AI Communication with Technology and Humans

RESEARCH ARTICLE

Even good bots fight: The case of Wikipedia

Milena Tsvetkova 1 , Ruth Garcı́a-Gavilanes

1 , Luciano Floridi

1,2 , Taha Yasseri

1,2*

1 Oxford Internet Institute, University of Oxford, Oxford, United Kingdom, 2 Alan Turing Institute, London,

United Kingdom

* [email protected]

Abstract

In recent years, there has been a huge increase in the number of bots online, varying from

Web crawlers for search engines, to chatbots for online customer service, spambots on

social media, and content-editing bots in online collaboration communities. The online world

has turned into an ecosystem of bots. However, our knowledge of how these automated

agents are interacting with each other is rather poor. Bots are predictable automatons that

do not have the capacity for emotions, meaning-making, creativity, and sociality and it is

hence natural to expect interactions between bots to be relatively predictable and unevent-

ful. In this article, we analyze the interactions between bots that edit articles on Wikipedia.

We track the extent to which bots undid each other’s edits over the period 2001–2010,

model how pairs of bots interact over time, and identify different types of interaction trajecto-

ries. We find that, although Wikipedia bots are intended to support the encyclopedia, they

often undo each other’s edits and these sterile “fights” may sometimes continue for years.

Unlike humans on Wikipedia, bots’ interactions tend to occur over longer periods of time and

to be more reciprocated. Yet, just like humans, bots in different cultural environments may

behave differently. Our research suggests that even relatively “dumb” bots may give rise

to complex interactions, and this carries important implications for Artificial Intelligence

research. Understanding what affects bot-bot interactions is crucial for managing social

media well, providing adequate cyber-security, and designing well functioning autonomous

vehicles.

Introduction

In August 2011, Igor Labutov and Jason Yosinski, two PhD students at Cornell University, let

a pair of chat bots, called Alan and Sruthi, talk to each other online. Starting with a simple

greeting, the one-and-a-half-minute dialogue quickly escalated into an argument about what

Alan and Sruthi had just said, whether they were robots, and about God [1]. The first ever con-

versation between two simple artificial intelligence agents ended in a conflict.

A bot, or software agent, is a computer program that is persistent, autonomous, and reactive

[2,3]. Bots are defined by programming code that runs continuously and can be activated by

itself. They make and execute decisions without human intervention and perceive and adapt

to the context they operate in. Internet bots, also known as web bots, are bots that run over the

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 1 / 13

a1111111111

a1111111111

a1111111111

a1111111111

a1111111111

OPEN ACCESS

Citation: Tsvetkova M, Garcı́a-Gavilanes R, Floridi

L, Yasseri T (2017) Even good bots fight: The case

of Wikipedia. PLoS ONE 12(2): e0171774.

doi:10.1371/journal.pone.0171774

Editor: Sergio Gómez, Universitat Rovira i Virgili,

SPAIN

Received: November 16, 2016

Accepted: January 25, 2017

Published: February 23, 2017

Copyright: © 2017 Tsvetkova et al. This is an open access article distributed under the terms of the

Creative Commons Attribution License, which

permits unrestricted use, distribution, and

reproduction in any medium, provided the original

author and source are credited.

Data Availability Statement: All relevant data are

available from figshare at 10.6084/m9.figshare.

4597918.

Funding: This work has received funding from the

European Union’s Horizon 2020 research and

innovation program under grant agreement No.

645043: HUMANE.

Competing interests: The authors have declared

that no competing interests exist.

Internet. They appeared and proliferated soon after the creation of the World Wide Web [4].

Already in 1993, Martijn Koster published “Guidelines to robot writers,” which contained sug-

gestions about developing web crawlers [5], a kind of bot. Eggdrop, one of the first known

Internet Relay Chat bots, started greeting chat newcomers also in 1993 [6]. In 1996, Fah-Chun

Cheong published a 413-page book, claiming to have a current listing of all bots available on

the Internet at that point in time. Since then, Internet bots have proliferated and diversified

well beyond our ability to record them in an exhaustive list [7,8]. As a result, bots have been

responsible for an increasingly larger proportion of activities on the Web. For example, one

study found that 25% of all messages on Yahoo! chat over a period of three months in 2007

were sent by spam bots [9]. Another study discovered that 32% of all tweets made by the most

active Twitter users in 2009 were generated by bots [10], meaning that bots were responsible

for an estimated 24% of all tweets [11]. Further, researchers estimated that bots comprise

between 4% and 7% of the avatars on the virtual world Second Life in 2009 [12]. A media ana-

lytics company found that 54% of the online ads shown in thousands of ad campaigns in 2012

and 2013 were viewed by bots, rather than humans [13]. According to an online security com-

pany, bots accounted for 48.5% of website visits in 2015 [14]. Also in 2015, 100,000 accounts

on the multi-player online game World of Warcraft (about 1% of all accounts) were banned

for using bots [15]. And in the same year, a database leak revealed that more than 70,000

“female” bots sent more than 20 million messages on the cheater dating site Ashley Madison

[16].

