M. Cavazza and S. Donikian (Eds.): ICVS 2007, LNCS 4871, pp. 38
–
50, 2007.
© Springer-Verlag Berlin Heidelberg 2007
Story Manager in ‘Europe 2045’ Uses Petri Nets
Cyril Brom
1
, Vít Šisler
2
, and Tomáš Holan
1
1
Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic
{brom,holan}@ksvi.mff.cuni.cz
2
Charles University, Faculty of Philosophy, Prague, Czech Republic
vsisler@gmail.com
Abstract. Europe 2045 is an on-line multi-player strategy game aimed at
education of high-school students in economics, politics, and media studies.
The essential feature of the game is that players face various simulated
scenarios and crises addressing contemporary key issues of the unified Europe.
These scenarios are branching and can evolve in a parallel manner. In this
paper, we present a technique for specifying plots of these scenarios, which
underpins the story manager of Europe 2045. The technique is based on a
modification of Petri Nets. We also detail one particular scenario concerning
the current crisis in Darfur. On a general level this paper discusses the strengths
and weaknesses of implementation of Petri Nets in virtual storytelling.
1 Introduction
The idea of using computer games to support training and learning objectives is more
than 30 years old [4]. Recent works have explored the potentialities of commercial
strategy games and simulations in formal education and their alleged advantages over
classical e-learning and edutainment tools, e.g. [6]. Indeed, many of such games have
been experimentally integrated to formal curricula in the last four years. Perhaps the
most prominent case studies have been conducted with The Sims 2, Civilization III,
and Europe Universalis II [6, 12, 15], but other attempts exists as well. The results
from these pilots are promising, but also ambiguous in some aspects, e.g. [15]. Hence,
so called “serious” or “educational” games are starting to achieve increasing amount
of attention. These games are, contrary to commercial games, intentionally developed
as educational tools, which makes their integration into formal education easier. For
example, a role-playing game prototype Global Conflicts: Palestine has been recently
evaluated in a Denmark high-school with positive outcome [7]. Another studies are
being conducted, including FearNot!, an anti-bullying educational game [1], and
Revolution, a multi-player educational role-playing game concerning American War
of Independence [5, 8].
As a part of European funded project “Integration of IT Tools into Education of
Humanities” we develop an educational game Europe 2045, which is likely the first
on-line multi-player strategy game worldwide aimed at education of high-school
students in economics, politics, and media studies. The implementation part is
finished and the game is presently being tested. Five preliminary studies have been
already carried out, each with about 10 high-school or undergraduate university
Story Manager in ‘Europe 2045’ Uses Petri Nets 39
students. A large-scale pilot evaluation is planed for November 2007 in a high-school
in Prague, Czech Republic. The game is intended to be fully applied in spring 2008.
Europe 2045 features three layers of game-play. Each student (1) represents one
EU member state in the game and is responsible for its governmental policies,
economical development, and social issues. Additionally, in cooperation with the
other players, (2) he or she is engaged in setting politics of the whole EU.
Nevertheless, the essential feature of the game is that (3) each player faces various
simulated scenarios and crises addressing contemporary key issues of the unified
Europe, including migration, population aging, international relations, and energy
independence. Not only have these scenarios a strong educational potential, but also
they introduce storytelling into the game. Storytelling has played an important role in
humanities education since the advent of formal schooling ([4]). Stories help to build
a learning context, through them the students can better understand the problematic,
they increase their involvement, and consequently their motivation.
Specifying plots of stories and controlling the course of a game in accordance with
these plots is a well known problem. It was indeed one of the most challenging goals
we faced during the development. Essentially, the game had to be designed in order to
meet the following requirements:
a) the story plots to be branching,
b) the story episodes to be both global, i.e. concerning the whole Europe, and local,
i.e. concerning a particular state or a set of states,
c) the episodes to can happen in parallel, because we have more than 20 countries,
which could be played simultaneously, each having defined different episodes,
d) the episodes to be triggered by various initial conditions depending on the time,
EU economy etc.,
e) the technique for specification of the plots to be intuitive enough for a high
school teacher or another user (typically an undergraduate university student of
humanities) to be able to design new scenarios for the game.
