linux.conf.au 2018 Sydney – A little bit of history repeating

This year, linux.conf.au 2018 headed back to Sydney, where it hasn’t been held since 2007. This year I skipped quite a few sessions due to having Linux Australia duties and tasks to do, and because the heat and humidity were exhausting. Thankfully, the videos by Next Day Video were released very quickly, so I’m spending “Week 2” of linux.conf.au catching up!

On reflection, several themes came through.

  • Volunteers, volunteering and volunteer labour – There are several free software and opensource organisations across the world, and they’re all vying for volunteer contributions. Moreover, the volunteer base itself is ageing; we’re getting older and having children and families and other family responsibilities – we simply don’t have the time to contribute that we once did. At the other end of the demographic curve, younger people don’t have the same passion and ‘fire in the belly’ for free and open source software. In one sense, that’s a product of the success of the free and open source software movement – because it’s been normalised; but on the other hand this leaves us with a gap in the ‘compelling-reasons-to-join-a-free-and-open-source-project’ list. As a concrete example, during the opening of linux.conf.au, no less than three organisations – Open Source Initiative, Free Software Foundation, and Code Club Australia – did a shoutout for volunteers. At the same time, Linux Australia – the auspicing body of the conference – had fewer nominations to its board than open vacancies. I want to be clear: The Organisers and Volunteers of linux.conf.au did a phenomenal job. They were dedicated, professional, resilient and awe-inspiring. As individuals, and as a conference team, amazing. Systemically though, open source has some major issues to address to avoid burnout, and worse, resentment.
  • Infrastructure-as-code continues to gain maturity – As more and more devices become internet-connected, and we’re managing more and more devices, we need better orchestration. We’re seeing this manifest in container-all-the-things, in MQTT for unified messaging and in our approach to IoT hardware and open hardware. Standards however remain a barrier to interoperability and greater maturity in code-based orchestration, as outlined brilliantly by Kathy Giori.
  • Open source touches many disciplines – the range of Miniconfs available this year sent a strong and undeniable message – free and open source software, hardware and practices are touching many disciplines. Art, genomics, games, galleries, libraries and museums (GLAM) – Linux and open source touch each of these in fundamental ways. Personally, I’m delighted to see this cross-pollination happening in our communities. Together, we do better.

On communities, volunteering and volunteer labour

“A division of labour in free software” – Molly de Blanc, Free Software Foundation

Molly’s talk used the results of different surveys of opensource communities to show visually that labour in free software is gendered, ageist, and that these schisms also apply to what is considered technical and non-technical work. The implications of these findings are that these patterns are repeated without intervening action, such as having quotas on leadership boards. Importantly, anecdotal data shows that we still value technical work over important non-technical work; people still justify their non-technical contributions to an opensource project by emphasising the technical contributions they do make.

This resonated strongly with me; as the leader of an organisation that turns over around $AUD 1 million a year – Linux Australia – there are a number of skills I need to have – budgeting, strategic communications, strategic and operational management – and of course, the ability to be an efficient administrator. None of these are technical skills; yet, as the leader of a technical organisation I am expected to have a strong grasp of technology issues. Even in a non-technical role, you’re not allowed to be non-technical.

https://youtu.be/6NDB2VFYlfg

“Dealing with Contributor Overload” – Holden Karau

Holden Karau is a core contributor to the Apache Spark project, and this war story and guidance was learned the hard way – when the project became so big that contributors were significantly overloaded. She provided a number of strong pieces of guidance for dealing with contributor overload, including:

  • Developing a contributor pipeline to allow users of the project to become contributors, and in time, core committers
  • Not ‘raising the bar’ for changes and requests because these have very unattractive downsides such as making the contribution pipeline harder and paradoxically increasing the contributor workload by increasing questions and requests for assistance.
  • The power of having clear roadmaps which make it clear what the core project is, and is not going to do, so that people can either start their own project, or plan around it. The Roadmap also helps guide contributions, and show how smaller tasks contribute to larger milestones.
  • Focussing on committer productivity – such as better tools to merge changes, making it easier to review changes, and more tests – can have significant long term dividends. Imagine what a 1% productivity increase would mean across say 10-20 committers? 50 committers? 100 committers?
  • Creating safe spaces to ask questions and contribute without being mocked – people who feel safe to fail are going to commit more.

https://youtu.be/BempWfBkvs8

 

“Burning Down the Castle” – Daniel Vetter, Intel (previous graphics kernel maintainer)

Daniel’s talk was an eye-opener. As a previous graphics kernel maintainer, Dan has seen a whole range of poor behaviours that contribute to maintainer burn-out, rage-quitting and other unproductive outcomes. His talk advocates for a kinder, gentler approach to maintaining a technically elite community.

