PyCon AU 2019

As folks probably know, this year I’m in Canberra at the 3A Institute at the Australian National University, as part of the first cohort of Masters in Applied Cybernetics – itself an experiment in helping to build a new applied science in taking artificial intelligence and cyber-physical systems safely to scale. I haven’t written much about the experience to date; nominally because the course is quite intensive, and doesn’t leave a lot of free time, and, importantly, because many of the reflections that surface belong within that experience.

That said, the course is broadly structured in two brush strokes; a theoretical foundation in multiple disciplines – anthropology, sociology, computer science, cybernetics, artificial intelligence, machine learning and human computer interaction – and a technical foundation in several components – Git, Arduino, 3D printing, and, importantly, Python.

To this end, many of us from the cohort headed to PyCon AU 2019 in Sydney as an excursion to gain a deeper understanding not just of the language, but the community, the ecosystem, its standards, its norms and its idiosyncrasies.

Members of the 3A Institute – Staff and students – at PyCon AU 2019
From left to right: Tom Chan, Matthew Phillipps, Zaiga Thomann, Katrina Ashton, Felicity Millman (seated), Elizabeth Williams, Zac Hatfield Dodds, Kathy Reid, Charlotte Bradley (seated), Sam Backwell, Danny Bettay, Alison Kershaw
Photo courtesy of E. Dunham

Aurynn Shaw – Making lasting change

Aurynn has written extensively on the topic of contempt culture; the default communications culture we have adopted both in Western society and in tech organisations in general. Contempt culture is where we identify ourselves by being pejorative to other languages, practices and disciplines. We create distance rather than intimacy – hostility rather than harmony. Aurynn’s keynote called for long lasting cultural change by understanding what businesses really need – they care about stability, and risk and outcomes. As technologists we need to communicate these pieces effectively. Businesses have stopped listening to us, because we haven’t cared about their needs in the past.

My key reflection here is that multi- and trans-disciplinary practice will be a key feature of work of the future. As practitioners, we will need competencies in understanding, responding, communicating and engaging with the needs of others in ways that build rather than erode social and intellectual capital.

Watch Aurynn’s presentation on YouTube here.

Mark Smith – It’s Pythons All The Way Down: Python Types & Metaclasses Made Simple

Even though I have a deeply technical background – see my toolchain if you’re in doubt – Python is still a bundle of dark magic in many ways. Mark’s talk was well structured, conceptually clear, and disambiguated types, classes and inheritance patterns in an engaging way.

The talk here covered basic data types in Python – Python is a strongly typed language – and then went deeper into the rabbit hole of types and meta classes by explicating the different types of inheritance that the language uses.

Kudos to Mark on an excellent presentation here – it’s a fantastic introduction to Python types and metaclasses for people transitioning from other programming languages.

Watch Mark’s presentation on YouTube here.

Amanda J Hogan – Pretty vector graphics – Playing with SVG in Python

Why would you do SVG in Python rather than Javascript? Excellent question, and Amanda’s talk shows why Python is a fantastic tool for experimenting with generative art. The talk covered basic SVG primitives, and went on to more complex constructs such as paths and symmetry. Her work with algorithmic generation of art using SVG was incredible – both the mathematical approach and the end result.

Amanda’s work is beautiful, engaging and an intriguing ‘in’ to working with generative art.

SVG linters

One comment I did ponder at length here was the lack of SVG linters available – in Python or otherwise. Having used Inkscape for several years – whose default output format is SVG, I was surprised that there weren’t more linters or validation approaches in this space. So, I went digging and found this linter on GitHub. Unfortunately my npm is segfaulting at the moment and I don’t have time to debug it, so it will have to wait.

Watch Amanda’s presentation on YouTube here.

VM Brasseur – The real costs of Open Source Sustainability

Grounded by Nadia Eghbal’s excellent “Roads and Bridges” report, Vicky’s talk started with a critique on the current methods for improving open source sustainability – for example the funding that comes in for donations needs to be managed – through legal entities, through tax. This can place additional burdens on maintainers.

Drawing from the experiences of other communities, she laid out a three-pillared plan for improving open source sustainability;

  • Contributing back: this can take the form of “time, talent or treasure” (per Tiffany Farriss) – contributions don’t have to be about just money. We need to find ways to attract non-technical talent to open source – and that starts by valuing it on an equal footing.
  • Human and environmental diversity: Diverse teams outperform non-diverse teams. We especially need to look at linguistic diversity – the open source world is English-centric. We also need to think about open source in the supply chain. What happens when open source companies are acquired and absorbed? Open source can also be single provider, single service, auspiced by a single foundation – and thus a single point of failure.
  • Community safety: Creating psychological safety for teams allows tough conversations to take place.

View VM’s talk online here.

Book review: Made by Humans by Ellen Broad

Made by Humans: The AI Condition by Broad, Ellen

My rating: 5 of 5 stars

Ellen Broad’s multi-faceted exploration of the many inter-twined aspects of artificial intelligence embarks and concludes at the same salient juncture; emerging technologies are conceived, shaped, used, governed and iterated by humans. Just as humans are inherently neither good nor bad, the systems we construct echo our moral plurality, our unconscious bias, and, frequently, our unwillingness to be critically interrogated.

That this is Broad’s first book – given its well-researched examples, coherent structure and intellectual incisiveness – is surprising. Its clarion call – for greater care, more rigourous thinking and a more holistic approach to the almost-infantile adoption of artificial intelligence, machine learning and autonomous decision-making – is not.

