Avoiding the Bubble – 10 ways to broaden your data visualisation horizons on social media


We know all too well that without proper care online communities can easily become bubbles, effectively becoming echo-chambers of opinion that, unchecked, can leave unwary users with a very distorted view of how real world opinion differs from that in their online community.

We seek out social contact within a relatively narrow set of views and ideologies; we are naturally attracted to people who share our views and actively shun those who don’t. This has, and is, playing out in the political world at the moment with many “remain” voter left reeling after the UK voted for Brexit. For me personally this meant I could debate and engage with only one Brexit voter in my network – did this really help me shape my opinion and attitudes? Did I affect anyone elses as a result of my discussions on the subject, or did we simply reinforce our own beliefs? The blunt truth is that I comprehensively failed to either appreciate or engage with any other viewpoint apart from those almost identical to my own. On the political spectrum the two camps were ideologically so far apart that this polarisation of views was reinforced on both sides and arguments that seemed obvious to one side failed to land on the other. This left the Remain vote in disarray as they failed to appreciate their own failings. In the US we are also seeing this play out with the lead of 60 / 70 points, predicted by many for Clinton, failing to materialise as Trump supporters continue to be disengaged by any alternative despite the Republican’s many “gaffes”.

I won’t dwell on it here as this echo-chamber effect has been been discussed by many, a particularly good article by David Byrne is well worth reading.

echo-chamber-7

Echo Chamber by Christophe Vorlet, 2016

 

Data Visualisation

Within data visualisation, my field of interest, it is easy to see the same issues play out. In the data viz world online communities have typically been built around software / solutions;  Tableau, Qlik, PowerBI, D3, R to name a few; as well as having a more general solution agnostic communities typically flourishing around experts / researchers or special interest such as sports data . Visualisations that might not get a second glance in one community can be lauded as the best thing since sliced bread in others – often praise revolving around the technical difficulties of producing the visualisation as opposed to their validity as a useful / interesting visualisation or analysis. The echo of what is “good” / “bad” can vary wildly between solutions and communities (though typically a hatred of a pie chart unites communities in a common rallying cry).

For new members of the data visualisation community it can be very easy to become distracted by these echoes and feel that certain techniques or visualisations offer more value (based on feedback from the community) than others. Of more concern is that without checks and balances communities can easily alienate those who don’t share similar opinions to those in the “bubble” leading to an increasingly narrow set of viewpoints, all reinforcing each other.

city-people-bubble-soap-large

How to avoid the Bubble

With this in mind I wanted to offer some tips to the discerning social media user in the Data Visualisation world, new or old, on how to avoid the bubble effect and ensure your timeline remains diverse.

  1. Remember you are in a bubble

Simply being aware of the fact that our online communities don’t reflect the real world is a start. Remember it. Try and actively switch your viewpoint to that of an outsider at regular intervals in order to try and see your community through a different lens.

  1. Be yourself

Online communities are seen y some as a means to end career and learning-wise but that doesn’t stop you developing an online personality and diversifying your posts. Showing people who you are outside the community will help people relate in a different way to your online self and give them confidence to challenge your views if they want to

  1. Diversify who you follow

Okay so this one is fairly obvious but it needs to be said: don’t just follow people who are likely to agree with you. Go out of your way to look for communities in other areas away from your chosen data visualisation solution – use Twitter lists if you wish to ensure your timeline doesn’t become cluttered.

Follow

Twitter recommendations serve to narrow, not broaden, your network.

Follow a wide range of genders and ages, go outside your normal circles, a diverse network of followers will server to provide a range of views to counterbalance yours.

  1. Diversify your followers

So this is harder, but you really need to make sure you have a wide range of different viewpoints in your follower list. That way your posts are more likely to be debated as opposed to be accepted at face value.  How do you do this? Post on different subjects away from your core solution, e.g. if you primarily post about Tableau then try to keep your posts generic, or try building visualisations in a range of different software to ensure you attract followers from different software vendors / solutions. Build a broad base of content but remain focused to ensure you appeal to your broader audience. Don’t be afraid to lose followers in this manner – personally I’d rather one follower who offers a counterpoint to my views than two who don’t.

  1. Diversify your inputs

There’s really no better way to open up your horizons than by drawing inputs from across multiples streams; Reddit, Twitter, LinkedIn, Facebook, Books, Blogs, Conferences, Periscopes, Meetups are all ways to try and seek out new contacts. Try to actively look for communities that do things differently, or might even actively disagree with you, and try to shift your perspective to theirs. There is no right or wrong solution and altering your perspective can make the world seem a very different place.

  1. Challenge the status quo

It’s okay to disagree now and then if you do it the right way – there’s a balance between being “that guy” and debating productively with someone who is willing to listen. Be especially careful of providing a dissenting voice if you’re new to a community e.g. a Qlik user in the Tableau community might see his/her views dismissed. However, don’t disagree on everything, it get’s tiresome in communities to see constant disagreement (further reading here from Ben Jones).

  1. Avoid being a fanboy / girl

In the same way then agreeing, retweeting and liking everything adds very little value to a community. Work out why you’re in the community; do you want to help new users, publish your own content, get help to solutions? Develop an online profile / personality around those interests and share content while adding your own comments. Followers will engage much more with cultivated, meaningful content that you have added value to.

download

  1. Don’t take feedback too much to heart

Positive feedback feels great, it’s sometimes overwhelming to have your visualisation praised by the community and it’s easy for it to go to your head but be aware that that is only likely to be one viewpoint, albeit a shared one. Learn to critique your own visualisations and rely less on likes / retweets / Viz of the Day as a way of judging a projects value.

