Social Bot Design Strategies.

At the time of writing social bots have still been barley examined. Only few academic research can be found on this topic and it is only possible to identify few design approaches of sophisticated social media bots. The few documented approaches have been published by researchers. It is to assume that there are other social bots that have not been identified yet.

We can identify social bots of three types:

1. Automated Social Engeneering bots, or ASE bots, have been used by Huber et al. (2009) to operate on the social network service Facebook, but can also be adopted for other SNSs. The main concept of these bots is the automation of random tasks such as posting direct or public messages and developing their own social network.
These bots can be compared to the chatter bots from the web 1.0, with the difference that they are adapted to a social network environment and therefore utilized its features and functions. When involved in a sophisticated conversation these bots are easy to identify since they are not very believable at imitating human conversation.
Similar bots where used in an experiment about the effects twitter bots can have on human to human conversation (Nanis et al., 2011). The social bots randomly used the given functionalities of a Social Network Service, Twitter in this case, to study the effect on their network.

A screenshot of the website of realboy tritter bot.

A screen shot of the Realboy developers web site.

2. Realboy is a twitter bot created by Zack Coburn and Greg Marra (Marra, Coburn, 2008) that is trying to imitate real human behavior, through means of copying. Its goal was to identify social graph clusters, infiltrate them, and achieve a 25% follow-back rate (Marra, 2011). This bot automatically followed five new users a day, refollowed everybody that followed him, and sent one tweet per day.
For the preparation of this bot a spider program was used to identify a great amount of twitter accounts and its connections. Within this network similiar algorithms as used by Amazon to suggest products based on what other people buy have been applied to identify various sub networks of people more densely connected to each other. From one of these sub networks the bot would then draw three generic key words such as “weather”, “running”, and “shoes” and do a normal twitter search on those three keywords. From the results of this search the bot would draw a tweet once a day and copy it into its own twitterfeed, without including hashtags or @replies.
By recycling the public (human) conversation available on twitter based on the key words from its own social network, this bot was able to talk about the topics most of his followers were interested in and develope a more consistent conversation over time. If the bots social networks interest shifted from running to food topics than this bot would also talk about food. In the example given by Marra (2011), he presents a bot that within six months shifted from being an expert on outdoor sports to diabetes topics and then to Affiliate Marketing.

3. Honeybot is a bot created by Lauringer et al. (2010) which is based on the concept “Man in the middle”, where a direct connection between two entities is secretly intruded by the attacker who is then able to eavesdrop on the victims and even intercept or introduce messages and links. “The general attack principle works with any chat system that allows the exchange of private messages.” (Lauringer et al., 2010: 2)
The honeybot was able to automatically bootstrap a conversation between, influence the topic of the ongoing conversation, make the participants click on links that were inserted into the conversation, and apply techniques to make conversations last longer.
In the study the bot was applied to two different environments, a classic Internet Relay Chat (IRC) environment and a social network service environment (Facebook). By design IRC does not support social networking such as Facebook. Still it a great amount of social conversation is transmitted via IRC.
This bot is propably the most intrusive concept as it directly interferes with an existing conversation, whereas the other two bot types described above rather created new conversation. As a result of the honeybot on Facebook four out of ten people clicked on the link that was inserted by the attacker.

HUBER, M. & KOWALSKI, S. & NOHLBERG, M. & TJOA, S. (2009). “Towards automating social engineering using social networking sites”. In CSE (3), p. 117–124. IEEE Comp. Soc., 2009.

LAURINGER, Thomas & PANKAKOSKI, Veikko & BALZAROTTI, Davide & KIRDA, Engin (2010). “Honeybot, Your Man in the Middle for Automated Social Engineering”, 3rd USENIX Workshop on Large-Scale Exploits and Emergent Threats (LEET), San Jose, April 2010

MARRA, Greg & COBURN, Zack (2008). “Realboy – Believable twitter bots”, from

MARRA, Greg (2011). “Gepetto’s Army: Creating International Incidents with Twitter Bots”, from, last modified March 15, 2011.

NANIS, Max & PEARCE, Ian & HWANG, Tim (2011). “PacSocial: Field Test Report “, from, last modified January 2012.

One thought on “Social Bot Design Strategies.

  1. Pingback: Effectiveness of Social Media Bots. | Social Media and Society

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>