 

https://www.youtube.com/watch?v=BB0luXmuo3g

 

“Mirror, mirror on the wall: testing Conway’s Law in open source communities” – Lindsay Holmwood

Lindsay provided an outline of Conway’s law of organisational communication patterns, and the concept of mirroring – the mapping between the organisational structure and the supporting technical structures for communication. Strong mirroring leads to strong ownership – you are led to the actors who own a system. Using an overview of the empirical literature on organisational development and he explained how organisations try to solve the problem of communication – using different structural strategies. But mirroring works poorly in unstable environments – those undergoing radical change and innovation. This has led to the rise of structures like guilds. These theories are then applied to open source to show that shifts away from the ‘core’ of an open source project can indicate a decline in the project itself. This necessitates a need to build a pipeline – again the pipeline – of people moving closer to the core in their contributions.

This talk was intense – but the key takeaway was that the way we design organisational structures has a significant impact on organisational outputs and long term organisation success. This is of particular importance for projects that are scaling up significantly; poor choices during scale up will lead to poor productivity later in the project’s lifecycle.

https://www.youtube.com/watch?v=xYkh1sAu0UM

 

Orchestrate all the things. With code.

“MQTT as a Unified Message Bus for Infrastructure Services” – Matthew Treinish

This was an excellent talk by Matt Treinish, who outlined the reasons behind the design of MQTT, which was originally designed for sensor telemetry. He goes on to show there are different levels of quality of service for the broker. An excellent introduction to how MQTT can be used as a unified messaging bus – as used in FireHose.

https://www.youtube.com/watch?v=y6xN6S407Xc

 

“What does the buyout of @arduino mean for #openhardware?” – Kathy Giori, IoT at Mozilla

I was truly disappointed not to be able to make it to Kathy’s presentation, as it came about partially because of a tweet I’d sent out to #lcapapers in mid-2017 – and which Kathy shouted out to me for. Thank you, and apologies for not being there in person.

Giori provided an overview of the corporate history of Arduino and how it’s now consolidated under one company; lamenting the drawn-out legal process that led to this point.

She continued to outline some of the challenges in licensing for open hardware and how manufacturers are being cheated by lower-quality knock-offs; with those same manufacturers then expecting the original author of open hardware / open software to provide ongoing support. This led to a discussion on the different levels of openness in open hardware, and the pros and cons of each.

Concluding the talk, Kathy provided an overview of the Mozilla Web of Things project, which is attempting to bring some standardisation and streamlining to the very fragmented IoT and open hardware space. There are competing standards, competing platforms, and the piece that I didn’t realise was that this is actually inflating costs for consumers. Because individual companies need to make hubs and supporting infrastructure for “their” range of IoT hardware, this means each endpoint device – light bulb, sensor, thermostat and so on – is quite expensive. Mozilla is seeking to have stronger interoperability in this space by creating the ‘Web of Things’:

“The “Web of Things” (WoT) is the idea of taking the lessons learned from the World Wide Web and applying them to IoT. It’s about creating a decentralized Internet of Things by giving Things URLs on the web to make them linkable and discoverable, and defining a standard data model and APIs to make them interoperable.”

If anyone can drive this, Mozilla can, but my personal feeling is that they’re going to come up against significant corporate interests in doing so – at a time when their own corporate mis-steps (Mr Robot, anyone) have significantly backfired. I live in hope.

https://www.youtube.com/watch?v=x2ltqoAqJbY

Cross-pollination, because together we do better

“The Future of Art” by J Rosenbaum

This was the mind-blowing talk of #lca2018 for me personally. Academic and artist J Rosenbaum took us through their research, which sits at the intersection of machine learning, neural networks and the production of art.

J’s talk started with an overview of machine learning projects, such as Botnik and Janelle Shae, and moved on to underscoring the collaboration between human and machine in generative art.

