fontaine presents an interesting idea with nice use of cues and positive reinforcement.

Peter Fung, McDonnell, Toronto, DS Minigames (see page 5)
Peter Fung's submission was a mixed bag of minigames with very detailed designs.


Matt Roberts, Project(ion)
What I am doing, and my submission for this week's game design challenge, is a meta-game I am calling Project(ion).

Right now it is a humble Facebook group run by a fictional character I am playing myself. I am padding the community with people who are "in on the joke" and developing the "tests" at the moment.

Project(ion) is an amalgamation of my experiences with cults, organized religion, social psychology, memetics, and paranoid conspiracy theory. Please feel free to join the Modern Shamanist Movement, led by Isaiah Crowley!

Phase 1: Set up online community, padded with several active members who are in on the joke. The online community provides positive feedback (reciprocal reinforcement: "I'm told I'm psychic." "Really? Me too!" "Wow, there's quite a few of us ... this must be a valid human ability"), a sense of long-term involvement (a simple application is not used and forgotten), and needed credibility.

Phase 2: Develop and release short Flash-based games with tunnels presented singularly in rapid succession, asking the user to determine which tunnel is safe and has a green gem, avoiding the other tunnels that house an approaching monster with red eyes. Introduce the exercise with a series of "focusing exercises" -- psychedelic colors and patterns billed to focus perception. Use a barely perceptible green tint in the black of the right tunnel, a red mixed into the wrong tunnels. Also, conflicting gradients on the cave walls (saturated to de-saturated, as in nature, for the correct tunnel) will draw the eye into the right tunnel and give a feeling of subconscious anxiety (that's not "right!") when looking at the wrong tunnel. There will be a time limit of 5 seconds, in order to prevent conscious study of the tunnels and ensure the users are feeling their way through the exercise. Some failure is expected and necessary for credibility.

At the end of the exercise, a semi-random percentage will be given, outlining the possibility of psychic ability. For instance, someone who gets half of the tunnel presentations correct will have a random number between 40 and 60 given as their percentile chance. This gives the illusion of a complex system in place analyzing the results. Players then discuss their results on the community forums, and are informed that the test gives preliminary results of potential; there is "obviously a component of coincidence to consider." Caution is reasonable, and reason is convincing - emphatically declaring everyone the next David Blaine is not.

Phase 3: Develop and release a second exercise aimed at remote communication. After the same long instructions and focus exercises as before, the user is given a "live feed" of a lizard trained in Pavlovian response. The lizard will only eat his grub when he is given a signal that it is safe, and the user is supposed to provide that signal from afar by either trying to put the sound in the lizard's head or by convincing him through feeling that it is safe. In reality, the user is just given one of several recordings. If the user clicks through the instructions quickly, skipping the exercises, etc., s/he will trigger a video wherein the lizard doesn't eat the grub. If the user spends a reasonable amount of time with the exercises, there will be a 70/30 chance of getting one of several clips of the lizard eating the grub versus one of the lizard not eating. This "experiment" is being conducted and monitored by "colleagues in Zimbabwe."

The need for a scheduled time will ensure that one person will see the clips at most once or twice a week, and the time of the experiment will be consistent with Zimbabwean time. The "leader" of the MSM will post findings, percentages, and the like as the experiment goes on at the web site.

Phase 4: After several people begin reporting greatly increased psychic ability (those in on the joke), we hint at a shadowy big brother apparatus attempting to censor the MSM. After rumors of I. J. Crowley's mysterious death, members will be given breadcrumb clues to follow to find the new web site. Live (faked) messages from Crowely, such as cryptic announcements full of statistics that all relate to, say, six and three (e.g., "We have 32 new members this month, with 23 scoring in the 60th percentile or more, half of whom...") will get people to "feel" that the next secret message will come at 6:30.

