Putting gaming data to good use
Gaming is a massive data generator. World of Warcraft (WoW) has over 5 million subscribers, The World of Tanks is played by over nine million fans, while Dota 2 has reached more than 826 thousand concurrent players on Steam. If you multiply these numbers by the actions taken in the game, such as character selection, spells, strategies, and maps, you end up with data that can help in developing new extensions or games. Furthermore, you can have a good idea about what pleases each user, and dynamically generate playing situations which are more engaging.
Data scientists from InDataLabs put together the following guide to help game developers make the most of this complicated yet rewarding discipline.
Data Science for Game Developers
Game developers gather playing logs from all the members, but for a long time these were overlooked and used mainly for error identification and debugging. The buzz around data science in marketing and retail has inspired developers to take a closer look at the stored data and use it as a way to create more balanced games and better maps, and fight against meta-playing. It means giving all players a fair chance at winning based on skill, not reverse engineering of the system.
A Balanced Map
A game should be hard enough to be challenging, but fair enough to offer at least some satisfaction. If your character dies as soon as the play deploys, there is no incentive to continue playing. Thus, developers should balance the power of every entity and those of groups to give everyone a chance to play, to grow, and to win.
This sounds logical, but as the developers of Halo 3 found out during their beta testing, it can be tricky. Their solution was to monitor all actions and use a heat map to identify the ones which proved that one of the players had a clear advantage.
Incentives to Play
Each player has some favorite scenarios, tools, moves or parts in a quest. By monitoring what they usually use as equipment, actions, and alliances, a profile emerges. Looking at sudden drop-offs during more difficult games can indicate what makes a player anxious and disgusted of the game. It is also important to see if after a logout they return to the quest.
A good game offers the player what they enjoy most in moderate doses and also puts some new things in the scenario to make it more challenging. If the game rolls out similar features to those they already like and use, they present the player with these in the hope that they will quickly adopt the new items. Data science is what help to make it a reality and connect the dots in player behavior.
Focus on Skill
Another problem is matching teams against each other. Data science can help evaluate the overall power of clans and generate a map which is suitable for their equipment and also gives each of them a fair chance at winning.
It is common for online gaming platforms to match slightly disproportional teams in the same game, for example, a level apart (a common practice in The World of Tanks). It helps the lower ranked team get some experience and some points to move to a higher level. It makes sense because to advance, you need to challenge and win in front of those who are more experienced instead of remaining in the same class.
A Heads Up on Addiction
This over-customization of gaming also has a darker side effect. Because players are getting very engaging content perfectly tailored to their needs, skill levels and interests make the gaming experience highly addictive. When the algorithm behind the game gets deep into your mind, desires, and triggers, it can turn from leisure to addiction.
As in most enjoyable activities, gaming releases endorphins and, together with some adrenaline, keeps people glued to the screen. In extreme cases, this turns players into zombies, and there were multiple cases of people dying from exhaustion while playing games for up to 40 hours continuously.
With this in mind, game development should be optimized with data carefully to keep the engagement levels healthy.
Data Science for Game Players
Not only developers use data science for games. On the other side, players have learned that looking at the right numbers can uncover tricks, cheats, and even items which are not yet officially released. The dance between developers and gamers resembles a bit that between security experts and hackers, each trying to be one step ahead of the other.
This has evolved into meta-gaming – a way to learn about hidden features to get advantages without increasing the skills or even spending more on equipment. It is a practice which is discouraged by game creators but has been a common practice ever since WoW was released 15 years ago.
WoW: A Case Study
An interview with Ryan Shwayder reveals that for a game designer, meta-gaming is a double-edged sword. On the one hand, it is flattering that players would go to great lengths to uncover what is new in the game by data mining. On the other side, it is much hard work concealing the secret about new releases while working on them.
The production director also highlights that although the development team doesn’t want to give any spoilers, they have to upload everything to the servers before a patch becomes available. There is a tradeoff between spoilers, testing and making sure everything runs smooth.
One funny aspect they highlighted is related to the names they give projects under development. Through data mining, these are leaked into the community’s gossip and news. The developers never act on these and just enjoy the free publicity around the rumors.
Even if data mining for games is already a few decades old, as in the case of AI, it has been in a hibernation stage for too long. With advanced use of data science, the industry will include far more personalization options, and gameplay will not only be optimized per clan or level but for individual players. If Amazon can do it for your shopping cart, why not expect the same from your favorite game developers?
Written by: Marta Robertson, BOSS Contributor
Marta Robertson has over 7 years of IT experience and technical proficiency as a data analyst in ETL, SQL coding, data modeling and data warehousing involved with business requirements analysis, application design, development, testing, documentation, and reporting. Implementation of the full lifecycle in data warehouses and data marts in the marketing industry.