The future is not man versus machine  – the future of art is man with machine.

https://www.youtube.com/watch?v=lTT2mq692JQ

 

“The Knitting Printer” by Sarah Spencer

Again a brilliant intersectional talk by Melbourne-based hobbyist and knitter, Sarah Spencer, in which she provides an introduction to knitting machines, and provides a breakdown of how she reverse engineered a hack to a 32-bit knitting machine to be able to get images from her computer to the knitting machine.

Massive respect, @chixor.

https://www.youtube.com/watch?v=Y6k15pdFTsA

 

“Wearing access: a story about open collections, a sewing machine and the nation’s secrets” – Bonnie Wildie

Bonnie’s talk, from the OpenGLAM Miniconf, was very much a hidden gem of the conference. She talked about the concept of redaction art, created from files that have been redacted – and remixed. Bonnie even turned the redaction art into a dress, which opened up a conversation on the politics and power of what we wear. Dress and costume become media for subversion. Much awesome.

https://www.youtube.com/watch?v=XhTzE67HrhE

linux.conf.au 2016 Geelong – LCA By the Bay

So, it’s been about six weeks now since linux.conf.au 2016 Geelong – LCA By the Bay concluded, and after a lot of sleep and catching up, it’s about time to pen some thoughts about the process, the experience and learnings.

Getting linux.conf.au to Geelong

When you think ‘epicentre of opensource’, Geelong is not what comes to mind. Well, it’s not what used to come to mind! So, how did we bring linux.conf.au to Geelong?

Late 2013

Firstly, we needed some core people to put a bid together. David Bell and I had worked on BarCampGeelong before, and had semi-seriously considered bidding previously. We felt that we had a complementary set of skills, and the drive, leadership and passion to make it happen.

We worked with Business Events Geelong, and the wonderful Terry Hickey, to put together a bid document, covering key aspects of what linux.conf.au in Geelong would look like. Business Events Geelong were able to assist with a professional bid document template, and with sourcing pricing to include in the bid document. A couple of hours later, and we had formally submitted our bid to host linux.conf.au!

linux-conf-au-geelong-bid (PDF, 1.5 Mb)

linux.conf.au is chosen by a committee of trusted senior members of Linux Australia, the organisation that umbrellas linux.conf.au and a stable of other events, such as PyconAU, Open Source Developers Conference and a number of WordCamp and Drupal events in Australia. Linux Australia calls for expressions of interest from teams interested in running linux.conf.au every year – bids – and forms a small committee to evaluate the submissions. This normally involves travelling to the bid city, and assessing elements such as;

  • accommodation
  • conference venue
  • transport to and from the conference
  • conference event locations

April 2014

Terry and the Business Events team were amazing at hosting the bid team, and showcased a number of Geelong’s leisure and recreation offerings, cementing the quality of our bid. It was a great opportunity to learn from the bid team, as they assessed our risk management, our planning and our ability to pull together such a large event.

Venue visit for #lca2015 #geelong. Beautiful.

A photo posted by @kathyreid_id_au on

Although Auckland were awarded linux.conf.au for the year ahead (2015), the decision was made to award Geelong linux.conf.au 2 years out. This was an excellent decision, and provided long term stability not only to the event, but also provided the conference team with a longer term planning horizon.

Woo! We won a bid for linux.conf.au! Now what?!

Once we had a strong idea of how the main conference venue (Deakin University’s Waterfront Campus) would work, we focussed our efforts on preparing to showcase Geelong as an outstanding venue at linux.conf.au 2015 in Auckland. Often, the next year’s conference prepares promotional material or flyers to help encourage conference attendees. We had decided on our conference theme of

life is better with linux

and in keeping with the theme, worked with Martin Print to have NFC keyrings printed.

Now the hard work began. Firstly, we needed to ensure that our conference management system was functional. linux.conf.au traditionally runs on a piece of software called ZooKeepr, and it needs a bit of maintenance each year. Luckily, we had Josh Stewart and James Iseppi to give us a bit of a hand, and with David Bell being generally awesome with anything technical, in no time we were able to get ZooKeepr ready for the Call for Papers.

Call for Papers (#CfP)

The Call for Papers (#CfP) happens about 6 months before the conference, and the challenging part for conference organisers is ensuring not only that there are a large volume of submissions, but that the quality of submissions is of a quality fit for an internationally renowned conference. One of the ways in which the conference spreads the word about #CfP far and wide is to reach out to all past Speakers of linux.conf.au and encourage them to make submissions. We also lean heavily on the Papers Committee, the group of senior and respected Linux Australia members who review the #CfP submissions and make recommendations to the conference team on which submissions should be accepted into the conference.