Finally, we have chosen a modification of Petri Nets [2] as the plot specification
technique. Although this modification is used for Europe 2045, the technique is quite
universal and can be use in other applications as well. The goal of this paper is to
present this technique and discuss its strengths and weaknesses. We first detail the
game Europe 2045 in Section 2, focusing on how the stories are narrated in the game.
In Section 3, we review previous research on methods of controlling a story in games
and storytelling applications in general. Section 4 details the Petri Nets modification
and its implementation. Section 5 demonstrates a part of a scenario concerning the
Darfur crisis. Section 6 concludes.
2 ‘Europe 2045’
This section details the game Europe 2045. We first describe the game from the
perspective of a player, focusing on its storytelling aspects, and then overview the
technical background.
Europe 2045 is an on-line multiplayer game in which each student (i.e. a player)
governs one state by setting its policies, taxes, and subsidies while discussing
40 C. Brom, V. Šisler, and T. Holan
European and global issues with other players. The game contains economical and
social model which simulates population aging, migration, evolution of the market,
transfers of industry and services, changes in environment, moods of citizens, and a
substantial number of other variables describing particular states and European Union
as a whole (e.g. culture, infrastructure, education, etc.).
On the European level, all the players are encouraged by the game to take active
part in decision making. The narrative structure of Europe 2045 serves for three
purposes. First, it introduces new topics and agenda for students’ discussions. Second,
unfolding new events in accordance with players’ previous decisions, it serves as a
global feedback for the students and as a method for sharpening the discussion. Both
these kinds of events are global, i.e. they are common for all the players and concern
EU as well as international issues (e.g. conflict in Darfur has intensified). The third
class of events provides individual players with a feedback about the results of their
previous actions concerning their own states; hence, these events are local (e.g.
citizens in France protest against university fees, or unemployment in Czech Republic
has reached 15%).
The game proceeds in rounds, one round is one game year. An atomic “beat” of a
scenario is called an affair. It is an event that takes place in one round and can be
triggered by players’ actions or results from the economical and social model or
affairs from previous rounds. An affair is communicated to the player via a textual
description in the game newspaper (NP news item) or via a short animation in TV,
which is being displayed at the beginning of every round (TV news item). In some
cases, an affair also has an impact on the economical and social model, i.e. it
influences state of a country or the whole EU. Typically, an affair can result in
increasing the EU budget, increasing the level of pollution in particular states,
crippling agriculture production, etc.
Some affairs introduce issues that require decision to be taken by the players (e.g.
accepting another state’s proposal, sending humanitarian mission to the area of a
conflict, etc.). These decisions are intended to be taken during a discussion, typically
in the class under the teacher’s supervision, and voted through a ballot. One affair
often triggers more ballots, each constituting precisely formulated question (“Do you
vote for sending European humanitarian mission to Darfur area?”) with three possible
answers (yes/no/abstain). The ballots chosen by the game designers aim to cover all
the main possible solutions usually proposed by real politics in similar cases. When
the answers can not be schematized to the yes or no option, the ballot contains
number (3-4) of more detailed solutions. The decision chosen by the players
influences the economical and social model and the affairs to be triggered in the next
round.
The game offers more different campaigns to be played, each of them focusing on
different problematic (e.g. energy independence, international relations,
environment). For each campaign, specific affairs and a scenario describing relations
between them have to be designed. New campaign also comprises distinctive
animations for the TV, articles for the newspaper, items for the in-game
encyclopaedia, teachers manual, and handouts for students.