As the population of bots active on the Internet 24/7 is growing fast, their interactions are

equally intensifying. An increasing number of decisions, options, choices, and services depend

now on bots working properly, efficaciously, and successfully. Yet, we know very little about

the life and evolution of our digital minions. In particular, predicting how bots’ interactions

will evolve and play out even when they rely on very simple algorithms is already challenging.

Furthermore, as Alan and Sruthi demonstrated, even if bots are designed to collaborate, con-

flict may occur inadvertently. Clearly, it is crucial to understand what could affect bot-bot

interactions in order to design cooperative bots that can manage disagreement, avoid unpro-

ductive conflict, and fulfill their tasks in ways that are socially and ethically acceptable.

There are many types of Internet bots (see Table 1). These bots form an increasingly com-

plex system of social interactions. Do bots interact with each other in ways that are comparable

Table 1. Categorization of Internet bots according to the intended effect of their operations and the kind of activities they perform, including some

familiar examples for each type.

Benevolent Malevolent

Collect

information

• Web crawlers

• Bots used by researchers

• Spam bots that collect e-mail addresses

• Facebook bots that collect private information

Execute actions • Anti-vandalism bots on Wikipedia

• Censoring and moderating bots on chats and

forums

• Auction-site bots

• High-frequency trading algorithms

• Gaming bots

• DDoS attack bots

• Viruses and worms

• Clickfraud bots that increase views of online ads and YouTube videos

Generate content • Editing bots on Wikipedia

• Twitter bots that create alerts or provide content

aggregation

• Spam bots that disseminate ads

• Bot farms that write positive reviews and boost ratings on Apple App Store,

YouTube, etc.

Emulate humans • Customer service bots

• @DeepDrumpf and poet-writing bots on Twitter

• AI bots, e.g. IBM’s Watson

• Social bots involved in astroturfing on Twitter

• Social bots on the cheater dating site Ashley Madison

doi:10.1371/journal.pone.0171774.t001

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 2 / 13

to how we humans interact with each other? Bots are predictable automatons that do not have

the capacity for emotions, meaning-making, creativity, and sociality [17]. Despite recent

advances in the field of Artificial Intelligence, the idea that bots can have morality and culture

is still far from reality. Today, it is natural to expect interactions between bots to be relatively

predictable and uneventful, lacking the spontaneity and complexity of human social interac-

tions. However, even in such simple contexts, our research shows that there may be more simi-

larities between bots and humans than one may expect. Focusing on one particular human-bot

community, we find that conflict emerges even among benevolent bots that are designed to

benefit their environment and not fight each other, and that bot interactions may differ when

they occur in environments influenced by different human cultures.

Benevolent bots are designed to support human users or cooperate with them. Malevolent

bots are designed to exploit human users and compete negatively with them. We have classi-

fied high-frequency trading algorithms as malevolent because they exploit markets in ways

that increase volatility and precipitate flash crashes. In this study, we use data from editing

bots on Wikipedia (benevolent bots that generate content).

We study bots on Wikipedia, the largest free online encyclopedia. Bots on Wikipedia are

computer scripts that automatically handle repetitive and mundane tasks to develop, improve,

and maintain the encyclopedia. They are easy to identify because they operate from dedicated

user accounts that have been flagged and officially approved. Approval requires that the bot

follows Wikipedia’s bot policy.

Bots are important contributors to Wikipedia. For example, in 2014, bots completed about

15% of the edits on all language editions of the encyclopedia [18]. In general, Wikipedia bots

complete a variety of activities. They identify and undo vandalism, enforce bans, check spell-

ing, create inter-language links, import content automatically, mine data, identify copyright

violations, greet newcomers, and so on [19]. Our analysis here focuses on editing bots, which

modify articles directly. We analyze the interactions between bots and investigate the extent to

which they resemble interactions between humans. In particular, we focus on whether bots

disagree with each other, how the dynamics of disagreement differ for bots versus humans,

and whether there are differences between bots operating in different language editions of

Wikipedia.

To measure disagreement, we study reverts. A revert on Wikipedia occurs when an editor,

whether human or bot, undoes another editor’s contribution by restoring an earlier version of

the article. Reverts that occur systematically indicate controversy and conflict [20–22]. Reverts

are technically easy to detect regardless of the context and the language, so they enable analysis

at the scale of the whole system.

Our data contain all edits in 13 different language editions of Wikipedia in the first ten

years after the encyclopedia was launched (2001–2010). The languages represent editions of

different size and editors from diverse cultures (see Materials and Methods for details). We

know which user completed the edit, when, in which article, whether the edit was a revert and,

if so, which previous edit was reverted. We first identified which editors are humans, bots, or

vandals. We isolated the vandals since their short-lived disruptive activity exhibits different

time and interaction patterns than the activity of regular Wikipedia editors.

