JeffConf Hamburg CfP/Agenda retrospective

superluminar recently hosted the JeffConf Hamburg (now known as serverlessdays) - a community conference focussed on serverless topics. Please find our experience report regarding the Call For Papers process and agenda here.

Agenda “committee”

The agenda team consisted of 3 people:

First of all, thanks to Susanne and Ben for co-organizing this community event! They sacrified a lot of their spare time to make it possible!

Numbers

We got 48 submissions via the CfP.

  • All submissions were in English.
  • The CfP Process was open for 3 month (started on 07.11.2017 and ended on 15.01.2018)
  • There was one month between the end of the CfP and the conference
  • The agenda included 17 talks including lightning talks

Provider / Content Diversity in the CfP

Content-wise we got the following distribution (separated by Cloud Providers):

  • 23 AWS (3 from AWS employees)
  • 3 Azure and 2 of them from Microsoft employees.
  • 1 Google
  • 1 IBM (from an IBM employee).

Speaker Diversity

Our internal diversity goal was 30% from people underrepresented in tech.

If we don’t count in workshop facilitators, that’s:

  • 3 of our 17 talks were from women, that’s 17%
  • 5 of our 17 talks weren’t from white males, that’s 29% (I am no expert here but I counted asian/indian as “not white male” - please correct me if that’s wrong - actually I don’t like this distinction at all since we are all humans, but it’s for the sake of statistics and diversity report here).

None of the CfP submitters were woman. In order to raise the women speaker rate from 0% we looked e.g. at speakerinnen.org for women speakers and we were really glad that we found Manu Rink there! Additionally we also scanned Twitter, LinkedIn, etcpp - and we were glad that Anna and Gillian joined us as well. In the end we contacted about 14 women and got 3 confirmed in total. We did not send them through a CfP process, though, but let them choose their topic.

CfP Process

  • We had a “committee” of 3 people. 3 seems to be good because it’s an odd number and gives us quorum.
  • The CfP process was anonymized, a feature of PaperCall.io.
  • The CfP was open for three month, AFAIR we did this because re:invent was in between and we wanted to give those people who were preparing for / exhausted from re:invent a chance to submit a proposal as well.
  • We did not have explicit criteria for the selection. E.g. how the content should be mixed.
  • We did a “preselection” in the middle of the running CfP. I think we did this as a result of the long running CfP. In hindsight we should not have done a ‘mid selection’ because that lowered the chance for later CfPs to be accepted.

Why so much AWS content?

One feedback that we got was that there was much AWS content. In fact every second talk had at least some AWS content. Here comes the (hindsight) explanation:

On the one hand, have a look at the CfP numbers. We as a committee were especially looking for practical use cases from practitioners of Serverless/AWS and it looks like there are the most on the AWS boat. This was why we chose for example “Rule-based dispatching of events to a serverless services armada” or “Evolutionary Serverless Architecture: From hackathon to product” which showcased use cases built with AWS.

In fact we did not get any “this is my serverless architecture” submissions from practitioners who run Serverless GCP/Azure/IBM stuff in production.

Another factor was that we also preallocated some talk slots with “famous” speakers before the CfP started. And in the end most of them also came up with AWS topics. This was clearly my fault since I gave away the slots personally so that the committee was a bit limited in the slots left for talks coming from the CfP.

What also played in here was the fact that we pre-allocated the slots for 3 speakers from underrepresented groups (a.k.a women in this case and it turned out that 2 of also showed up with AWS content. We did not send them through the official CfP process. Don’t get me wrong here. This was clearly an issue on our side.

Learnings

  • Serverless is big enough, we don’t have to fear that we don’t get enough submissions. We don’t need to “pre-allocate” slots. Thus we can also better control content diversity and have a more transparent workflow for the submitters.
  • We need to do our homework as the committee.
    • We had only 1 month between end of CfP and the conference. We got the feedback that this is not enough and there should at least 2 months in between. The reason is that 1 month might be too little time for planning / talk preparation and so it does not make the event more inclusive. In fact it even restricts one from submitting the conference at e.g. callbackwomen.org who have a strict policy.
    • We need to have a clear and publicized timeline how the CfP process works. This should include:
      • start/end date, voting/selection phase dates, notifing speakers dates. Date of published agenda.
      • how many slots will be available, how many are reserved, e.g. for diversity initiatives?
  • We did not use the CfP tool (PaperCall.io) very extensive. For example instead of the native voting feature we used another spreadsheet where we could see the votes of the other “committee” members. That might have resulted in bias. I can’t remember anymore why we did this.
  • We need to make sure the content is more diverse, e.g. even if the market is clearly dominated by AWS, we should have more content from other providers or other startups in the serverless space. What can we do regarding Content Diversity and talks from practitioners who use providers other than AWS? Maybe we could for example ask Azure, Google and IBM if they could spread the word in their local communities.
  • Don’t have the CfP open for so long. Don’t have re:invent or other big conferences interfere with your CfP. Or simply don’t sacrifice a 2 month minimum interval between end of CfP and conference for these reasons.
  • Gender diversity is still hard, but it’s worth the effort and I was proud that we got three women on stage even if we failed our internal goal of 30%. Their talks have been perceived really useful and high quality. Thanks again Manu, Gillian and Anna!
  • The next time we can also make more explicit statements about the travel expenses policy. This time we only gave a hint that we might be able to cover expenses for some speakers because we had no idea how our budget would look like in the end.
  • We could use the “Invite Speaker” PaperCall feature the next time for “direct invites”. This makes sure that we can both invite speakers AND still have the same process for everyone.

By the way, the serverlessdays London organizers did a great job setting up their CfP and already incorporated our learnings :)

comments powered by Disqus