Spatial Structure as an Element of Motivation in Location-Based Games

Abstract

Location-based games have emerged with the ubiquitous deployment of GPS-enabled devices. These games allow players to blend over the digital sphere with real-world whereabouts, implying that design choices made may affect the way players approach non-digital geographical spaces. This paper addresses scoring systems used to reward players for solving in-game tasks. We investigate how spatially structured scoring systems influence playing behaviours. The game we focus on is StreetComplete, an app that allows to collect and complement the OpenStreetMap database. We investigate indicators like walking speed, distance walked, and the game duration. Our results identify interesting effects spatial structure has on the game duration as well as the extent to which players are willing to explore an area geographically, hinting on interesting motivational elements of location-based games.

Full text

40. Wissenschaftlich-Technische Jahrestagung der DGPF in Stuttgart – Publikationen der DGPF, Band 29, 2020
290
Spatial Structure as an Element of Motivation
in Location- Based Games
HEINRICH LOREI1, BERNHARD HÖFLE2 & RENE WESTERHOLT3
Abstract: Location-based games have emerged with the ubiquitous deployment of GPS-
enabled devices. These games allow players to blend over the digital sphere with real-world
whereabouts, implying that design choices made may affect the way players approach non-
digital geographical spaces. This paper addresses scoring systems used to reward players for
solving in-game tasks. We investigate how spatially structured scoring systems influence
playing behaviours. The game we focus on is StreetComplete, an app that allows to collect
and complement the OpenStreetMap database. We investigate indicators like walking speed,
distance walked, and the game duration. Our results identify interesting effects spatial
structure has on the game duration as well as the extent to which players are willing to explore
an area geographically, hinting on interesting motivational elements of location-based games.
1 Introduction
Location-based games are characterised by a conflation of physical and digital spaces. They allow
players to enrich camera views with additional information, to switch attention between digital
and physical environments, and to situate and contextualise themselves on a scale beyond the
immediate context. Popular examples of location-based games include Pokémon Go, Ingress, and
more traditional Geocaching applications. In some cases, the entertainment factor of such games
is combined with scientific purposes like data collection, turning games into so-called serious
games. This is the case with apps like Kort and StreetComplete that allow users to playfully collect
OpenStreetMap data. This way, location-based games can be useful devices for the targeted
collection of missing geodata, and to augment or update existing repositories. This paper focuses
on the latter type of serious games, and on how the spatial arrangement of digital gaming elements
influences the playing behaviour in a real-world setting.
The gaming element we investigate is scoring systems attached to in-game tasks a player may have
to solve (e.g. answering questions about geographic features). Spatial structure refers to non-
random geographic patterns like clustering or dispersion. We compare the influence of such
spatially systematic deployments of scores to spatially random setups. The knowledge gained is
not only important for research, but also for game designers and players themselves. Game
designers get hints and detailed insights into the effects of spatial patterns in scoring systems. This
knowledge can be used to design games more efficiently. For players of location-based games, our
1 Metropolregion Rhein-Neckar GmbH, M1 4–5, D-68161 Mannheim, E-Mail: Heinrich.Lorei@m-r-n.com
2 Heidelberg University, Institute of Geography, 3DGeo Research Group, Im Neuenheimer Feld 368,
D-69120 Heidelberg, Germany, E-Mail: hoefle@uni-heidelberg.de
3 TU Dortmund University, Faculty of Spatial Planning, Spatial Modelling Lab, August-Schmidt-Straße 10,
D-44227 Dortmund, Germany, E-Mail: rene.westerholt@tu-dortmund.de