Structured in three distinct parts – Humans as Data, Humans as Designers and Making Humans Accountable, the book covers much territory. From systemic and cultural biases in how data used by machine learning is selected and captured, to the errors that are introduced to data sets by humans, to decisions made about system tradeoffs, what privacy means in different contexts, how open a system is to inspection and intelligibility, how diverse that system is, to who is accountable for the impacts of a system, real life examples are interwoven with provocative and often confronting questions.

Broad does not set out – at least in this tome – to answer these questions – rather, she lays a foundation for examining each of these questions in more depth. Personally I’d like to see a follow-up to this that covers attempts to standardise practices in machine learning and artificial intelligence – the frameworks and benchmarks – often competing – that have been proposed – alongside efforts at industry adoption and (likely) the barriers that are faced.

linux.conf.au 2019 Christchurch – The Linux of Things

linux.conf.au 2019 this year went over the Tasman to New Zealand for the fourth time, to the Cantabrian university city of Christchurch. This was the first year that Christchurch had played host and I sincerely hope it’s not the last.

First, to the outstanding presentations.

NOTE: You can see all the presentations from linux.conf.au 2019 at this YouTube channel

Open Artificial Pancreas System (OpenAPS) by Dana Lewis

See the video of Dana’s presentation here

Dana Lewis lives with Type 1 diabetes, and her refusal to accept current standards of care with diabetes management led her to collaborate widely, developing OpenAPS. OpenAPS is a system that leverages existing medical devices, and adds a layer of monitoring using open hardware and open software solutions.

This presentation was outstanding on a number of levels.

As a self-experimenter, Dana joins the ranks of scientists the world over putting their own health on the line in the strive for progress. Her ability to collaborate with others from disparate backgrounds and varied skillsets to make something greater than the sum of its parts is a textbook case in the open source ethos; moreover the results that the OpenAPS achieved were remarkable; significant stabilization in blood sugars and better predictive analytics – providing better quality of life to those living with Type 1 diabetes.

Dana also touched on the Open Humans project, which is aiming to have people share their medical health data publicly so that collective analysis can occur – opening up this data from the vice-like grip of medical device manufacturers. Again, we’re seeing that data itself has incredible value – sometimes more so than the devices which monitor and capture the data itself.

Open Source Magnetic Resonance Imaging: From the community to the community by Ruben Pellicer Guridi

You can view the video of Ruben’s presentation here

Ruben Pellicer Guridi‘s talk centred on how the Open Source MRI community has founded to solve the problems of needing more MRI machines, particularly in low socio-economic areas and in developing countries. The project has attracted a community of health and allied health professionals, and has made available both open hardware and open software, with the first image from their Desktop MR software being acquired in December.

Although the project is in its infancy, the implications are immediately evident; providing better public healthcare, particularly for the most vulnerable in the world.

Apathy and Arsenic: A Victorian era lesson on fighting the surveillance state by Lilly Ryan

You can view the video of Lilly’s presentation here

Lilly Ryan’s entertaining and thought-provoking talk drew parallels between our current obsession with privacy-leaking apps and data platforms and the awareness campaign around the detrimental effects of arsenic in the 1800s. Her presentation was a clarion call to resist ‘peak indifference’ and increase privacy awareness and digital literacy.

Deep Learning, not Deep Creepy by Jack Moffitt

You can view the video of Jack’s presentation here

Jack Moffitt is a Principal Research Engineer with Mozilla, and in this presentation he opened by providing an overview of Deep Learning. He then dug a little bit deeper into the dangers of deep learning, specifically the biases that are inherent in current deep learning approaches, and some of the solutions that have been trialled to address them, such as making gender and noun pairs – such as “doctor” and “man” – equidistant – so that “doctor” is equally predictive for “man” and “woman”.

He then covered the key ML projects from Mozilla such as Deep Speech, Common Voice and Deep Proof.

This was a great corollary to the two talks I gave;

Computer Science Unplugged by Professor Tim Bell

You can view Tim’s presentation here

Part of the Open Education Miniconf, Tim‘s presentation covered how to teach computer science in a way that was fun, entertaining and accessible. The key problem that Computer Science Unplugged solves is that teachers are often afraid of CS concepts – and CS Unplugged makes teaching these concepts fun for both learners and teachers.

Go All In! By Bdale Garbee

You can view Bdale’s talk here

Bdale’s talk was a reinforcement of the power of open source collaboration, and the ideals that underpin it, with a call to “bet on” the power of the open source community.

Open source superhumans by Jon Oxer

You can view Jon’s talk here

Jon Oxer’s talk covered the power of open source hardware for assistive technologies, which are often inordinately expensive.

Other conversations

I had a great chat with Kate Stewart from the Linux Foundation and the work she’s doing in the programmatic audit of source code licensing space – her talk on grep-ability of licenses is worth watching – and we covered metrics for communities with CHAOSS, and the tokenisation of Git commits to understand who has committed which code, specifically for unwinding dependencies and copyright.

Christchurch as a location

Christchurch was a wonderful location for linux.conf.au – the climate was perfect – we had a storm or two but it wasn’t 45 C burnination like Perth. The airport was also much bigger than I had expected and the whole area is set up for hospitality and tourism. It won’t be the last time I head to CHC!