This model demonstrates how gambling-like behavior can emerge in lootbox spending within gaming communities. A lootbox is a purchasable mystery box that randomly awards the player a series of in-game items. Since the contents of the box are largely up to chance, many players can fall into a compulsion loop of purchasing, as the fear of missing out and belief in the gambler's fallacy allow one to rationalize repeated purchases, especially when one compares their own luck to others. To simulate this behavior, this model generates players in different network structures to observe how factors such as network connectivity, a player's internal decision making strategy, or even common manipulations games use these days may influence a player's transactions.
Each tick, every agent decides if they want to purchase or not based on the decision structure defined by Decision Strategy. These strategies are ALWAYS_BUY, COIN_FLIP, and PRICE. A player following ALWAYS_BUY will make a purchase every round regardless of their satisfaction with their previous luck. A player following COIN_FLIP will generate a number 1-10, and if it greater than their current buyThreshold, they will make a purchase. For the Price decision structure, the average price value of their previous boxes is calculated and divided by the average rarity of their previous boxes. If the rarity is better than the price the player has been paying for the boxes,they will decide to buy - if not, the player will resort to COIN_FLIP to determine if they should make the purchase.
If any manipulations are in play, they are processed next, skewing the luck of players boxes or awarding them free ones depending on the strategy selected.
After a player has purchased the box, it is added to their intenal lootbox history, which serves as a short term memory of their recent lootboxes. Their buyThreshold is then updated depending on the value of the new box.
By default, if the new lootbox they have purchased is worse than the last, their buyThreshold is decreased, reducing their likelyhood to purchase in the future. If the new box is better than the old one, buyThreshold is increased so they are more likely to buy again in the future. If a player is using the PRICE decision strategy, their threshold is updated according to a formula. The price of the old box is divided by its rarity, and the same is done for the new lootbox. A better box will have a smaller value returned from this formula. If the old box has a smaller value than the new box, buyThreshold is decreased, and vice verse. This ends the turn for the player who decided to purhcase.
If a player decides not to make a purchase, they instead compare themselves to their peers. The networks generated are all directed, and players will compare themselves to a node they are looking at, rather than nodes looking at them (successors). If a node has no successors, a random node is selected from the network. A player's lootbox rarity history is then averaged and compared to the successor node. If the player has better history than the node, the strength of the connection between them is decreased (or the edge is removed if it is equal to 0), and buyThreshold is decreased. If the node has better history than the player, the strength of the connection is incremented, and the player's buyThreshold is incremented by the weight of the connection between the nodes. If any manipulations are present, they are processed after this calculation. This ends the turn for the player who decided not to purchase.
After selecting the parameter values for a specific run, click the blue power button at the top of the GUI to initialize the model. If you would like to slow down the model, on the bottom left panel labled "Run Options" increase the "Scheduled Tick Delay". Click the blue play button to start the model once you have selected the run options you want.
Num Players: the number of player agents interacting in the model
Decision Strategy: The buying strategy each agent utilizes to decide if they want to make a purchase
Manipulation:
Network Type:
Stop Time: the number of ticks the model will run for before terminating.
Many studies on adolescents (N Schuhmacher, P Van Geert, L Ballato) have shown that they demonstrate a tendency to adopt behaviors similar to those of their friends, or those who share interests with them. They also engage in risk-taking behaviors if peers who are similar to them participate in those same events. This model could be extended by allowing players to be influenced more strongly by those who are similar to them in such areas as age, gender, work-status, etc.
Weighing the impact of each lootbox on buyThreshold is another improvement that could be made to the model. If a player spends a small amount of money and gets the rarest item possible, that should impact their future decisions much more sharply than if they recieved an average box for an average sum of money.
Another improvement to the model concerns the drop rates of specific types of lootboxes. Implementing a rarity generation method which allows users to input the real life rarity of different types of boxes from real games may allow for very interesting observations and projections.
Agent-based model of risk behavior in adolescence, retrieved from: https://www.comses.net/codebases/3844/releases/1.1.0/
El Farol, Netlogo Model Library
Dr. Murphy for his area expertise and boundless patience!
Hing, N., Russell, A., Tolchard, B. et al. Risk Factors for Gambling Problems: An Analysis by Gender. J Gambl Stud 32, 511534 (2016). https://doi.org/10.1007/s10899-015-9548-8
N Schuhmacher, P Van Geert, L Ballato (2019, April 08). Agent-based model of risk behavior in adolescence (Version 1.1.0). CoMSES Computational Model Library. Retrieved from: https://www.comses.net/codebases/3844/releases/1.1.0/
North, MJ, NT Collier, J Ozik, E Tatara, M Altaweel, CM Macal, M Bragen, and P Sydelko, "Complex Adaptive Systems Modeling with Repast Simphony", Complex Adaptive Systems Modeling, Springer, Heidelberg, FRG (2013). https://doi.org/10.1186/2194-3206-1-3
Romanowska, Iza, et al. Agent-Based Modeling for Archaeology: Simulating the Complexity of Societies. Santa Fe Institute Press, 2021.
Zendle David, Meyer Rachel and Over Harriet (2019) Adolescents and loot boxes: links with problem gambling and motivations for purchaseR. Soc. open sci!