40. Wissenschaftlich-Technische Jahrestagung der DGPF in Stuttgart – Publikationen der DGPF, Band 29, 2020
291
results can be useful, as appropriately designed games allow players to better explore and learn
local areas, improve their physical activity and thus positively influence the gaming experience.
2 Motivation and Objective
Location-based games utilise real world geographies as playing fields. Their game experience is
thus strongly influenced by contextual spatial and non-spatial motivational factors. The
attractiveness of an area, for instance, can have an impact on how extensive players explore a
playing field (SCHEIDER & KIEFER 2018; WEBER 2017; HARTEVELD 2011). Similarly, the
integrability of a location-based game with everyday life can support gaining new perspectives on
otherwise routine places and activities (MATYAS 2011; MÄYRÄ & LANKOSKI 2009). Another
closely related aspect is the degree to which a game is adapted to local conditions. The strong
alignment of gaming elements with local geographical features that players may be familiar with
often improves the gaming experience through evoking a so-called ‘pride of place’ (SCHLIEDER
2014; WILL 2013; COLEMAN et al. 2009). The geography of a given playing field clearly influences
the motivational ability of a location-based game.
In this study, we investigate the influence of spatial patterning on the playing experience of
location-based games. The focus is on scoring systems used to reward players for completing
individual tasks during the game, such as annotating geographical features with attribute
information. It is known that location-based games are affected by the physical and mental
capabilities of players like their physical endurance, spatial cognition, and navigational capabilities
(SCHLIEDER 2014; WILL 2013; JACOB & COELHO 2011). Our underlying hypothesis is therefore
that players may (consciously or unconsciously) be affected by systematic structure found in the
spatial distribution of scores attached to in-game tasks. The game we use in our investigations is a
modified version of the open access application StreetComplete, and the tasks we look at are
attached to actual OpenStreetMap features.
3 Literature Review
Location-based games take place in real geographic environments. Therefore, the geography of an
area becomes an integral part of such games, including the related contextual factors (SCHLIEDER
et al. 2006). Several factors that influence game behaviour are of an ambient nature. For example,
it has been found that the time of day is important not only because a player's attention varies with
it, but also because other characteristics, such as the buzz of streets, are strongly correlated with
time (CARRIGY et al. 2010). Similarly, the actual physical and perceived conditions during the
game have an impact on the motivation of players. Weather conditions are one example, but stress
factors such as noisy environments (KNÖLL et al. 2014) and traffic-related air pollution are also
important in evaluating the gaming experience. Some contextual factors are directly related to the
morphology of a playing field. There is evidence that complex urban street layouts are
demotivating, as they not only make it difficult for players to find their way around, but also to
understand a local morphology (OLIVEIRA 2016, BEDÖ 2017). For similar reasons, the availability
of prominent landmarks makes a difference, too (BESTGEN et al. 2017; RICHTER & WINTER 2014).
Players can use them to orient themselves in potentially complex urban areas. This has an

H. Lorei, B. Höfle & R. Westerholt
292
influence, consciously or unconsciously, on how comfortable players of location-based games feel
with real-world playing fields.
Location-based games have the advantage that they can be played literally anywhere and 'on the
go'. This makes it easy for players to integrate these types of games into their everyday lives. One
aspect of this integration, which has proved to be particularly positive in terms of the geoliteracy
of the players, is the re-experience of seemingly familiar places (WEBER 2017; MATYAS 2011;
MÄRYÄ & LANKOSKI 2010). This gives the players a new perspective on their own everyday
activity spaces, which in turn enriches their everyday life with new experiences that they would
not have had without location-based games. This re-experiencing of familiar places is supported
by games that offer a broad distribution of game elements over an area (WEBER 2017; FRÄNTI et
al. 2017; SCHLIEDER 2014; WIll 2013; CELINO et al. 2012). The wider the geographical distribution
of the game elements, the stronger the positive influence a game can have on increasing the daily
level of exercise, exploring an area and gaining new experiences. Furthermore, adapting the game
elements to the local conditions of an area can further improve the game experience and thus
motivate the players to play even more. Our research provides additional evidence for the
importance of geography in location-based games. We shed light on the influence that a systematic
geospatial pattern in scores can have on player behaviour.
4 Methodology
4.1 The game StreetComplete
The game we used for our research in this paper is StreetComplete. This game is developed by
Tobias Zwick, a German software developer from Hamburg. StreetComplete is an open source
application and therefore the code is freely available on Github1. The main goal of the app is to
give also inexperienced OpenStreetMap users the opportunity to participate in the OpenStreetMap
project. For this reason, the questions (called quests) that are asked during the game are relatively
simple, such as dichotomous yes/no questions. New quest types are suggested by the wider
OpenStreetMap and StreetComplete communities and are collected and discussed on Github. Once
approved by the community, the quests are implemented by volunteers and automatically attached
to current issues on the map, for example based on missing tags. StreetComplete is an open-ended
game. This means that players do not work towards a specific goal, but rather altruistically collect
or improve data. The only form of reward is a counter of already solved quests. This is an
advantage for our study, as it ensures that none of our players (see Section 4.3) has previously used
a more complex form of scoring in the context of StreetComplete. The fact that StreetComplete is
open source has further allowed us to modify the game and tailor it to our needs. This has allowed
us to introduce various forms of spatially structured scoring systems.
4.2 Spatially-Structured Scoring System
Our analysis is based on two trial groups of players. While the task locations were the same for
both groups (uniformly distributed across the map to avoid visual clustering on the map), the scores
attached to those locations differed with respect to their spatial patterning. One group of subjects
played the game with a spatially randomised scoring system. That is, we generated scores on the
range [0, 100] from a spatially autoregressive model with the spatial parameter adjusted to ρ=0.01.