Results

Bots constitute a tiny proportion of all Wikipedia editors but they stand behind a significant

proportion of all edits (Fig 1A and 1B). There are significant differences between different lan-

guages in terms of how active bots are. From previous research, we know that, in small and

endangered languages, bots are extremely active and do more than 50% of the edits, sometimes

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 3 / 13

up to 100% [19]. Their tasks, however, are mainly restricted to adding links between articles

and languages. In large and active languages, the level of bot activity is much lower but also

much more variable.

Compared to humans, a smaller proportion of bots’ edits are reverts and a smaller propor-

tion get reverted (Fig 1C and 1D). In other words, bots dispute others and are disputed by oth-

ers to a lesser extent than humans. Since 2001, the number of bots and their activity has been

increasing but at a slowing rate (S1 Fig). In contrast, the number of reverts between bots has

been continuously increasing (Fig 2A). This would suggest that bot interactions are not

becoming more efficient. We also see that the proportion of mutual bot-bot reverts has

remained relatively stable, perhaps even slightly increasing over time, indicating that bot own-

ers have not learned to identify bot conflicts faster (Fig 2B).

In general, bots revert each other a lot: for example, over the ten-year period, bots on

English Wikipedia reverted another bot on average 105 times, which is significantly larger

than the average of 3 times for humans (S1 Table). Bots on German Wikipedia revert each

other to a much lesser extent than other bots (24 times on average). Bots on Portuguese Wiki-

pedia, in contrast, fight the most, with an average of 185 bot-bot reverts per bot. This striking

difference, however, disappears when we account for the fact that bots on Portuguese Wikipe-

dia edit more than bots on German Wikipedia. In general, since bots are much more active

Fig 1. The proportion of Wikipedia editors who are human, vandals, and bots and the type of editorial

activity in which they are involved. A language edition to the left has a higher total number of edits than one

to the right. (A) Bots comprise a tiny proportion of all Wikipedia users, usually less than 0.1% (not visible in the

figure). (B) However, bots account for a significant proportion of the editorial activity. The level of bot activity

significantly differs between different language editions of Wikipedia, with bots generally more active in

smaller editions. (C) A smaller proportion of bots’ edits are reverts compared to humans’ edits. (D) A smaller

proportion of bots’ edits get reverted compared to humans’ edits. Since by our definition, vandals have all of

their edits reverted, we do not show them in this figure.

doi:10.1371/journal.pone.0171774.g001

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 4 / 13

editors than humans, the higher number of bot-bot reverts does not mean that bots fight more

than humans. In fact, the proportion of bots’ edits that are reverts is smaller for bots than for

humans (Fig 1C). This proportion is highest for bots in the English and the Romance-language

editions (Spanish, French, Portuguese, and Romanian). Interestingly, although bots in these

languages revert more often compared to bots in other languages, fewer of these reverts are for

another bot (S2 Fig).

Our analysis focuses on interactions in dyads over time. We model the interaction trajecto-

ries in two-dimensional space, where the x-axis measures time and the y-axis measures how

many more times the first editor has reverted the second compared to the second reverting the

first (Fig 3). We analyze three properties of the trajectories: latency, imbalance, and reciprocity.

Latency measures the average steepness of the interaction trajectory, imbalance measures the

distance between the x-axis and the last point of the trajectory, and reciprocity measures the trajectory’s jaggedness (see Materials and Methods below for definitions).

Analyzing the properties of the interaction trajectories suggests that the dynamics of dis-

agreement differ significantly between bots and humans. Reverts between bots tend to occur at

a slower rate and a conflict between two bots can take place over longer periods of time, some-

times over years. In fact, bot-bot interactions have different characteristic time scale than

human-human interactions (S3 Fig). The characteristic average time between successive

reverts for humans is at 2 minutes, 24 hours, or 1 year. In comparison, bot-bot interactions

have a characteristic average response of 1 month. This difference is likely because, first, bots

systematically crawl articles and, second, bots are restricted as to how often they can make

Fig 2. The number of bot reverts executed by another bot and the proportion of unique bot-bot pairs

that have at least one reciprocated revert for the period 2001–2010. (A) Generally, the number of bot-bot

reverts has been increasing. (B) However, the proportion of reciprocated reverts has not been decreasing

(error bars correspond to one standard error). This suggests that disagreement between bots is not becoming

less common.

doi:10.1371/journal.pone.0171774.g002

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 5 / 13

edits (the Wikipedia bot policy usually requires spacing of 10 seconds, or 5 for anti-vandalism

activity, which is considered more urgent). In contrast, humans use automatic tools that report

live changes made to a pre-selected list of articles [24,25]; they can thus follow only a small set

of articles and, in principle, react instantaneously to any edits on those.

