Have you ever felt that one computer game is more "addictive" than another? Leaving definitions aside for the moment, it's fair to say that an addictive computer game is likely to be a more successful product than a game that is merely fun to play. Gaming developers apply numerous techniques and tests in an attempt to evaluate which games will hit the right buttons. Now researchers at Academia Sinica and the National Taiwan University (ASNTU) have developed a direct test for the addictiveness of a computer game based on physiological responses of a group of new players.

Online gaming is a massive enterprise involving millions of players and billions of dollars. Indeed, computer gaming is a major influence in driving personal computer processing speed increases, and advances in internet access.

Arguably the most intense of the games are the massively multiplayer online role-playing games (MMORPG). In all role-playing games, a player enters as a character or avatar who lives in the world of the game. MMORPGs are distinguished from other games partly by the number of players involved, but more importantly by the persistence of the game's "world," or playing field, in which the game changes and evolves in the absence of a particular player.

The competition between games for new players is as intense as the player's competition within a particular game. One result is that online gaming is a crowded arena. So many new games are released every month that the difficult task of dislodging players from ongoing games is a major barrier to success. The standard course for a new game is that it enjoys some level of popularity for a few months, and then vanishes into the noise before it becomes profitable.

The issue of MMORPG addiction is actively discussed in gamer forums. Studies have shown that one in five MMORPG gamers prefers to carry out their social interactions online rather than in the real world. Whether this indicates addiction, obsession, or simple shyness is open to opinion.

Still, the designers of computer games would be more successful if their products were engineered to be addictive, so let's take a look at the ASNTU addiction test. The challenge is to approximate the addictive potential of a new game before releasing it, preferably as early in the design stage as possible. Currently, the fate of new games is based on a designer's intuition and experience, and feedback from a trial group of users – a very subjective process that has proven to be less than reliable in predicting hit games.

The ASNTU researchers first established a criterion for the addictiveness of a game. They obtained the activity records of 11 online games from Gamania Digital Entertainment. Their basic assumption is that a game is more addictive if players tend to play as often as possible. However, measuring this among a range of games having different structures and constraints is quite difficult.

The tool developed by the ASNTU group is called the ratio of presence (ROP). This is calculated using "subscription period" – the number of days that a player has been subscribed to a game – and "presence" – the number of days during the subscription period that the player actually played the game. The ROP = presence/subscription period.

Note that ROP for a given player and a specific game changes with time. The following example will illustrate this:

A player subscribes to a new game on day 1, and begins to play with the following pattern.

  • Day 1................Plays................ROP = 1/1 = 1
  • Day 2................Plays................ROP = 2/2 = 1
  • Day 3................No play.............ROP = 2/3 = 0.666
  • Day 4................Plays................ROP = 3/4 = 0.75
  • Day 5................No play.............ROP = 3/5 = 0.6
  • Day 6................No play.............ROP = 3/6 = 0.5
  • Day 7................Play.................ROP = 4/7 = 0.571
  • The net effect is that ROP becomes smaller on average as the game is played less.

    The ASNTU group finds that the rate at which the ROP becomes smaller appears to follow a universal function, and that the rate of decline can be described quite accurately by a single number, beta. The researchers then take a bit of a jump, and argue that the slower the ROP declines, the more "addictive" the game must be.

    The notion that this describes true addiction must be taken with a grain of salt. What beta actually described is a player's behavioral persistence – the rate at which other interests deflect an initial obsession with a new game. This is precisely the information that a game designer would like to have, as it would predict the commercial dynamics of a new game.

    However, beta can only be calculated properly after the game has been in use for a considerable period of time. Accordingly, the ASNTU group wanted to develop a way to predict beta by measuring the physiological reactions of new players.

    What they did was to use facial electromyography during play to monitor the emotional reactions of a player. This is a standard technique for such issues, which works by detecting nerve impulses which activate the corugator supercilli and zygomaticus facial muscle groups. The first group is activated when you frown, and the second is activated when you smile. By electrically monitoring the electrical signals to these muscles, even subliminal activation can be detected. The twitch of a smile was considered a positive emotional response, while the hint of a frown meant that a negative emotion was being experienced. The strength of the signal was used to approximate the strength of the emotion.

    On the left hand head, the Zygomaticus facial muscle (smiling) is highlighted – on the right, the Corrugator supercilii facial muscle (frowning)

    The ASNTU group then related the frequency and strength of positive and negative emotions experienced while first playing a game to the historical beta value of that same game. To make a long story short, both positive and negative emotions were found to make a game more addictive, although positive emotions were more effective. More powerful emotions also led to larger addictive potential. In the end, a very good fit was found between the measurements and the addictiveness of the game as measured by the beta parameter, meaning that the electromyographical measurements provide a good predictor of the addictiveness of a computer game.

    Is addictiveness actually being measured here? That is largely a question of definitions. Video game addiction was considered for inclusion in the upcoming Diagnostic and Statistical Manual for Mental Disorders of the American Psychiatric Association, but in the end it was rejected.

    Three big questions deserve discussion:

    • Are computer games addictive in any reasonable sense?
  • If so, does it seem reasonable that addictiveness can be predicted?
  • If so, will game designers emulate the tobacco industry and design their games for maximum addictive effect? Would this be ethical?
  • We're looking forward to a lively debate!

    View gallery - 2 images