40. Wissenschaftlich-Technische Jahrestagung der DGPF in Stuttgart – Publikationen der DGPF, Band 29, 2020
293
Analogously, the spatially structured scores were generated from the same model but with ρ=0.99
leading to a strongly spatially autocorrelated scoring system. Figure 1 visualises both outlined
scoring systems and how the tasks are distributed across the investigation area. The scores
generated were not made visible on the map interface of the game to avoid introducing visual
confounding factors. This way, we have been able to isolate the effect of spatially structured scores
and to compare two groups of players under different spatial scenarios.
4.3 Subjects and Playing Field
Both scoring systems were deployed under controlled conditions. Our subjects comprised 40
geography students. This choice may limit the scope of the results to a specific target population,
but it homogenises with respect to demographics and educational level, as well as technical
proficiency. Those 40 subjects were randomly assigned to the two trial groups (20 each) playing
the two different spatial scoring systems. The playing field is comprised of an urban area of 1.3 km²
in size, located close to the centre of the city of Heidelberg (Fig. 2). The size chosen is motivated
by findings from a prior study recommending 1.5 km² to be optimal for playing times of 30 to 60
min (SCHLIEDER 2014), a duration we considered appropriate to test our hypothesis. The area is
diverse on a small scale comprising quiet zones like backyards but also busy roads. This allowed
us to diminish the effect of subjects being more likely to move to pleasant parts of the area only.
4.4 Indicators of the Playing Behaviour
We have tested a range of parameters of the players’ behaviours. All of those are indicators of the
players’ engagement and how motivated players remain during the game. Our indicators assessed
include playing time, distance walked (normalised by playing time), variety of road types explored,
numbers of tasks per minute solved, standard deviational ellipses, and a detour factor (ratio of
shortest and trajectory-derived path, see Fig. 3). We investigated these indicators for significant
mean differences between the two trial groups. Following Shapiro-Wilk tests, mean testing was
performed using the non-parametric Mann-Whitney U test to account for non-normality. The only
approximately normal variable is the numbers of tasks, which we tested by means of a t-test.
5 Results
A range of parameters did not differ significantly between the two trial groups (Tab. 1), but we
were able to disclose two systematic differences. The group exposed to spatially structured scores
played the game significantly longer than the control group (39 min vs 30 min, p=0.02). A second
though slightly weaker result is that the group with spatially structured scores explored the area
more extensively than the members of the control group did (detour factor of 5.02 vs 4.04, p=0.09,
t-test). These results indicate that the ways in which scoring systems are laid out spatially could be
an interesting way to tweak location-based games such that players remain motivated over time
and space during the play.

H. Lorei, B. Höfle & R. Westerholt
294
Fig. 1: Overview of the playing field. a) Endowed with a spatially random pattern of scores. b) Endowed
with a spatially structured pattern of scores; clusters of the highest scores highlighted in orange.
The background maps are based on OpenStreetMap data copyrighted by the OpenStreetMap
contributors and available from https://www.openstreetmap.org.