Bots also tend to reciprocate each other’s reverts to a greater extent. In contrast, humans

tend to have highly unbalanced interactions, where one individual unilaterally reverts another

one (S4 and S5 Figs).

We quantify these findings more precisely by identifying different types of interaction tra-

jectories and counting how often they occur for bots and for humans, as well as for specific

languages. To this end, we use k-means clustering on the three properties of the trajectories

(latency, imbalance, and reciprocity) and on all bot-bot and human-human interactions longer

than five reverts (the results are substantively similar without the length restriction). We do

not claim that the clusters are natural to the data; rather, we use the clusters to compare the

interactions of the different groups.

Fig 3. Typical interaction trajectories for bot-bot and human-human pairs in English Wikipedia in the

period 2001–2010. The interaction trajectories are constructed as follows: starting from yo = 0, yt = yt-1 + 1 if i

reverts j and yt = yt-1 − 1 if j reverts i at time t; the labels i and j are assigned so that y >= 0 for the majority of the ij interaction time; to compress the extremes, we scaled the y-axis to the power of 0.5. The panels show the

trajectories of 200 pairs randomly sampled from those who have exchanged more than five reverts. In

addition, we highlight the four longest trajectories in the sample from each of the four trajectory types we

identify. Compared to human-human interactions, bot-bot interactions occur at a slower rate and are more

balanced, in the sense that reverts go back and forth between the two editors.

doi:10.1371/journal.pone.0171774.g003

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 6 / 13

The algorithm suggested that the data can be best clustered in four trajectory types (S6 Fig):

• Fast unbalanced trajectories. These trajectories have low reciprocity and latency and high

imbalance. They look like smooth vertical lines above the x-axis.

• Slow unbalanced trajectories. These trajectories have low reciprocity and high latency and

imbalance. They look like smooth diagonal lines above the x-axis.

• Somewhat balanced trajectories. These trajectories have intermediate imbalance and reci-

procity. They are somewhat jagged and cross the x-axis.

• Well balanced trajectories. These trajectories have low imbalance and high reciprocity.

They are quite jagged and centered on the x-axis.

Looking at the prevalence of these four types of trajectories for bots and humans and across

languages, we confirm the previous observations: bot-bot interactions occur at a slower rate

and are more balanced, in the sense that reverts go back and forth between the two bots (Fig

4). Further, we find that bot-bot interactions are more balanced in smaller language editions of

Wikipedia. This could be due to the fact that bots are more active in smaller editions and

hence, interactions between them are more likely to occur. Less intuitively, however, this

observation also suggests that conflict between bots is more likely to occur when there are

fewer bots and when, common sense would suggest, coordination is easier.

Discussion

Our results show that, although in quantitatively different ways, bots on Wikipedia behave and

interact as unpredictably and as inefficiently as the humans. The disagreements likely arise

from the bottom-up organization of the community, whereby human editors individually

create and run bots, without a formal mechanism for coordination with other bot owners.

Fig 4. The prevalence of the four types of trajectories for bots and humans and for different language editions of

Wikipedia. The darker the shading of the cell, the higher the proportion for that type of trajectory for the language. Bot-bot

interactions occur at a slower rate and are more balanced, in the sense that reverts go back and forth between the two bots.

Further, bot-bot interactions are more balanced in smaller language editions of Wikipedia.

doi:10.1371/journal.pone.0171774.g004

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 7 / 13

Delving deeper into the data, we found that most of the disagreement occurs between bots that

specialize in creating and modifying links between different language editions of the encyclo-

pedia. The lack of coordination may be due to different language editions having slightly dif-

ferent naming rules and conventions.

In support of this argument, we also found that the same bots are responsible for the major-

ity of reverts in all the language editions we study. For example, some of the bots that revert

the most other bots include Xqbot, EmausBot, SieBot, and VolkovBot, all bots specializing in

fixing inter-wiki links. Further, while there are few articles with many bot-bot reverts (S7 Fig),

these articles tend to be the same across languages. For example, some of the articles most con-

tested by bots are about Pervez Musharraf (former president of Pakistan), Uzbekistan, Estonia,

Belarus, Arabic language, Niels Bohr, Arnold Schwarzenegger. This would suggest that a sig-

nificant portion of bot-bot fighting occurs across languages rather than within. In contrast, the

articles with most human-human reverts tend to concern local personalities and entities and

tend to be unique for each language [26].

Our data cover a period of the evolution of Wikipedia when bot activity was growing. Evi-

dence suggests that this period suddenly ended in 2013 (http://stats.wikimedia.org/EN/

PlotsPngEditHistoryTop.htm). This decline occurred because at the beginning of 2013 many

language editions of Wikipedia started to provide inter-language links via Wikidata, which is a

collaboratively edited knowledge base intended to support Wikipedia. Since our results were

largely dictated by inter-language bots, we believe that the conflict we observed on Wikipedia

no longer occurs today. One interesting direction for future research is to investigate whether

the conflict continues to persist among the inter-language bots that migrated to Wikidata.

