Beware of Personalised Search! Information Dystopia and Online ‘Filter Bubbles’

Last year Eli Pariser spoke at TED, covering a topic that we’re all becoming increasingly affected by while navigating the web. Eli introduced his theory on the online ‘filter bubble’, explaining that through the continuous quest to ‘personalise’ search results, our online environments are becoming less and less diverse.

Hi everyone, I am Ned and this is my first (proper) post on State of Search. I’ve become fairly annoyed lately with many of the websites I visit regularly and spend a lot of time on, like: Amazon, Facebook and Google (after all, I am an SEO), that are becoming incredibly ‘samey’ places to hang out and discover information online. So much of the content that I see every day is either something I’ve already seen or, worse, nothing to do with what I was looking for in the first place! This is one of the effects that Eli refers to when talking about online ‘filter bubbles’.

Filter bubble [Definition] – “The unique universe of information that you live in online”

The filter bubble is an invisible force and one that I feel, it is also a counter-intuitive force that goes against the very powerful nature of the Internet that has, up until now, enabled the borderless and free ability to create and share vast amounts of information. Instead, this invisible force filters the information at your disposal, ultimately limiting not what exists on the Internet as a whole, but what you have the ability to find on ‘your Internet’. Such that it creates a form of information dystopia.


Information Filtering in practice


One area where you may be most familiar with this happening is Facebook. Facebook’s ‘EdgeRank’ is an algorithm that determines what people see when they log in to Facebook. In layman’s terms, this algorithm determines what content to show you based on:

  • Who the individual is that shared the content;
  • How you are connected to that individual;
  • and the certain types of content that you interact with most (liking, sharing etc).
  • Essentially it assigns values based on 3 variables: Affinity, weight, and time decay.

Facebook Edgerank Algorithm

I won’t dwell on this too much in this post but Kelvin Newman covered the topic very thoroughly in his post on algorithm marketing.

Mark Zuckerberg Quote on Relevancy

The problem here is that Facebook is removing results from you, without your consent.


Amazon were also one of the first adopters to personalising results based on users profiles, dedicating a large part of their homepage to product based on:

  • Recently viewed products
  • Recent purchases
  • Other similar users buying/shopping habits
Amazon Personalised Recommendations
My personalised homepage on

This is an intelligent approach to personalisation and many believe (myself included) that this is one of the reasons for Amazon’s phenomenal success. However, it is not without it’s flaws, the amount of information I am potentially missing out on is vast. We must challenge this personalised approach and question whether or not it is really providing to best experience for the user.

Learn a little more about how Amazon’s recommendation system works by reading the top answer here on Quora or looking at the top answer over on StackOverflow.


A large amount of Google’s recent algorithmic changes have been focused on creating far more personalised results for the user; this trend only looks to continue. As SEOs and online marketers, we must understand how Google will look to do this. In the talk, Eli says that a Google engineer confirmed there are 57 signals that Google considers to help serve you results specific to your query, (even when you’re logged out!). Anything from:

  • The browser that is being used
  • Where you are accessing the information
  • What computer/device is being used

Google Signals for Personalised Results

This means that the results shown for any two people searching individually for the same keyword/keyword term can be very different. Even if these people are searching for the exact same term, at the exact same time, in exactly the same city, their results will be influenced by a large variety of factors and information that Google know’s about you. This is a trend that’s been happening for a while and we’re going to see considerably more of in the future; there is no ‘standard’ Google anymore, search has become personalised.

Why we should all beware of the ‘filter bubble’

The likes of Facebook, Amazon, Google and Netflix aren’t the only sites that are doing this, in fact a large amount of sites have began personalising their results. However, despite their brilliance, technologies such as Amazon and Netflix’s vastly intelligent recommendation systems need to give us (as the recipient for this information) some control, we need to be able decide what is right and wrong for us. We don’t want to live in a ‘web of one’, but rather a ‘web of many’.

“The Internet is showing us what it thinks we want to see but not necessarily what we need to see”

The problem with Internet sites tailoring their results to our personal likes and dislikes based on algorithms, is there is a dangerous unintended consequence of us all getting trapped inside our own ‘filter bubbles’.

The Filter Bubble

The problem about this personalisation of search, as Eli points out is that “You don’t decide what gets in [to your filter bubble], and more importantly you also don’t actually see what gets edited out.”

The Effect of the Filter Bubble

Over-Personalisation: The Robots Need a Human Touch

Roger Mozbot

Recommendation engines are a staple for the Internet. Using algorithms, websites can accumulate vast amounts of user-specific data and tailor their offerings to your likes, what you’ve spent time looking at, what your friends like etc. But this can be a flawed. Eli discusses the shift from the classical editor type figure of the print revolution; these were largely removed when the Internet developed into mainstream use, allowing a raft of bloggers and amateurs being able to publish their content without the need of editorial approval. This process has caused a trend to replace these humans with robots. The content we are served when logging onto some sites like: Mashable, the Huffington Post, Yahoo News and Facebook has been selected by algorithms. But algorithms should not be curators of the almost infinite amount of information on the Internet. Robots don’t do ethics. They need a human touch.