40. Wissenschaftlich-Technische Jahrestagung der DGPF in Stuttgart – Publikationen der DGPF, Band 29, 2020
295
Fig. 2: Location and geographical context of the playing field. a) Location of Heidelberg in the wider
context. b) Location of the playing field within the city of Heidelberg. The background maps are
based on data from DIVA-GIS (H
IJMANS
et al. 2001) and OpenStreetMap data copyrighted by
the OpenStreetMap contributors and available from https://www.openstreetmap.org.
Fig. 3: Illustration of the detour factor as the ratio of the lengths of the GPS tracks and the shortest
paths passing through all tasks visited. The background map is based on OpenStreetMap data
copyrighted by the OpenStreetMap contributors and available from
https://www.openstreetmap.org.
Interpreting our results in more detail reveals that some of them are not only statistically significant
but also notable with respect to their effect strengths. The players exposed to spatially structured
scores on average played the game longer than the control group. In addition, the same group
added the whole length of a shortest path distance to their distance walked according to our

H. Lorei, B. Höfle & R. Westerholt
296
assessed detour factors. Considering also the other indicators that do not differ significantly in a
statistical sense (which may be an effect of the limited numbers of participants in the groups)
reveals that the players exposed to a spatially structured scoring system consistently solved more
tasks, walked slower and longer distances on average, and traversed a higher diversity of different
road types. The results obtained are thus highly indicative of an interesting relation between the
players’ motivation and the spatial layout of the game.
The parameter that differed most statistically and in terms of the absolute mean deviation is the
duration of play. This is an important result because the game StreetComplete is unlimited in time
and players could end the game at any time. The fact that the players from the experimental group
played the game almost 33% longer than the players who were presented with the spatially random
point system ceteris paribus is therefore a strong indication of a systematic influence of the spatial
structure in the scores on the motivational aspects of the game. The random allocation of game
versions to the players additionally supports this finding, since no obvious distortions result from
the investigation structure. This result is significant beyond the case of location-based games. It
supports the previous evidence that shows how important it is for players to be able to understand
a game in order to be and keep motivated (LEE et al. 2017). Our results add a geographical
dimension in the form of spatial structure.
Another very important parameter that was tested is the detour factor. As with the duration of the
game, this parameter varied significantly and strongly between the two groups. While the longer
game duration indicates a general tendency towards higher game motivation, the detour factor
provides information about the motivation of the players not only to play the game but also to
explore the field. Compared to the control group, the players in the trial group added the length of
an entire "optimal" shortest path to their trajectories. This not only shows how important a
comprehensible point system is for the effective design of location-based games, but also points
to the relevance of geographically rewarding areas in general and beyond the present context, for
example regarding quality of stay. Players from the experimental group have become more
involved in the playing field, which, transferred to cases beyond playing, shows that they might
also be more likely to explore areas if there are incentives to do so. The results obtained here are
thus also of importance for urban planners and related researchers and practitioners.
Tab. 1: Mean values and their differences between the indicators calculated for the two trial groups.
Statistical significance is flagged for confidence levels α=0.10 (*) and α=0.05 (**).
Parameter Group with spatial
pattern
Randomised control
group
p-value
Playing time [min] 39.20 30.20 0.02**
Normalised distance walked [m] 2587 1954 0.46
Walking speed [m/s] 0.92 0.99 0.40
Area: standard deviational ellipse [km²] 0.114 0.102 0.86
Variety of road types traversed [%] 79 72 0.29
Tasks solved [1/min] 17.80 12.55 0.38
Detour factor 5.02 4.04 0.09*