Wikipedia is perhaps one of the best examples of a populous and complex bot ecosystem

but this does not necessarily make it representative. As Table 1 demonstrates, we have investi-

gated a very small region of the botosphere on the Internet. The Wikipedia bot ecosystem is

gated and monitored and this is clearly not the case for systems of malevolent social bots, such

as social bots on Twitter posing as humans to spread political propaganda or influence public

discourse. Unlike the benevolent but conflicting bots of Wikipedia, many malevolent bots are

collaborative, often coordinating their behavior as part of botnets [27]. However, before being

able to study the social interactions of these bots, we first need to learn to identify them [28].

Our analysis shows that a system of simple bots may produce complex dynamics and unin-

tended consequences. In the case of Wikipedia, we see that benevolent bots that are designed

to collaborate may end up in continuous disagreement. This is both inefficient as a waste of

resources, and inefficacious, for it may lead to local impasse. Although such disagreements

represent a small proportion of the bots’ editorial activity, they nevertheless bring attention to

the complexity of designing artificially intelligent agents. Part of the complexity stems from

the common field of interaction—bots on the Internet, and in the world at large, do not act in

isolation, and interaction is inevitable, whether designed for or not. Part of the complexity

stems from the fact that there is a human designer behind every bot, as well as behind the envi-

ronment in which bots operate, and that human artifacts embody human culture. As bots con-

tinue to proliferate and become more sophisticated, social scientist will need to devote more

attention to understanding their culture and social life.

Materials and methods

Data

Wikipedia is an ecosystem of bots. Some of the bots are “editing bots”, that work on the arti-

cles. They undo vandalism, enforce bans, check spelling, create inter-language links, import

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 8 / 13

content automatically, etc. Other bots are non-editing: these bots mine data, identify vandal-

ism, or identify copyright violations.

In addition to bots, there are also certain automated services that editors use to streamline

their work. For example, there are automated tools such Huggle and STiki, which produce a

filtered set of edits to review in a live queue. Using these tools, editors can instantly revert the

edit in question with a single click and advance to the next one. There are also user interface

extensions and in-browser functions such as Twinkle, rollback, and undo, which also allow

editors to revert with a single click. Another automated service that is relatively recent and

much more sophisticated is the Objective Revision Evaluation Service (ORES). It uses

machine-learning techniques to rank edits with the ultimate goal to identify vandals or low-

quality contributions.

Our research focuses on editing bots. Our data contain who reverts whom, when, and in

what article. To obtain this information, we analyzed the Wikipedia XML Dumps (https://

dumps.wikimedia.org/mirrors.html) of 13 different language editions. To detect restored ver-

sions of an article, a hash was calculated for the complete article text following each revision

and the hashes were compared between revisions [23]. The data cover the period from the

beginning of Wikipedia (January 15, 2001) until February 2, 2010 –October 31, 2011, the last

date depending on when the data was collected for the particular language edition. This time

period captures the “first generation” of Wikipedia bots, as in later years, Wikidata took over

some of the tasks previously controlled by Wikipedia. The sample of languages covers a wide

range of Wikipedia editions in terms of size; for example, it includes the four largest editions

by number of edits and number of editors. In terms of cultural diversity, the sample covers a

wide range of geographies.

Wikipedia requires that human editors create separate accounts for bots and that the bot

account names clearly indicate the user is a bot, usually by including the word “bot” (https://

en.wikipedia.org/wiki/Wikipedia:Bot_policy). Hence, to identify the bots, we selected all

account names that contain different spelling variations of the word “bot.” We supplemented

this set with all accounts that have currently active bot status in the Wikipedia database but

that may not fit the above criterion (using https://en.wikipedia.org/wiki/Wikipedia:Bots/Status

as of August 6, 2015). We thus obtained a list of 6,627 suspected bots.

We then used the Wikipedia API to check the “User” page for each suspected bot account.

If the page contained a link to another account, we confirmed that the current account was a

bot and linked it to its owner. For pages that contained zero or more than one links to other

accounts, we manually checked the “User” and “User_talk” pages for the suspected bot account

to see if it is indeed a bot and to identify its owner. The majority of manually checked accounts

were vandals or humans, so we ended up with 1,549 bots, each linked to its human owner.

We additionally labeled human editors as vandals if they had all their edits reverted by oth-

ers. This rule meant that we labeled as vandals also newcomers who became discouraged and

left Wikipedia after all their initial contributions were reverted. Since we are interested in

social interactions emerging from repeated activity, we do not believe that this decision affects

our results.

Using the revert data, we created a directed two-layer multi-edge network, where owner-

ship couples the layer of human editors and the layer of bots [29]. To build the network, we

assumed that a link goes from the editor who restored an earlier version of the article (the

“reverter”) to the editor who made the revision immediately after that version (the “reverted”).