These secret videos released by I. J. Crowley in hiding will cue followers in to meeting places and times, missions that need to be completed, people to be monitored, and all sorts of meta-game greatness. Progress reports will be issued to give the illusion of an ongoing struggle. A safe haven that happens to be a large corporation of chain coffeehouses will become a focal point, ergo a strong impetus for corporate sponsorship.


Sina Jafarzadeh, media and computer science student at the Technical University of Dresden in Germany, Psychic!
"Psychic!" combines two aspects to let the player believe he has psychic powers:
  1. Right use of probabilities
  2. Right use of visual tricks

In this game the player competes against the CPU in a card duel. Each combatant starts with the same amount of life-points (15 LP) and has a set of 4 cards in front of them. 2 of them are hit-cards, which cause damage to the opponent and 2 of them are blank ones with no effect. Although the probabilities for selecting a hit-card are the same as a coin-flip, the fact that there are 4 cards increases the satisfaction if a right one was chosen and a possible pattern was discovered.

One round consists of 2 draw phases for each combatant with 4 cards at any time. The player starts to choose a card and the effect of the card immediately takes place. Then it is the CPU's turn. During the second draw phase of the CPU, the player can also select a card and if it is the card the CPU has selected the effect of the card is neutralized.

A combatant wins if an opponent has no life-points anymore. Should the player loose against the probabilities and despite of the additional tricks there is always the excuse that he wasn't concentrated enough this time.

Probabilities
Because of the following reasons the player is in advantage and has better chances to win the duel:

The player starts first and the effects take place immediately. Hence the CPU can't finish a round if he has lost all life points.

Player chances:

CPU chances:
Although the difference between the probabilities for the player and the CPU doesn't seem so serious, it will deploy its effect on a long-term-basis:
Player chances to do minimum 15 damage in 21 rounds
CPU chances to do minimum 15 damage in 21 rounds
Visual Tricks
Additionally, the advantage of the player can be increased by using some visual tricks to make a card more attractive. Of course these tricks have to be subtle. Assuming that through the visual manipulation chances of selecting the right card rise by 5%, the following chances result:
Player chances:
Besides of that the chances for the CPU sink thanks to our mechanics:
Assuming 10%, 15% or 20% rise following chances results:
p(hotPV10)=88.56%
p(hotPV15)=93.04%
p(hotPV20)=96.05%
p(hotCV10)=47.55%
p(hotCV15)=46.31%
p(hotCV20)=45.08%


As you can see this has a huge effect on a long-term-basis, which is the goal. The following two figures show manipulated sets of cards. Guess which card could be the hit-card. (Contrary to the game there is only one; Please use the email attachments because of the size.) Afterward, there is the explanation of the visual tricks. There are the solutions to Figures 1 and 2.

Figure 1 The second card


Figure 2 The third card


Visual Tricks Used
Symmetry. Symmetry is seen as a pleasant condition by human beings. The dots of the hit-card are arranged symmetrically while one dot of the fail-cards destroys the symmetry (the second one top left if the card is on the left of a hit-card or the second one top right if the card is right to a hit-card)

Color perspective. In paintings red tone is used for near objects while blue tone is used for far objects. The color of the dots tends to red for hit-cards and it tends to blue for fail cards.

Brightness. Hit-cards are slightly brighter than Fail-card. Humans can detect changes of brightness better than changes of color (they have much more rods than cones). This fact is also used in JPEG compression (chroma subsampling).

Space. The space between fail-card and hit-card is greater than the space between fail-card and fail-card. This distinguishes the hit-card.

Visual and Audio Tricks that Can be Used (subliminal messages)

Video. Although James Vicary has manipulated his notorious "Eat Popcorn - Drink Coke" experiment, at least subliminal messages could attract the attention of the player to the hit-card and support the above-mentioned tricks. This could work as follows: one frame in a to-be-determined frame range would be black with only the space where the hit-cards are colored white.

Audio. Adding infrasound to the ambient background music when the player's mouse courser is above a fail-card. There is a study that 17Hz infrasonic can cause uneasiness although it is beneath the lowest limits of human hearing. (This may also be the reason for "ghost" encounters.)