40. Wissenschaftlich-Technische Jahrestagung der DGPF in Stuttgart – Publikationen der DGPF, Band 29, 2020
297
6 Conclusions
Based on our results obtained, we conclude that a spatially comprehensible layout is likely to
support higher levels of motivation with location-based games. In this sense, our results support
and add to prior results achieved in non-spatial settings demonstrating the importance of traceable
scoring systems that players can make sense of (either consciously or subconsciously) (LEE et al.
2017). Future research should investigate other types of spatial structures in scoring systems to
identify optimal layouts for game designs. Further, other gamification elements beyond scoring
systems may be tested for spatial effects in similar ways. This way, it will be possible to optimise
location-based games and to better utilise them for research purposes such as data collection. Also,
research in the nexus of gamification and spatial analysis may contribute to the revealing of
interesting, general psycho-geographic mechanisms.
Notes
1 https://github.com/westnordost/StreetComplete
7 References
BEDÖ, V., 2017: Size and Shape of the Playing Field: Research Through Game Design Approach.
Playable Cities, Nijholt, A. (ed.), Springer, Singapore, 67-86.
BESTGEN, A. K., EDLER, D., KUCHINKE, L. & DICKMANN, F., 2017: Analyzing the effects of VGI-
based landmarks on spatial memory and navigation performance. KI-Künstliche Intelligenz
31(2), 179-183.
CARRIGY, T., NALIUKA, K., PATERSON, N. & HAAHR, M., 2010: Design and evaluation of player
experience of a location-based mobile game. Proceedings of the 6th Nordic Conference on
Human-Computer Interaction: Extending Boundaries, ACM, 92-101.
CELINO, I., CERIZZA, D., CONTESSA, S., CORUBOLO, M., DELL’AGLIO, D., VALLE, E. D., FUMEO, S.
& PICCININI, F., 2012: Urbanopoly: collection and quality assesment of geo-spatial linked
data via a human computation game. Proceedings of the 10th Semantic Web Challenge, 148-
163.
COLEMAN, J.D., GEORGIADOU, Y. & LABONTE, J., 2009: Volunteered Geographic Information: The
Nature and Motivation of Produsers. International Journal of Spatial Data Infrastructures
Research 4(1), 332-358.
FRÄNTI, P., MARIESCU-ISTODOR, R. & SENGUPTA, L., 2017: O-Mopsi: Mobile orienteering game
for sightseeing, exercising, and education. ACM Transactions on Multimedia Computing,
Communications, and Applications 13(4), 56.
HARTEVELD, C., 2011: Triadic game design: Balancing reality, meaning and play. Springer Science
& Business Media, Heidelberg.
HIJMANS, R. J., GUARINO, L., CRUZ, M. & ROJAS, E., 2001. Computer tools for spatial analysis of
plant genetic resources data: 1. DIVA-GIS. Plant Genetic Resources Newsletter 127, 15-19.
JACOB, J.T.P.N. & COELHO, A.F., 2011: Issues in the development of location-based games.
International Journal of Computer Games Technology, 2011, 495437.

H. Lorei, B. Höfle & R. Westerholt
298
KNÖLL, M., NEUHEUSER, K., VOGT, J. & RUDOLPH-CLEFF, A., 2014: Einflussfaktoren der gebauten
Umwelt auf wahrgenommene Aufenthaltsqualität während der Nutzung städtischer Räume.
Umweltpsychologie 18(2), 84-103.
LEE, C.-I., CHEN, I.-P., HSIEH, C.-N. & LIAO, C.-N., 2017: Design aspects of scoring systems in
game. Art and Design Review 5(1), 26-43.
MATYAS, S., 2011: Gemeinschaftliche qualitätsgesicherte Erhebung und semantische Integration
von raumbezogenen Daten. University of Bamberg Press, Bamberg.
MÄYRÄ, F. & LANKOSKI, P., 2010: Play in hybrid reality: Alternative approaches to game design.
Digital Cityscapes: Merging digital and urban playspaces, De Souza e Silva, A. & Sutoko,
D. (eds.), Peter Lang Publishers, New York, US, 129-147.
OLIVEIRA, V., 2016: Urban morphology: an introduction to the study of the physical form of cities.
Springer, Cham, Switzerland.
RICHTER, K. F. & WINTER, S., 2014. Cognitive aspects: How people perceive, memorize, think and
talk about landmarks. Landmarks, Richter, K. F., & Winter, S. (eds.), Springer, Cham,
Switzerland, 41-108.
SCHEIDER, S. & KIEFER, P., 2018: (Re-) Localization of location-based games. Geogames and
geoplay, Ahlqvist, O. & Schlieder, C. (eds.), Springer, Cham, Switzerland, 131-159.
SCHLIEDER, C., 2014: Geogames Organizer’s Guide v1.0. Technical report, Universität Bamberg,
Germany.
SCHLIEDER, C., 2018: Geogames – Gestaltungsaufgaben und geoinformatische Lösungsansätze.
Informatik-Spektrum 37(6), 567-574.
SCHLIEDER, C., KIEFER, P. & MATYAS, S. 2006. Geogames: Designing location-based games
from classic board games. IEEE Intelligent Systems 21(5), 40-46.
WEBER, J., 2017: Designing engaging experiences with location-based augmented reality games
for urban tourism environments. Dissertation, Bournemouth University, UK.
WILL, C., 2013: A Pattern Language for Designing Location-based Games. Dissertation, RWTH
Aachen, Germany.