Technically, the game is a client-server application; the students play the game via
the Internet (Fig. 1). The server part comprises PHP scripts generating the game
interface, the story manager written in PHP as well, and the social-economical
Story Manager in ‘Europe 2045’ Uses Petri Nets 41
simulation, which is written in Java. Almost all parts of the interface are programmed
in Flash (see Fig. 2). The social-economical simulation features a simplified model of
EU economy. Technically, the model is a multi-agent simulation [16], where each
agent is either a country, or an abstract representation of an EU industry, like travel
industry, mining industry, agriculture etc. In a simplified fashion, at the end of each
round, an agent–country computes next state of the country, while an agent–industry
carries out decisions in which country to build new factories, mines etc. based on
particular variables of the countries (e.g. mining industry agent would prefer countries
with low environmental tax, travel industry agent would prefer countries with nice
environment, high culture and developed infrastructure). We remark that this
simulation is coarse grained in the sense that it does not feature human-like agents.
Fig. 1. Architecture of Europe 2045 2a
2b 2c
Fig. 2. a) The TV news. b) The interface, through which the player governs its country. c) The
balloting interface.
3 Related Work
The issue of generating/controlling stories in games and storytelling applications is
notoriously known. Most techniques come from games and experimental simulations
featuring human-like actors. As said above, Europe 2045 does not employ such
actors. Rather, the story events are more abstract; they deal with whole populations,
with a country economy etc. However, formally, the problem is very similar. We
remind that we had several requirements on the storytelling technique in Europe 2045
42 C. Brom, V. Šisler, and T. Holan
ranging from parallelism of the stories to the technique to be of use to undergraduate
university non-IT students (see Sec. 1).
A well known branch of techniques for specifying plots are deterministic finite-
state machines (dFSMs) [e.g. 13, 14]. Each state represents a story episode, and a
transition is a trigger that detects the end of the episode and starts a next one. Natural
advantage of dFSMs is that they are formal, and yet graphical (Fig. 3), which makes
them easily intelligible. However, a classical dFSMs was not suitable for us, since
they cannot cope with the issues of parallelism (req. b), c)). On the other hand, non-
deterministic FSMs can cope with it, but they are not easily comprehensible (e)).
Similarly to dFSMs, we encountered these parallelism difficulties when considering
adopting the “beat approach” by Mateas [9].
In the field of emergent narrative, planning formalism is often used [1, 3, 11]. This
technique can cope well with the requirements a), b), c), d), but it is not too friendly
for a non-AI expert (req. e)). Since we are interested in pre-specified plots but not in
automatic story construction (because the story must fit into the formal curricula), to
use the HTN (“hierarchical-task network”) formalism would be like using a
sledgehammer to crack a walnut, with having the unintelligibility disadvantage. To
tackle e), one could introduce a “presentation layer” for an HTN system to disguise
the underlying representation and develop an authoring interface; however, this is
time-consuming activity.
We needed something that would have natural comprehensibility advantage like
the dFSMs, but could cope with the parallelism at the same time. The best candidate
technique we found was Petri Nets, which is a specification technique frequently used
in software engineering. Petri Nets have been already employed in storytelling.
Natkin & Vega [10] used them to a retrospective analysis of a computer game story.
In our previous work [2], we used it to prototype a story plot of a large simulation
featuring human-like agents. However, none of these work implemented a story
manager for a real full-fledged game, which is the case of Europe 2045.
Actually, many variants of Petri Nets exist. For our purposes, we specified our own
modification, which will be detailed in the next section.
Fig. 3. Story plots as dFSMs. The linear plot is on the left, the branching on the right.
4 Petri Nets in ‘Europe 2045’
This section gives description of our modification of Petri Nets. Generally, Petri Nets
consist of containers (or places, represented by a circle: ), tokens (“the pellets”: ),
actions (or transitions,
), and transition function ( ). The containers contain the
tokens. If a sufficient number of tokens is contained in specific containers, an action
is triggered. After firing an action, the tokens that helped to fire this action are