References
Cones: https://en.wikipedia.org/wiki/Cone_cell
Rods: https://en.wikipedia.org/wiki/Rod_cell
Chroma subsampling: https://en.wikipedia.org/wiki/Chroma_subsampling
Color perspective [PDF ]
Subliminal message: https://en.wikipedia.org/wiki/Subliminal_message
Infrasound: https://en.wikipedia.org/wiki/Infrasound


Honorable Mention
Louis Fontaine, MSc AI student at the University of Amsterdam, Precog Outlaw
Shorthand description: FPS shooting gallery
Format: web Flash game
Relies on: cues of various subtlety (ranging from subliminal to unmissable), conditioning, and (partially unwarranted) positive reinforcement

The player controls an on-screen crosshair and can shoot slow projectiles that explode after a set time. The levels contain various objects (blocks, walls, furniture) where enemies can be hiding behind. Enemies can pop in and out of their hiding places randomly. When an enemy pops out, the player does not have enough time to shoot it down: the enemy will pop back before a projectile will have reached him and explode. This means that players need to shoot in places where they expect an enemy to appear, rather than where they see one.

Players are made aware that enemies are about to appear by multiple cues. One cue is very obvious, say a clear symbol accompanied by an obvious sound. The other cue(s) are more subtle, e.g., small animations in the background or certain soft noises. Moreover, these subtle clues relate to where the enemy is about to appear. The mapping from a clue to a position isn't obvious however, and the player will learn this mapping through persistent play. In fact, the cues should be so subtle that the player won't pay much attention to them and unconsciously makes the mapping from cues to positions. Optimally, in the opening stages the players won't even be aware of the subtle cues, and instead rely on the obvious cue.

Players are assigned a score per hit based on how accurate they were. When a projectile explodes very close to an enemy, they get awarded full points for that shot. When a shot is removed somewhat from the enemy but still manages to kill or hurt it due to the explosive blast, he gets only some points for that shot. In other words, the more accurate the hit, the more points.

In the first level, there will only be four places where enemies will pop up. Moreover, the enemies are positioned so, that even when a player anticipated an enemy to appear in a position next to the one where it actually appeared (i.e., the player fired a shot in that direction), the blast of the shot will still kill the enemy. Thus, in the first level, the chance of killing an enemy is 50 percent per enemy, if the player shot at one of the four positions when an enemy was to appear. In later levels, the number of enemy positions increases.

In the first set of levels, the primary enemy cue (symbol, sound) is very obvious. As the player progresses, this cue becomes more and more muted, and will eventually be absent. At this point, hopefully, the player has been conditioned to see the subtler cues as premonitions of enemies appearing. In cases where this has happened, it is to be expected some players will become aware of the subtle cues they are reacting upon. The others however, may think they have acquired precognitive power through their experience.

A side character will constantly evaluate the performance of the player, but will be overly positive; only the worst of shots will be called poor - and still with tact and consideration. Evaluations range from: "Hmm, you were a little off this time", to "Brilliant shot! I didn't even see it coming." This positive reinforcement will help in making believers out of those players that improved through conditioning.


Honorable Mention
Peter Fung, Toronto, McDonnell
System: Nintendo DS
Genre: Minigames series
This game is designed to mislead players into thinking they have Extra-Sensory Perception by careful manipulation of data and by increasing their chances to win in games of chance without them knowing.

Inaddition, the pace of the minigames will be extremely fast-pace to force the player to rely on intuition and random guesses. Data gathered from the player's initial profile setup and the questions they will answer in the process will be crucial as they will be used in the minigames. The final step into tricking the player into thinking they have special abilities will be the presentation of baseline statistic to make their results seem above average. The overall game will provide quick, easy to access entertainment and leave a lasting impression on the player.

Datamining Techniques
Profile creation. Before a player can play (except in Guest Mode), the player is given a generic male or female avatar to represent him/herself. The way the player dress up their character, the colour scheme they use for the clothing, their hand of preference and date of birth will be incorporated into correct answers in the minigames.