Similarly just because a project gets negative (or worse no) feedback then that doesn’t mean it doesn’t have value. Social Media can be very fickle, things on a populist theme will get much more attention than anything of genuine business value.

  1. Seek out feedback from alternative sources

Seek out alternative feedback from different communities or on different platforms. One of the best ways is to ask for honest feedback from one or two trusted contacts / experts privately where they are more likely to give your work time and energy as opposed to glancing over it, or simply hitting the retweet button. The value of one meaningful critique like this is not to be underestimated, 140 characters is isn’t enough to get any meaningful feedback and so many people won’t bother.

  1. Don’t just take my word for it

Do you agree? Look for other methods of avoiding the bubble online, a lot has been written about the social media bubble. Discuss, comment, argue and debate with me – I’d love to hear from you. I’m happy to be wrong.

What does this mean for me personally?

I’ll be the first to admit I’m well inside the bubble myself. Very few of my contacts and peers in the world of data visualisation come from outside the Tableau world. I rarely use any other solution to build data visualisations and I fail to engage with any media away from Twitter and LinkedIn professionally. I could do a lot more.

Over the next few months I need to ensure I broaden my horizons using the tips above, work with new people and seek out their opinions. I intend to work with new communities and with new solutions to see the world from their point of view….and I’ll be richer for it.

.

Advertisements

2 thoughts on “Avoiding the Bubble – 10 ways to broaden your data visualisation horizons on social media

  1. Fascinating post, thanks. I got very interested in the bubble after hearing about Eli Pariser’s book “The Filter Bubble”, which obviously goes into a fair bit of detail on the subject.

    Brexit is a great example. If one was to base predictions on my social network (offline, but moreso online) then it was inevitably the case that the vote would be 95% remain. Apparently that didn’t reflect the UK population’s view!

    Similarly to your experience, there was only really 1 person whose Brexit case I found credible, largely because it was so disparate from my image of a Brexiteer. Since then I’ve seen a couple of “Lexit” articles that gave me some sympathy for it but, even if they had changed my mind (as it happened, they haven’t, but they have added nuance), it was too late. Likewise it’s hard for me at first glance to imagine Trump can win – but the fact he is in the position he is in is evidence in itself that it really could happen.

    I suspect this is increasingly the case in all fields of endeavour, yes, even dataviz 🙂 To some extent, I think this is probably an issue around the paradox of choice, leading to the problem of specialisation.

    We have so much potential knowledge and experience available to us so easily that it simply isn’t possible to know the entire scope of most fields, so we break into natural divisions that are more mentally manageable. For dataviz, probably most relevantly , as you say, into specific tools or techniques. It takes a lot of effort, worthwhile though it is, to tour other communities even at a light level, what with all the other bits and pieces life keeps us busy with.

    For instance, I must admit it has been a worryingly long time since I took a serious look at Qlik for instance. Not necessarily a great thing when perhaps what I’m paid for is really to derive the most insights out of data irrespective of method, rather than become the best ever user in a given tool (to be fair, workplace priorities obviously may be legitimately different for people working in a Tableau consultancy for instance!).

    But even the first version of Powerpivot, much as I didn’t love it, had a feature or two that I wish had been brought to other tools faster. Spotfire had some abilities that were way more advanced in certain domains that Qlik and Tableau had years ago, even though it didn’t make my favourites list at the time. Tableau is not objectively “better” than Qlik etc. for all use cases.

    There is some merit in most approaches I’m sure, and a quick review of e.g. the Gartner Magic Quadrants shows the range even within the top regarded tools in the field.

    So that’s one thing we can try and do; review other options within the same field. Few people choose to use a tool or technique that they think substandard after all, so most successful companies must have some merit! If want to go all out, we could try and steel-man the argument for “Tool [X] better than [Y]” substituting your favourite tool in for [Y].

    In a more generic sense, I think you are right re online communities. Facebook is obsessively filter-bubbly in my experience. You can even find out bits about what bubble it’s put you in – have a look at https://www.facebook.com/ads/preferences when you’re logged in for either some good laughs or worryingly accurate insights

    Twitter is less so (yet), but the need to explicitly follow people tends to have you recreate your own bubble, possibly an even worse one. I am new to it, but so far my forays into Reddit make me think that may be more representative (but representative of the type of people that can and choose to go on Reddit, not humanity as a whole!).

    Search engines suffer the same issues; Google tries to personalise, but some don’t (duckduckgo?), usually citing privacy as the issue, but a lesser bubble has to be a side effect. There’s also tools to try and “confuse” websites into unbubbling you – e.g. Trackmenot https://cs.nyu.edu/trackmenot/ (I haven’t used it). But of course, as with everything, there may be a downside in convenience; personalised search results are sometimes simply more useful than the non-personalised variety!

    Another thing I feel has somewhat helped me is to study what seems to be referred to as “intellectual traditions”, especially philosophy. In trying to broaden my mind beyond a more tech/data focus, I did a couple of courses and read into that. There’s a lot one could say, but I do think that it has helped me to “question everything” and have some insight into the psychological, logical or rhetorical fallacies that make the bubble so effective. A study of statistics,design of experiments, and particularly how to trick people with them, is probably useful, and also happens to be very good knowledge for data people.

    There are naturally some downsides – not least a realisation that I may be less right about some things than I thought!

    I also try and keep in mind that, on the whole, nobody thinks they themselves are evil or incompetent, and most people aren’t any more wilfully ignorant than I am. Books like Haidt’s ‘The Righteous Mind’ help reinforce this.

    Anyway, please excuse the rambling length of this comment; I look greatly forward to hearing about your adventures combating the bubble!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s