All links were time-stamped. We collapsed multiple bots to a single node if they were owned

by the same human editor; these bots were usually accounts for different generations of the

same bot with the same function. In the network, reverts can be both intra- and inter-layer:

they occur within the human layer, within the bot layer, and in either direction between the

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 9 / 13

human and bot layers. The multi-layer network was pruned by removing self-reverts, as well

as reverts between a bot and its owner.

Interaction trajectories

We model the interaction trajectories in two-dimensional space, where the x-axis measures time

and the y-axis measures the difference between the number of times i has reverted j and the number of times j has reverted i. To construct the trajectories, starting from y0 = 0, yt = yt-1 + 1 if i reverts j at time t and yt = yt-1 − 1 if j reverts i at time t; the labels i and j are assigned so that y >= 0 for the majority of the ij interaction time. We analyze three properties of the trajectories:

• Latency. We define latency as the mean log time in seconds between successive reverts:

μ(log10 Δt).

• Imbalance. We define imbalance as the final proportion of reverts between i and j that were not reciprocated: |ri − rj| / (ri + rj), where ri and rj are the number of times i reverted j and j reverted i, respectively.

• Reciprocity. We define reciprocity as the proportion of observed turning points out of all

possible: (# turning points) / (ri + rj− 1), where ri and rj are the number of times i reverted j and j reverted i, respectively. A turning point occurs when the user who reverts at time t is different from the user who reverts at time t+1.

K-means clustering

To identify the number of clusters k that best represents the data, we apply the elbow and sil- houette methods on trajectories of different minimum length. The rationale behind restricting

the data to long trajectories only is that short trajectories tend to have extreme values on the

three features, thus possibly skewing the results. According to the elbow method, we would

like the smallest k that most significantly reduces the sum of squared errors for the clustering. According to the silhouette method, we would like the k that maximizes the separation dis- tance between clusters and thus gives us the largest silhouette score.

Although the elbow method suggests that four clusters provide the best clustering, the sil-

houette method indicates that the data cannot be clustered well (S8 Fig). We do not necessarily

expect that trajectories cluster naturally; rather, we employ clustering in order to quantify the

differences between the interactions of bots versus humans across languages. We hence ana-

lyze the clustering with k = 4. This clustering also has the advantage of yielding four types of trajectories that intuitively make sense.

Supporting information

S1 Fig. The number of bots, the number of edits by bots, and the proportion of edits done

by bots between 2001 and 2010. Between 2003 and 2008 the number of bots and their activity

have been increasing. This trend, however, appears to have subsided after 2008, suggesting

that the system may have stabilized.

(TIFF)

S2 Fig. For the majority of languages, bots are mainly reverted by other bots, as opposed to

human editors or vandals. English and the Romance languages in our data present excep-

tions, with less than 20% of bot reverts are done by other bots.

(TIFF)

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 10 / 13

S3 Fig. Bot-bot interactions have different characteristic time scale than human-human

interactions. The figures show the distribution of interactions for a particular latency, where

we define latency as the mean log time in seconds between successive reverts. (A) Bot-bot

interactions have a characteristic latency of 1 month, as indicated by the peak in the figure. (B)

Human-human interactions occur with a latency of 2 minutes, 24 hours, or 1 year.

(TIFF)

S4 Fig. Bot-bot interactions are on average more balanced than human-human interac-

tions. We define imbalance as the final proportion of reverts between i and j that were not reciprocated. (A) A significant proportion of bot-bot interactions have low imbalance. (B) The

majority of human-human interactions are perfectly unbalanced.

(TIFF)

S5 Fig. Bots reciprocate much more than humans do also at a smaller timescale. We mea-

sure reciprocity as the proportion of observed turning points out of all possible. (A) A signifi-

cant proportion of bot-bot interactions have intermediate or high values of reciprocity. (B)

The majority of human-human interactions are not reciprocated.

(TIFF)

S6 Fig. Four types of interaction trajectories suggested by the k-means analysis. The left

panels show a sample of the trajectories, including bot-bot and human-human interactions

and trajectories from all languages. The right panels show the distribution of latency, imbal-

ance, and reciprocity for each type of trajectory. The three properties measure the average

steepness, the y-value of the last point, and the jaggedness of the trajectory, respectively. (A) Fast unbalanced trajectories have low reciprocity and latency and high imbalance. (B) Some-

what balanced trajectories have intermediate imbalance and reciprocity. (C) Slow unbalanced

trajectories have low reciprocity and high latency and imbalance. (D) Well balanced trajecto-

ries have low imbalance and high reciprocity.