40 Questions. After the avatar setup, the player is asked 40 random questions interlace with questions regarding their favourite colour, number, animal, fruit, season, etc... These "favourite" questions should appear in completely random sequence and the questions must be localize to the region to ensure that the questions don't become obviously irrelevant.

Minigames: Conditions

Restrictions and Purposes. The player gets 1 additional point regardless if the answer is correct or not, and 0 points if they did not provide an answer. This is to ensure that even if the player does not actually know the answer, s/he will at least guess. As some of these questions provide the probability of a correct answer, it will help reinforce the illusion that the player has certain abilities. Inaddition, all minigames have a fixed timer of 2 to 8 seconds to answer. This is to force the player to rely more on their intuition and guessing ability as opposed to actually solving the minigames; the tactile nature of the NDS will help greatly as it reduces the response time used on the answer. A sense of anxiety will be created by playing an incrementally louder sound on every passing second to reinforce the idea the player must pick an answer immediately. This will help in disguising the manipulation of correct answers as the player will be too focus in trying to get the next answer as quickly and correctly as possible.

Two Versions of Minigames.
There will be two version of the same games, this is to ensure that if the resulting score don't deviate too greatly from the baseline score as to not rouse suspsicion. The first version, which will be used 70 percent of the time in a session, will be a truly randomize version where the representation of the answers bear no relevance to data mined previously. The second version, used 30 percent of the gaming session, will have answers based on data previously gathered from the player.

Minigames
Multiple Choice Impossible Math. A 4 to 7 digit number divided by an 1 to 3 digit number and ask for the answer. In the altered version, a number pertaining to the users favourite number or date of birth will be used as one of the multiple choice answers.

Number Guessing. A random range and size of numbers will be used in the normal version of the game. In the altered version, a smaller range containing either parts of the player's birth, age or favorite number will be used.

Card Guessing/Ball in a Cup/Three Card Monte. 1 object, out of 3 - 7, will be revealed to the player and then mixed back in. The player will have to find the shown object among the others. The modified version will actually have more of the same shown object in the mix. This work on the player's assumption that there's only 1 correct answer, as the game does not claim that there's only 1.

Down the Ladder. Popular decision-making game. There's 4 vertical lines with multiple horizontal connecting between each vertical line. Player select one of the 4 line to follow from the top, each time the player encounter a horizontal line, the player must cross it. The bottom of the ladder is not visible when the game begins. Each vertical lines will be colour coded differently. The player's favourite colour or the colour used on the avatar will be used on the correct line in the altered version. In the normal version, the colours and its variety will be completely random.

Modified Pachinko. In this minigame, the player will try to drop a ball from the top of the box and try to reach a predetermine destination. This will be difficult as the ball can go from one side to the other quite quickly as the path will be randomized. In the altered version, the pathing will actually lead the ball to the general area where the predetermined area is no matter where the player drop the ball from. Even in this case, the player will not be guaranteed a win since the pathing will only increase the probability of winning.

Which is Bigger/the Same/Smaller. An irregular shape is shown in the example, the player will have to find an identical piece in the answer. This game relies on optical illusion to confuse and disorientate the player. In the modified version, there will be more than 1 correct answers.

Fruits Basket Turnover. Before the problem picture is shown, the rule pertaining to that particular puzzle will be explained. As soon the player clicked ready, the problem picture will be shown and the player will have to pick the correct answer. The puzzle may ask what fruit should be called to ensure A will reach B, ensure the center will get a chair, or other conditions. In the modified version, there will be more than 1 correct answers.

Left of Right/Top or Bottom/Heads or Tail
s. The game just simply guess if the answer is the Right object or the Left object. In the modified version, 90 percent of the time the answer will be the player's hand of preference.