Humans need to consume information that we don’t agree with to understand different points of view; we need to be challenged; and we need to be made uncomfortable. If we don’t have transparency or control over the filters that are restricting us from seeing this information, we remain trapped inside our ‘filter bubbles’, blind to the kaleidoscope of people; different ideas; new ideals and perspectives on the Internet, in a ‘web of many’. The increased adoption of filter bubbles means that, instead of being social and integrated we’re actually heading backwards, toward a closed off, limited point of view. We must watch this change carefully, to quote Darwin:

“It is not the strongest or the most intelligent who will survive but those who can best manage change.” [Charles Darwin]

Eli has written a book on The Filter Bubble if you’re keen on reading more, but I’d love to hear your thoughts on personalisation of search results?

  • Do you think personalised results are a good or bad thing?
  • What sites have you noticed personalise results particularly well? Which do you think don’t personalise results well?
  • Do you think personalised results have started to limit the information you’re seeing on the web?

About Ned Poulter

Ned Poulter is CEO & Founder of Pole Star Digital, a Digital Marketing Consultancy he runs alongside a number of entrepreneurial ventures. He spends his time developing and overseeing digital strategy for a number of well reputed international clients, developing creative approaches to achieving organic search results, increasing inbound traffic, establishing and managing paid search campaigns and ultimately driving revenue.

10 thoughts on “Beware of Personalised Search! Information Dystopia and Online ‘Filter Bubbles’

  1. Ironically Google’s implementation of personalised search is one of the ‘least bad’, because it allows you to opt-out with a single click and thus see the ‘regular’ SERPs. Having said that, those SERPs still have some measure of personalisation, most commonly geographic personalisation…

    But yes, it’s a serious potential problem and a good reason for everyone to regularly step outside of their information consumption habits and venture towards websites and platforms you’d not normally visit.

    A monthly catch-up on Al Jazeera is not a bad idea…

    1. I’d have to agree Barry, Google is definitely one of the ‘least bad’ examples out there, I think to Eli’s point though when you start considering that you ‘don’t actually see what is getting edited out’ it’s when it begins to get Orwellian and downright scary.

      Also, while it’s not something I picked up on in this post but I’ve found that geographic personalisation of results can also be an unnecessary nuisance. I was also keen to use ‘SEO’s on Google+’ as a great case study for this. I think collectively, those working in search and related disciplines are some of the most vulnerable to this potential problem, the trends within our information consumption habits are likely to draw up a very polarised example of the problems with this in the wild…

  2. I’m on-board with personalization. I know it’s scary to think about our privacy at risk and that we might be missing out on information that we are seeking, however, I’d much rather see ads/articles about motorcycles and SEO than tampons and shoes.

    I’m looking forward to more insight at the Hangout tomorrow.

    1. Hi Matt, fair point – I do think that even though there are many scaremongers regarding data privacy (part of a whole different argument) I’m also fairly morally torn when it comes to shopping and remarketing, in short I’m a fan.

      But when it comes to threatening the ability to consume information from sources that may fall outside your own ‘filter bubble’, likely defined algorithmically and without us knowing what is being removed, I have major issues.

      Look forward to the Hangout to hear yours and others opinions too.

  3. Filter bubbles genuinely scare me and excite me simultaneously.

    If you remember back a few years, we all spoke about our “internet
    neighbourhood” when we used to “surf” the web for hours just exploring.
    The theory, which was personally true was that I, and other visited very
    few websites in a normal surf. And only added more into that
    neighbourhood by recommendation or services like stumbleupon.

    Personalisation can be amazing, clever and if they get it right our
    satisfaction will actually increases and they will be unsung heroes. The
    risk is that we see less, interact with the same few people, their
    networks and it becomes self fulfilling.

    Then to break out, we either get recommendations from our active and
    engaging networks OR business pay to be promoted. See what FB is doing
    with their recent edgerank changes to see that one coming over the hill

    My fear is that by being helpful and curating the flow of ‘stuff’, that
    actually we get a more safe, but boring web experiences that limits new
    people, brands and surprises coming on in.

    Small does feel safe for most people, but like many online decisions taken for us, we sleepwalk right into it.

    Just my 2 cents of ramblings.

    1. Hi Adrian, funnily enough I share the same opinion, I’m both scared and excited by the notion of filter bubbles. Personally, while I’m fascinated by the potential of algorithmically arranged content based on individual user profiles, the scales are tipped toward being scared because of an indescribable level of FOMO (fear of missing out) not just on social interactions with certain individuals I may not communicate with regularly, but also on information from sources I haven’t visited.

      I remember the conversation about the Internet neighbourhood and ‘good old surfing’, but the Internet has changed since these days, it’s far more social and there’s simply far more content. The problem with this is, by presuming we like things purely by a fairly unintelligent metric like ‘how many times you viewed this website’ it closes the shutters on our very ability to access information from wider, more diverse sources. I agree, as you rightly put it, “it feels more safe… but limits the surprises coming in”, I for one like an element of risk and the occasional surprise too. Most importantly I want the ability to decide when I can have these, not have them blindly filtered out by machines. To paraphrase Parker from Alien “It’s a robot…. A god damn robot!”

  4. Imagine the only online interaction, source of news, commerce and statistical information is limited to those with whom you currently enjoy on twitter or, facebook. That is Google and, she’s becoming more adept each day.

Comments are closed.