This year, the conference team decided to add another type of submission to the mix – Prototypes – alongside the standard 45-minute Presentation and 110-minute Tutorial. This worked out wonderfully and some of the most popular talks of the whole conference were submitted as Prototypes – including the crowd-favourite Linux-powered microwave by David Tulloh.

https://youtu.be/R3DADx5z-XY

Thanks to the efforts of Papers Committee and past Speakers, we received almost 300 submissions, and the overall quality was excellent. The Papers Committee spent a day in Sydney in in August making some very tough decisions, and after around 10 hours we had our Schedule! I was incredibly impressed by the talent in the room, and the generosity of the Papers Committee to give up their time – and in many cases their own coin – to travel and attend.

Schwag

While I was busy liaising with Speakers and getting travel organised, and David was busy with event venues for our conference events, Sae Ra Germaine was being a superstar with our schwag. She found an excellent supplier for our conference bags, Ecosilk, and designed a contemporary yet simple t-shirt for our delegates (navy) and volunteers (orange). She also worked to ensure that we had sunscreen  and hand sanitiser as part of the Schwag bag.

#lca2016 all tired out.

A photo posted by ms_mary_mac (@ms_mary_mac) on

Sponsors

David took a strong leadership role in Sponsorship, and developed a Sponsorship Prospectus, and negotiated sponsorship agreements with all of our fabulous sponsors. Many of our Sponsors support linux.conf.au year after year – without them, the conference wouldn’t happen. One of the challenges the conference has is having to re-establish sponsor relationships year after year, and our Ghosts debrief session and good documentation helps to ensure continuity.

Venue and catering

Deakin University’s Waterfront Campus and Costa Hall are beautiful architecturally, and provide a wonderful environment for collaboration and learning. However, the campus cannot hold 600 conference delegates in a 5 stream conference easily. So, we worked with the National Wool Museum, located a block away from Deakin, who had a conference room available. Another benefit of this arrangement was that delegates were able to see the jacquard loom – programmed via punch cards that the Museum had in their collection.

Patching a bug on a two story high computer. #lca2016 A photo posted by John Dalton (@varrqnuht) on

We worked with Waterfront Kitchen to arrange lunch options for delegates, and arranged to have menus placed in the Schwag bag. WFK also handled all catering for the conference, including morning and afternoon teas. We also made the decision to have core team and volunteer lunches fully catered, so that we could free up time during the busy conference period, and this proved to be a wise choice. We received nothing but positive feedback from our delegates regarding WFK’s catering – the variety, the attention to detail and handling of special dietary requirements.

By November, organisation of event venues was in full swing. linux.conf.au has three traditional conference events – Speakers Dinner, Professional Delegates’ Network Session (PDNS) and the main conference dinner, the Penguin Dinner. Speakers Dinner was held at the fabulous Balmoral at Fyansford, with the Limoncello String Quartet for music.

PDNS was held at the fabulous Little Creatures Brewery, and it was perfect. Great beer, great food and great company. It was amazing to see over 300 people of linux and opensource having a great night out.

 

Our Penguin Dinner was at the fabulous The Pier restaurant, and was an amazing night out for all concerned.

AV and Networking

A great conference needs great AV and networking, and we were fortunate to have some wonderful people, including Andrew and Steven working with us. The networking crew laid over 200m of fibre optic to the Wool Museum so that they could have solid internet, and we utilised the services of AARnet for our on-campus networking. Deakin University also provided phenomenal support, working with AARNet to provide strong wireless across the conference venues.

A great team

There were so many different parts of linux.conf.au that had to come together to make it an excellent conference, and the entire team needs to take credit for that. Aaron, who co-ordinated our childcare arrangements, which was greatly appreciated by attendees, Brittany whose excellent accountancy skills kept us very well budgeted, Michael whose social media prowess ensured we trended nationally, George who provided a helping hand where it was needed, Erin who was our Rego super-hero, Josh who helped us keep ZooKeepr and our payment gateway under control, Daniel our stellar volunteer co-ordinator and Brett whose photographic talents and video production blew us away – every single person was part of an amazing, productive, motivated and awesome team that I was so incredibly proud to be a part of.