Others. There are many other games available which can be added, adding more in this proposal will be redundant as many games used will share the same principle of relying on chance, brain power/calculating abilities and/or illusions. It should be noted that different regions should have their own type of game. For example, for western cultures, Fruits Basket Turnover can be changed to Musical Chair.

Scoring
Scores will given out in different category, they are listed as intuition, math, perception, luck, and instinct. Some minigames will affect only 1 category whereas others will affect more than 1 as it depends on the nature of the minigame itself.

Intuition.
This measures how intuitive a player is. In minigames like Left or Right where it relies purely on the player's guessing ability, the percentage of being correct will influence these scores.

Math. Measures a player's mathematic prowess. Although this is not actually a part of the ESP definition, it helps create a sense that the player is "special" if they have a high score in this area.

Perception.
This measures the ability of player to see through illusions and tricks. Games like Which is Bigger/the Same/Smaller will affect this score.

Luck. How lucky a player is. Of course, due to the altered minigames' nature, this score be slightly inflated. But the idea that a person is lucky can be contributed to ESP abilities. Even if not, who wouldn't feel good about being lucky?

Instinct. Instinct can be loosely translated into decisiveness. How quickly a player answer a question relative to the time limit will affect their score. Instinct is often regarded as a creature "sixth-sense" and assist in snap decision-making.

Score Presentation
The player's score will be presented on a pentagram with each vertices representing a category. This style is used in many other games such as DDR, Prince of Tennis, Ragnarok Online (hexagon in this case), and many other games.

Comparitive Presentation
Using a double-blind test, volunteers will play the game (without the avatar or 40 questions, of course, the player won't find out until way after ... ) and generate several average category scores and 6 overall mean score pentagram, 5 with emphasis on each category, 1 that does a mean average of those 5. The purpose of this test is to make it seems that this is done in a scientific manner and make the player assume that the results are legitimate and can be trusted to compare with this false baseline.

Score comparisons using the test will be as followed: Category scores can be compared placing the player's category score using a graph chart for each category individually. These category score can be displayed all at once in the graph chart format.

Overall score will be compared with the player's score pentagram appear on top of the 1 of the 6 overall mean score pentagram (the pentagrams are translucent to show the difference between the two) chosen by the player.

Positive Reinforcement
When the player scores low in one area, words of encourage should be used by giving out examples of statistical anomalies. For example, a coin might flip 13 times in a row, but if you flip the coins enough time, the percentage chance will be near 50 percent. Encouragement like this will help ensure the player will continue playing. When the player scores slightly above average, hint that they are special and might possess ESP or a sixth sense. This is the seed that idea into the player's mind if it isn't already there.

Revelation
After the player pull far enough away from the false baseline, the game will start telling players to visit certain webpages for more information. Initially, these webpages will refer to ESP, statistic anomalies, sixth sense, animal instinct and so forth. Eventually, the game will give webpages that argue against the previous pages. When the player's score is significantly above the false baseline and after a period of 90 active play sessions on different dates (this is to prevent spoilers on the internet), the truth about this game will be reveal slowly. Each of these revelation will allow the player to remove the particular type of minigames from their session. When all the tricks of this game is revealed, a final webpage will explain that although this game was intended to trick the player in believing they have ESP, they have in fact gained an ESP abilitiy. The reason being that by playing these fast paced games, they have tuned into their instincts/gut feelings/etc... and to rely to trust it more for making decisions. Now the players will have a true appreciation of what ESP is and isn't. Most importantly, they will be able to differentiate between mentalists and frauds as oppose to the genuine possibility that someone truly has ESP.

Legals: Legal terms such as "This game is meant for entertainment purpose only" should be prominently display before or during the game startup to prevent lawsuits.

References
https://en.wikipedia.org/wiki/Fruit_Basket_Turnover
https://en.wikipedia.org/wiki/Ponzo_illusion
https://en.wikipedia.org/wiki/Three-card_Monte
https://en.wikipedia.org/wiki/Pachinko

Return to the web version of this article
Copyright © UBM TechWeb