Having recently realised how much disagreement there can be in the social and behavioural sciences re. the concepts of internet echo chambers and filter bubbles, I got thinking about what a good conceptualisation should look like. I think it is fair to say that the terms are not usually used in a purely descriptive way, and partly because of this there is room for reasonable confusion or conflict over their appropriate extension. Of course the terms have an extension fixed by their inventors and users – but they will need some usefulness beyond this literature if this literature is to be defended from critics, and if the terms are to be of enduring value. I think that a good conceptualisation is do-able, but it might not please everyone. I’ll try to explain as I go:
The Metaphors
First some quick definitions: In both terms, ‘chamber’ and ‘bubble’ are doing similar work, suggesting that a person is isolated or separated from some kind of an ‘outside’ world. The ‘filter’ in filter bubble refers to an information filter (or something like information – maybe suggestions about values, political signals, etc) – only some kinds of information are able to get into the bubble, or at least, some sorts of information have trouble getting in. (Eli Pariser coined the term to refer to a specifically online phenomenon, but let’s put aside that restriction for the moment.) The ‘echo’ in echo chamber also refers to information (or something like it), but in this case we do not just passively receive the information which filters in, but we have a more constructive role in influencing what we receive, as the ideas we project into the chamber are ‘echoed’ back to us.
Every person might have their own filter bubble(s), but the paradigmatic echo chamber (the sort that concerned Marc Sageman and Cass Sunstein, who popularised this usage of the metaphor) is a group of people who are amplifying or exaggerating a certain message to each other.
To use the creators’ own examples: The classic filter bubble is a set of algorithms on internet search and social media sites which lead the user to whatever sorts of stimulus they most want to see (or that someone else wants them to see), giving them a skewed worldview in the process. The classic echo chamber is a group of extremists who believe that their views are more normal or rational or reasonable than these views really are, because their main epistemic resource is each other’s testimony. (To maintain an echo chamber here, it helps to install a filter bubble of some sort around the group.)
A Problem
That takes us about as far as we can get by picking through the metaphors, but we are left with a far broader account of the terms than their creators intended. With a little reflection we will see that ‘filter bubbles’ are ubiquitous, and with a little more reflection we will see that ‘echo chambers’ are too.
The first is the familiar problem that we all necessarily see the world from some perspective. We each live in a certain time and place, we are peculiarly social and visual creatures, various cultural and linguistic cues will direct our attention in some ways rather than others. In some sense the only escape from a filter bubble is the ‘view from nowhere’ (the ‘God’s eye view’), and that is a notoriously difficult perspective to take up. This may be an interesting issue, but it is certainly not the one that the term was designed to highlight.
The ubiquity of echo chambers may be a little harder to see at first, but it is certainly worth stopping to appreciate it. As Wittgenstein pointed out, all language use is a fundamentally social activity, and one of the reasons for this is that language is rule-based, and so needs to be testable for rule violations: “it is not possible to obey a rule “privately”: otherwise thinking one was obeying a rule would be the same thing as obeying it”. (Philosophical Investigations §202) In contrast, if we wanted to maintain a language without the use of an ‘echo chamber’:
“(Maintaining consistent usage) is done precisely by the concentrating of my attention; for in this way I impress upon myself the connection between sign and sensation. But ‘I impress it upon myself’ can only mean: this process brings it about that I can remember the connection right in the future. But in the present case I have no criterion of correctness. One would like to say: whatever is going to seem to me right is right. And that only means that here we can’t talk about right.” (PI §258)
Much the same thing can be said of our epistemology. We may be able to get quite a way reasoning and observing privately, but the benefits of regularly testing what we think by communicating with others are so great that it is pretty much unthinkable to avoid this forever. So, generally we learn and reason through communication, and if this is real communication then our thoughts when beginning a conversation will be ‘echoed’ back to us over the course of it, to some extent.
What Else is Needed?
What is missing so far from my definition is something to separate the troubling echo chambers and filter bubbles from the rest. And this probably means building values in right from the start. I will examine some options for what could be added to these definitions now.
Wrong conclusions: Worth a mention is the idea that a chamber/bubble is problematic (and so deserves our attention) because it reinforces a mistaken view. I don’t think this is particularly promising on its own, but doubtless this idea complements the others. It is unusual to talk about the need for an echo chamber to preserve good beliefs, and such a claim should probably raise eyebrows.
Cultural best-practice epistemology: In some sense almost any contemporary internet-nazi is far less contained within a filter bubble or echo chamber than a villager in mediaeval Europe. The nazi could hardly be as constrained as the villager if they tried. Yet it is usual to use the terms for the nazi’s situation, not the villager’s. We might explain this by saying that the nazi has far more opportunity to acquire novel information and engage with a variety of different perspectives than the villager, and it is the failure to take these opportunities which makes their chamber/bubble problematically insular (and so, which makes the label deserved).
Active v. Passive: Related to the above, we might say that someone is in a chamber/bubble if they are there willingly, rather than unknowingly or unavoidably. Focusing on the agent in the chamber/bubble, we will be interested in how ‘visible’ the chamber/bubble is, and how they ended up in it. We might also be particularly interested (as Eli Pariser encourages us to be) in the responsibilities of the creators of the chamber/bubble, if there is one independent of those inside it. This sort of definition focuses on culpability or motive, not as much on how restricted or perverted the person’s epistemology actually is.
Encounter: Perhaps a villager in mediaeval France is not in a problematic chamber/bubble because they will not have to spend much time interacting with people in different bubbles. They were doubtless aware that foreigners might believe different things, but this has less practical import – they will never have to rely on sharing a worldview with these people. The problem today is that we may easily end up in a chamber/bubble, but then have to share a society with others whose worldview we cannot understand.
Role-based or expectation-based: There are usages of the term ‘echo chamber’ prior to Sageman’s to refer to poor practice by mainstream journalists. In these cases the public expects novel reportage and commentary on an event, but they are actually given a rehash of what is being circulated by other journalists. So the benefits of diversity in the informational environment have been lost, and weak claims may appear far more credible than they actually are (because far fewer journalists are actually working ‘at the coalface’ than the public expects). The journalist has a professional role and a professional ethic, but we can also talk of ‘roles’ such as citizen, community member, student, or creator. I think that this is a promising addition to the simple and overly-general definition, which intersects nicely with some of the other options I have raised here.
Definitions?
My attempt to reign in the extensions of the terms has not resulted in two nice neat definitions which can tell us definitively who is and is not in a chamber or bubble. This may appear troubling to some. It may now seem that those using the terms are not being very ‘objective’. If this is motivated by a concern that social science is not real science unless it deals in some sort of Platonic absolutes, then it is severely misguided. But there is legitimate concern that too-vague definitions will lead to science which is not replicable, falsifiable, or predictive.
The solution to this is not to impose a rigid definition across all investigations of the concepts, but rather for those working with them to be clear about their interests and concerns. I think there is no doubt that writers in this literature let their own values creep into their work. But this is only problematic insofar as the values are kept implicit rather than being set out for critical appraisal. Those looking for single right definitions might begin in the vicinity of my ‘best-practice epistemology’, ‘encounter’, and ‘expectation-based’ considerations, but I highly doubt that the ‘objectivity’ benefits of such definitions would outweigh the loss of all of the other interesting lines of enquiry that would be precluded.
The ArachneanMetaxion Point 