Meme to Web is an agent based model I developed along with Mario Paolucci at LABSS to explore the effects of internet filtering technologies on information diversity.
The idea of building this model came to us at a time when Web2.0 applications triggered a second wave of optimism on the democratizing virtues of internet technologies. The low or non existent barriers to entrance posed by the “read-write-web” meant that everybody would be finally able to express their thought and potentially circulate it among millions at no cost.
However, at a closer look, it was clear to us that a disproportionate role in gatekeeping and selecting which information was being actually passed on, was played by a plethora of intermediate technologies: search engines, collaborative content filters, social networking applications, even the network structure itself – as the work of Barabasi on the nature of scale-free networks made clear – has a fundamental role in selecting information. All these applications act as collective filters on knowledge, where the community of users decides in a participatory, implicit and emergent fashion on the quality and relevance of every single piece of content introduced in the system.
We were interested in the effect that such cultural circulation paradigm has on the diversity of information available to mankind through the Internet.
The debate on the influence of intermediate technologies on cultural diversity has been reignited recently with the filter bubble argument raised by Eli Pariser in his popular book and, among the others, by Evgeny Morozov with an insightful article on the death of the cyber flaneur
The power of seemingly transparent, neutral, technologies to actually determine the nature of the cultural artifacts one has access to has been finally acknowledged. With this work we wished to explore the intrinsic tendency of each of the main information filtering platforms towards cultural pluralism and draw a history of the internet in terms of the easiness to access novel, diverse content that it has granted since its inception.
We modelled individual culture employing the notion of meme, in a weaker version of the original formulation of Richard Dawkins. The model implements a society of agents, each endowed with a particular meme pool, engaged in the production and consumption of cultural artifacts. The cultural artifacts are no more than aggregations of memes – an abstraction of artwork, writing, poetry, music… – stemming from the memetic configuration of the creator.
At each step:
- a fraction of the population (the “publishers”) has the chance of publishing an artifact and linking it with other artifacts
- all the agents “consume” some of the ‘most relevant’ artifacts as filtered by the content filtering mechanisms implemented [None; PageRank; Reddit-like; Pure popularity; and a mix of the above.] and retain a subset of the memes endowed in the artifacts.
The retaining of the memes happens on a soft cognitive fashion: some memes are memorized, the most reiterated memes of them are then retained in a belief base containing ‘accepted’ memes. Both the memorization and the acceptance take place under a homophily constraint: the more memes the agent’s mind and the artifact being read have in common, the more likely the agent will retain other memes in the artefact.
There are two versions of this model. One coded in NetLogo implements three conditions:
- flaneur condition (bare network)
- PageRank condition – Only Google acting as filter on content
- reddit condition – Google and a vote-based collaborative content filter
Another version of the model was coded in Java/Repast and, to the former, adds two more conditions:
- twitter – following/follower