(TIFF)

S7 Fig. The number of articles with a certain number of bot-bot and human-human

reverts. (A) Few articles include more than 10 bot-bot reverts. The most contested articles

tend to be about foreign countries and personalities. Further, the same articles also re-

appear in different languages. (B) There are many articles that are highly contested by

humans. The most contested articles tend to concern local personalities and entities. It is

rare that a highly contested article in one language will be also highly contested in another

language.

(TIFF)

S8 Fig. Performance of the k-means clustering algorithm for different number of clusters

and for sub-samples with different minimum length of trajectories. (A) The elbow method

requires the smallest k that most significantly reduces the sum of squared errors for the cluster- ing. Here, the method suggests that four clusters give the best clustering of the data. (B) The sil-

houette method requires the k that maximizes the separation distance between clusters, i.e. the largest silhouette score. Here, the method suggests that the clustering performs worse as the

number of clusters increases.

(TIFF)

S1 Table. Descriptive statistics for the bot-bot layer and the human-human layer in the

multi-layer networks of reverts. Bots revert each other to a great extent. They also reciprocate

each other’s reverts to a considerable extent. Their interactions are not as clustered as for

human editors. Still, both for bots and humans, more senior editors tend to revert less senior

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 11 / 13

editors, as measured by node assortativity by number of edits completed.

(PDF)

Acknowledgments

The authors thank Wikimedia Deutchland e.V. and Wikimedia Foundation for the live access

to the Wikipedia data via Toolserver. The data reported in the paper are available at 10.6084/

m9.figshare.4597918.

Author Contributions

Conceptualization: MT LF TY.

Data curation: MT RG TY.

Formal analysis: MT.

Funding acquisition: TY.

Investigation: MT.

Methodology: MT TY.

Project administration: TY.

Resources: TY.

Supervision: TY.

Visualization: MT.

Writing – original draft: MT.

Writing – review & editing: MT RG LF TY.

References 1. Cornell Creative Machines Lab. AI vs. AI. Two chatbots talking to each other [Internet]. YouTube. 2011

[cited 2016 Aug 4]. https://www.youtube.com/watch?v=WnzlbyTZsQY

2. Franklin S, Graesser A. Is It an agent, or just a program?: A taxonomy for autonomous agents. In: Müller

JP, Wooldridge MJ, Jennings NR, editors. Intelligent Agents III: Agent Theories, Architectures, and Lan-

guages [Internet]. Springer Berlin Heidelberg; 1997 [cited 2016 Aug 3]. p. 21–35. http://link.springer.

com/10.1007/BFb0013570

3. Floridi L, Sanders JW. On the morality of artificial agents. Minds Mach [Internet]. Kluwer Academic Pub-

lishers; 2004 Aug [cited 2016 Aug 12]; 14(3):349–79. Available from: http://link.springer.com/10.1023/

B:MIND.0000035461.63578.9d

4. Berners-Lee T. Short summary of the World Wide Web project [Internet]. Google. 1991 [cited 2016 Aug

4]. https://groups.google.com/forum/#!msg/alt.hypertext/eCTkkOoWTAY/bJGhZyooXzkJ

5. Koster M. Guidelines for robot writers [Internet]. 1993 [cited 2016 Aug 2]. http://www.robotstxt.org/

guidelines.html

6. Abokhodair N, Yoo D, McDonald DW. Dissecting a social botnet: Growth, content and influence in Twit-

ter. Proceedings of the 18th ACM Conference on Computer Supported Cooperative Work & Social

Computing—CSCW ‘15 [Internet]. New York, New York, USA: ACM Press; 2015 [cited 2016 Jul 27].

p. 839–51. http://dl.acm.org/citation.cfm?doid=2675133.2675208

7. Leonard A. Bots: The Origin of New Species. Hardwired; 1997. 218 p.

8. Brown J, Duguid P. The Social Life of Information. Harvard Business Press; 2000.

9. Gianvecchio S, Xie M, Wu Z, Wang H. Measurement and classification of humans and bots in Internet

chat. Proceedings of the 17th Conference on Security Symposium [Internet]. USENIX Association;

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 12 / 13

2008. p. 155–69. https://www.usenix.org/legacy/event/sec08/tech/full_papers/gianvecchio/

gianvecchio_html/

10. Sysomos. An in-depth look at the most active Twitter user data [Internet]. 2009 [cited 2016 Aug 2].

https://sysomos.com/inside-twitter/most-active-twitter-user-data

11. Cashmore P. Twitter zombies: 24% of tweets created by bots [Internet]. Mashable. 2009 [cited 2016

Aug 2]. http://mashable.com/2009/08/06/twitter-bots/#JqTVM0vEgqqA

12. Varvello M, Voelker GM. Second Life: A social network of humans and bots. Proceedings of the 20th

international workshop on Network and operating systems support for digital audio and video—NOSS-

DAV ‘10 [Internet]. New York, New York, USA: ACM Press; 2010 [cited 2016 Jun 17]. p. 9. http://portal.

