The goal of this study is to research the disparities in character positions between clients and server when playing an online game. The data needed was gathered by letting three players play a game made by me against each other, using extrapolation methods like the Kalman Filter on the characters’. During the play-through each client saved all characters positions together with the input made by the players. The clients logged the information every network update, in synch with the server. When the time came, all clients sent their information to the server, where it was collected, analyzed and compared with the information the server had registered. By calculating the difference in position of the server and clients characters, a disparity value could be extracted. This value is what was used to calculate a disparity value between the server characters and all clients’ counterparts. The same value is also what was used to answer the questions on how much impact the different extrapolation methods have on a game, as well as how big of an impact input made have on the delay of the game. An important part of the study was to make sure that the information gathered was collected at the same time on the clients and the server, as well as to be able to enable and disable parts of the methods. Therefore the whole game used in this dissertation was built focused on this study. All extrapolation methods are toggle-able and the information gathered is synched using time.windows.com.

Under de senaste decennierna har datorspel etablerats som ett legitimt forskningsområde. Inom denna forskning har den artificiella intelligensen (AI) varit av stort intresse framförallt i relation till de karaktärer inom spel som kallas för non-playable characters (NPC). Fokus har legat på att göra dessa karaktärer mer trovärdiga och på så sätt göra spel mer engagerande och potentiellt främja immersion. I och med detta bör alla aspekter av NPC:er utforskas för att säkerställa att alla bidragande faktorer har identifierats och kan därefter användas för att ytterligare utveckla dessa. Då social interaktion inom datorspel har tidigare kopplats till en spelares immersion är det viktigt att förstå alla komponenter som uppgör denna. Den mänskliga kommunikationen är grundad i både verbala och icke-verbala uttryck vilket tillsammans skapar den sociala interaktionen. Därför undersöker denna studie de aspekter av en karaktärs beteende som kan kategoriseras som icke-verbala uttryck. Detta utförs med hjälp av en kvalitativ observationsmetod för att identifiera vilka variationer som existerar, hur de uppfattas, samt om de påverkar karaktärers trovärdighet och därefter spelarens känsla av immersion. / In recent decades, computer games have been established as a legitimate area of research. In this research, artificial intelligence (AI) has been of great interest especially in relation to the characters in games called non-playable characters (NPCs). The focus has been on making these characters more believable and thus making games more engaging and potentially more immersive. In doing so, all aspects of NPCs should be explored to ensure that all contributing factors have been identified and can then be used to further develop them. As social interaction in computer games have previously been linked to a player's immersion, it is important to understand all the contributing components. Human communication is based on both verbal and non-verbal behaviors, which together create social interaction. Therefore, this study examines the aspects of NPC behavior that can be categorized as non-verbal expressions. This is achieved using a qualitative observation method to identify what variations exist, how they are perceived, and whether they affect the believability of characters and the player's immersion.