LCA2016 - Wednesday

Open Source Systems – Mobile FOSS workshop

The last day of the Open Source Systems conference was dedicated to two workshops – one on free / libre and open source software (FLOSS) in education, and the other to the same in the mobile sector – denoted ‘mFOSS’.

Led by Tony Wasserman, a professor at the Silicon Valley campus of Carnegie Mellon University, the goal of the workshop was the creation of a research agenda for open source software in the mobile environment. He set the scene by illustrating how the number of mobile applications had grown from zero to almost a million in just a few years, and placing this in the context of cloud computing; the code used to write the application itself may be open source, the platform it’s targeted at may (or may not be open source), the storage may be open source and the development tools used may (or may not be) open source.

The workshop raised several key discussion points.

Where are the open source mobile apps?

The participants discussed why we are not seeing more mobile applications released under open source licenses. This included barriers such as difficulty with the current dominant mobile app stores – Apple’s iTunes platform and the Google Play store – with open source licensed applications being disallowed with Apple. This gave rise to discussion on whether there would be the rise of a ‘SourceForge’ or ‘GitHub’ style of mobile application store geared toward open source apps. It was also noted that many apps were hosted using open source resources – whether it be storage, platform or even development tools – but they themselves as end products were not open source.

What makes mobile different?

Discussion was also centred around differentiating mobile application development from traditional application development – and a long list was generated;

  • In mobile applications, there are many hardware and software platforms to cater for, and the users of them expect a continuous or familiar experience.
  • From a user interface perspective, there are many form factors to cater for.
  • There are many development platforms and languages to choose from when developing, and although there are some dominant ones, the market is essentially in a state of flux.
  • Traditional applications are not required to be context aware – and mobile applications must effectively sense and use context through sensors such as accelerometer, touch and swiping, GPS and location based services, cameras and voice input.
  • Applications are much more distributed – the hosting may be in one country, the storage in another and users in several other countries
  • This leads to a related issue of the complexity of testing mobile applications – with different operators and different gateways in different countries – essentially the number of intermediaries between the device and the internet is greater.
  • Internet connectivity on mobile devices is not continuous – and mobile applications need to be able to gracefully handle loss of connectivity. A related issue is the availability of 3G and LTE/4G connections in some communities – particularly remote communities – where data services are not available all the time. Can the application detect the network on which it’s running and adapt to the internet connectivity (or lack thereof) that prevails? Effectively this calls for ‘responsive design’ – not to the form factor but to the network. As a related point, SMS rather than 3/4G is the transport layer of choice to reach the most people, and in developing countries there tends to be a high penetration of mobile phones, but they are basic feature phones, not smartphones. Designing around SMS is very different to internet based applications.
  • While mobile device technology has advanced significantly in recent years, it’s still not perfect. Battery life and device power consumption needs to be a key factor in application design – as it’s possible to build a fantastic app that drains the device in a matter of minutes, effectively rendering it useless.
  • The social element of app design is becoming ubiquitous – and must be factored in – such as for social shopping.

What technical challenges remain?

There are very few standards for mobile devices – for screen resolutions, data handling and even privacy and security. Touch interfaces and other sensors differ between hardware and software platforms. There is scope for research to identify ways that effort for cross platform development could be minimised.

Testing remains a challenge also given the heterogeneity of hardware and software. What works well on one device under one se tof conditions may not work well on another.

Load is also an issue, particularly when the user base of a mobile application grows large. How is app availability under high loads assured? This has implications for platform services like EC2 and Amazon web services.

One point of discussion was differentiating between ‘building’ applications and ‘engineering’ them – the teepee versus the Taj Mahal. The key question of course comes back to fitness for purpose – is the application doing what it was designed to do, and is it doing it well?

Another challenge is the nature of application feedback. If applications break, they garner negative reviews, which persist even after the bug might be addressed. This introduces the notion of community – which forces a degree of testing and quality assurance.

Monetisation also remains a challenge, with people very reluctant to buy mobile applications, even though their handset costs a significant amount of money – so different business models are likely to emerge.

Privacy and security of personal data are also issues worthy of attention, particularly with the distributed nature of mobile applications.

In general, this was a worthwhile session which captured a number of the issues currently facing the mFOSS community.