Signs of Triviality

Opinions, mostly my own, on the importance of being and other things.
[homepage] [index] [] [@jschauma] [RSS]

Beware the Conference Echo Chamber

March 29th, 2013

feedback loop Lately I've been thinking about conferences a lot. More specifically, about how the concept of the filter bubble applies in this world. Following my Twitter timeline, I often see a lot of the people I follow enthusiastically attending or talking about the same conferences -- ironic evidence of precisely the condition I'm lamenting. In short: most conferences today appear to be nothing but a positive feedback loop rather than really pushing forward their particular discipline.

This observation is restricted to my very own filter bubble, of course. That is, the conferences I'm talking about are mostly those in the System Administration, DevOps, WebOps, and popular Security area, and I believe that those of us who organize these conferences could learn a few things from academia (which, to be clear, is likely to suffer various problems of its own).

I've talked about this at this year's DevOpsDays NYC, but let me summarize my points made there. Reasons given when asked why we like to attend conferences include:

  • To share research and experiences.

  • To network with peers.

  • To learn new things.

But there are other reasons, if we're honest, though perhaps those are just different interpretations of the above. After all, we often like to present ourselves differently from our honest motivations[1].

We attend the conferences that cover the areas we work in and hear speakers talk about topics that we believe are relevant to our current interests. Most of the time we seek out validation of our own ideas, and so it is not surprising that most conferences have a very homogenous program. Within each community, we cater to the masses.

This reinforces each attendant's feeling that they get their money's worth out of a conference: if you know before attending that you will agree with everything (or at least the majority) that's said, then you later on are likely to rate the conference a smashing success. Which in turn leads to the conference organizers receiving lots of great feedback, patting their program committee on the back and planning an identical (but bigger!) conference the next year.


I understand that every conference caters to a certain group and that a certain like-mindedness is a prerequisite, but the degree of homophily exhibited is stark, and in my opinion detrimental to the idea of fostering new ideas. Look at the speaker line-ups and topics presented at the last couple of BlackHat, OSCON, Surge, or Velocity conferences[2], and ask yourself just how much new stuff you will actually find at this year's version. You can observe a similar echo chamber effect at much smaller conferences as well: Monitorama's speaker line up and attendee list seems largely congruent with the regular DevOpsDays events.

Now hanging out with people who agree with you and discussing common issues can be a lot of fun. And who doesn't want to consider themselves a "thought leader" by way of association with the popular crowd? But here are a few things that I'd like to see conference organizers to adopt (perhaps only in part), as I believe they'd make conferences much more diverse and interesting:

  • use a single track -- multiple tracks create competition amongst speakers (forcing them to present more "popular" talks) and frustration amongst attendants (who have to choose, easily letting the appeal of the familiar lead them to what they already know)

  • no speakers who presented the previous year -- this helps make sure that you are not just repeating last year's conference; this also helps more people with little speaking experience apply and new ideas be presented

  • no previously given talks -- standard in academic environments, our industry would do well to require original work not presented elsewhere

  • double-blind proposal submission -- knowing who will evaluate my speaking proposal makes it easy for me to (try to) write the proposal specifically with the interests of the committee in mind; on the flip side, double-blind submission procedures have been shown to significantly reduce the effects of (the committee's members') status, race, and gender biases

  • invite keynote speakers with opposing views -- every conference supports (consciously or unconsciously) certain core beliefs; expose your audience to the diametrically opposed viewpoint, attempt to pierce your audience's filter bubble

Many people choose conferences to attend and talks to listen to primarily by speaker. We want to be entertained first[3]; subconsciously we seek approval by our peers, while simultaneously ensuring that we don't venture too much out of our comfort zone. Since we're guaranteed to have a great time, we are actively reinforcing the idea of what we think makes for a good conference, closing the feedback loop. I'm as guilty of following this pattern as anybody else, but I try hard to at least be aware of my own filter bubble and periodically try to peek past the perimeter.


March 29th, 2013

[1] On a recent visit to the Twitter NYC office, Hilary Mason noted the discrepancy between the links what we retweet versus what we click on and actually read: we retweet those that seem profound, clever, and conscientious, but we actually read the profane, the gossipy, the sensationalist.

[2] I've attended and spoken at some of these myself. They're all wonderful fun, and I've enjoyed the program, the speakers and familiar faces in the audience; I'm sure, I'd have a great time attending the next iteration. But knowing this for certain is precisely the problem.

[3] While I agree that good speakers (and/or teachers) often have to be good entertainers, I disagree that this is their primary function. Attending a hilarious talk in which I'm not challenged or do not learn anything new is like watching a soap opera: easy to get sucked into, and easy to repeat, but forgettable and ultimately less rewarding than a movie that makes you think.

Related links:

[Yahoo!'s "Must Not Work From Home" Mistake] [Index] [Sharing Secrets Using SSH Keys (II)]