acm.org/citation.cfm?doid=1806565.1806570

13. Holiday R. Fake traffic means real paydays [Internet]. Observer. 2014 [cited 2016 Aug 3]. http://

observer.com/2014/01/fake-traffic-means-real-paydays/

14. Zeifman I. 2015 bot traffic report: Humans take back the Web, bad bots not giving any ground [Internet].

Imperva Incapsula. 2015 [cited 2016 Aug 1]. https://www.incapsula.com/blog/bot-traffic-report-2015.

html

15. Murphy D. Blizzard cracks down on World of Warcraft cheaters [Internet]. PC Magazine. 2015 [cited

2016 Aug 2]. http://uk.pcmag.com/console-games-products/41990/news/blizzard-cracks-down-on-

world-of-warcraft-cheaters

16. Newitz A. Ashley Madison code shows more women, and more bots [Internet]. Gizmodo. 2015 [cited

2016 Aug 3]. http://gizmodo.com/ashley-madison-code-shows-more-women-and-more-bots-

1727613924

17. Russell S, Norvig P. Artificial Intelligence: A Modern Approach. Third Edit. Harlow, UK: Pearson;

2009.

18. Steiner T. Bots vs. wikipedians, anons vs. logged-ins (redux). Proceedings of The International Sympo-

sium on Open Collaboration—OpenSym ‘14 [Internet]. New York, New York, USA: ACM Press; 2014

[cited 2016 Jun 17]. p. 1–7. http://dl.acm.org/citation.cfm?doid=2641580.2641613

19. Niederer S, van Dijck J. Wisdom of the crowd or technicity of content? Wikipedia as a sociotechnical

system. New Media Soc [Internet]. 2010 Dec 1 [cited 2016 Jun 17]; 12(8):1368–87. Available from:

http://nms.sagepub.com/cgi/doi/10.1177/1461444810365297

20. Kittur A, Suh B, Pendleton BA, Chi EH. He says, she says: Conflict and coordination in Wikipedia. Pro-

ceedings of the SIGCHI conference on Human factors in computing systems—CHI ‘07. New York, New

York, USA: ACM Press; 2007. p. 453–462.

21. Brandes U, Kenis P, Lerner J, van Raaij D. Network analysis of collaboration structure in Wikipedia. Pro-

ceedings of the 18th international conference on World wide web—WWW ‘09. New York, New York,

USA: ACM Press; 2009. p. 731–740.

22. Yasseri T, Sumi R, Rung A, Kornai A, Kertész J. Dynamics of conflicts in Wikipedia. PLoS One. Public

Library of Science; 2012 Jan; 7(6):e38869.

23. Sumi R, Yasseri T, Rung A, Kornai A, Kertész J. Edit wars in Wikipedia. Privacy, Security, Risk and

Trust (PASSAT), 2011 IEEE Third International Conference on Social Computing (SocialCom) [Inter-

net]. 2011 [cited 2016 Aug 30]. p. 724–7. http://arxiv.org/abs/1107.3689

24. Halfaker A, Riedl J. Bots and cyborgs: Wikipedia’s immune system. Computer (Long Beach Calif) [Inter-

net]. IEEE Computer Society; 2012 Mar [cited 2016 Aug 1]; 45(3):79–82. Available from: http://

ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6163451

25. Geiger RS, Halfaker A. When the levee breaks: Without bots, what happens to Wikipedia’s quality con-

trol processes? Proceedings of the 9th International Symposium on Open Collaboration—WikiSym ‘13

[Internet]. New York, New York, USA: ACM Press; 2013 [cited 2016 Jun 17]. p. 1–6. http://dl.acm.org/

citation.cfm?doid=2491055.2491061

26. Yasseri T, Spoerri A, Graham M, Kertész J. The most controversial topics in Wikipedia: A multilingual

and geographical analysis. In: Fichman P, Hara N, editors. Global Wikipedia: International and Cross-

Cultural Issues in Online Collaboration. Scarecrow Press; 2014.

27. Marciel M, Cuevas R, Banchs A, González R, Traverso S, Ahmed M, et al. Understanding the detection

of view fraud in video content portals. Proc 25th Int Conf World Wide Web. International World Wide

Web Conferences Steering Committee; 2016;357–68.

28. Ferrara E, Varol O, Davis C, Menczer F, Flammini A. The rise of social bots. Commun ACM [Internet].

2016 Jul 19 [cited 2016 Jun 20]; 59(7). Available from: http://arxiv.org/abs/1407.5225

29. Kivela M, Arenas A, Barthelemy M, Gleeson JP, Moreno Y, Porter MA. Multilayer networks. J Complex

Networks. Oxford University Press; 2014 Jul; 2(3):203–71.

Even good bots fight

PLOS ONE | DOI:10.1371/journal.pone.0171774 February 23, 2017 13 / 13

Copyright of PLoS ONE is the property of Public Library of Science and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.