Emergency response agencies that need to care for large crowds in real-life events, report a constant
lack in human volunteers in large numbers for training purposes. Moreover, existing computerized
training aides either totally omit affected crowds in their scenarios, or represent them as computer
generated models. A potential solution that can provide real human input in large numbers for
training purposes can be found in the form of Massive Multi-user Online Role-Playing computer
Games (MMORPGs) that attract millions of users on a daily basis.
In order to evaluate the use of MMORPGs as an emergency simulation platform I had to examine
the in-game behaviour of participants, usability issues, data collection methods, and data reliability. I
did so by constructing a multi-user computer game that included food shortage and a pandemic
spread scenarios. Data collected included every possible item that could be technically logged, both
qualitative (questionnaires, user’s self tagging of events) and quantitative (all in-game actions and
their coordinates, players and virtual environment in-game status).
The abundance of data enabled easy triangulation and verification. The main findings were:
participants attention span was about 90 minutes, they demonstrated only a narrow range of
behaviours necessary for their in-game survival, and this behaviour followed loosely real life
behaviour patterns. Usability wise participants ignored interface components and in-game tasks that
interfered with their game flow. Data reliability: unlike other methods that rely solely on participants
accounts, the game had the ability to compare between actions to questionnaire answers, and was
able to detect inconsistencies between people’s actions within the game and their accounts of their
actions. The ability to create spatial maps of event types enabled a fast way to visually analyze data.
The research concludes that MMORPGs can be used as an emergency simulation platform if: 1) its
duration fits the participants’ attention span (as a result aspects of human behaviour that happen over
a prolonged period of time will not be demonstrated); 2) the demographic composition of
participants fits that of the population examined by the simulation; 3) participants should be properly
reimbursed for their time; 4) it is known that participants’ in-game behaviour might be negatively
influenced by lack of real-life experience of similar events; 5) in-game rules and mechanisms are set
to filter out game abuse; 6) preliminary sessions are run to determine ideal attention span and data
skewing factor.
| Identifer | oai:union.ndltd.org:canterbury.ac.nz/oai:ir.canterbury.ac.nz:10092/4335 |
| Date | January 2010 |
| Creators | Schwartz, Ilan |
| Publisher | University of Canterbury. Natural Hazards Research Centre |
| Source Sets | University of Canterbury |
| Language | English |
| Detected Language | English |
| Type | Electronic thesis or dissertation, Text |
| Rights | Copyright Ilan Schwartz, http://library.canterbury.ac.nz/thesis/etheses_copyright.shtml |
| Relation | NZCU |
Page generated in 0.0017 seconds