A Dangerous Hypothesis, A Seismic Sonification, A Collection of Oscillating Cycles
Hello. I’m Duncan and this is “Hello from Duncan”, a newsletter I send every ten days where I talk about the work I’m doing and the things I’ve found interesting. Want more? Here’s more. I love to hear from readers, so hit reply and tell me what you like about it and what you don’t like it about it.
Subscribers who’ve been around for a while will remember that back in May I published a narrative data visualization made with my pen plotter. Titled “Boom & Bust”, it ran through the history of global oil prices and consumption over the last 50 years. The editor-in-chief of the Data Visualization Society’s publication, Nightingale, said it was “like watching Hans Rosling at his first TED talk except, with a digital pen plotter”.
Well, now the Washington Post is jumping on the bandwagon. They streamed a ten-hour video of a hanging pen plotter drawing a map of the 2020 US election as the results came in.
Unfortunately it took rather longer than ten hours to get a result. And they had some major alignment issues with the different coloured pens, which is a problem that pen plotter artists know all too well. But it’s fun to see the result anyway - and I love the idea of streaming the creation of a physical data visualization live as results come in.
You might be familiar with the famous experiment on selective attention where you need to watch a clip of students passing a basketball to each other, and count the number of passes made by players wearing white clothes. If you’ve not tried it yourself, then go do that before you read on. It only takes about a minute.
Done? Okay. Well, there was an interesting scientific paper published recently as part of a discussion of “night science” - the behind-the-scenes stuff that the public rarely hears about. The paper extends the selective attention experiment into the realm of data analysis. A team of researchers asked students to analyse a dataset containing the step counts and body-mass index data of about 1800 people, men and women. The students were split into two groups - one was asked to consider specific hypotheses - that there is a statistically significant difference between men and women, that there is a negative correlation between the number of steps and the BMI for women, and that this correlation is positive for men. The other group was simply asked “What do you conclude from the dataset?”
The most notable thing to be found in the dataset is that if you simply plotted the step counts on the x axis of a scatterplot, and the BMI on the y axis, then you’d see an image of a gorilla waving at you. But here’s the interesting thing: students in the first group - with a specific hypothesis to test - were five times less likely to spot the gorilla than those in the second group.
The takeaway, of course, is that when embarking on an investigation, it’s best to start with no preconceptions. But unfortunately that’s rarely how data science is done (and in fact, the entire scientific method focuses on hypothesis testing). The research paper is actually very readable, and I recommend going through it if this is a subject that interests you. Here’s a particularly good bit:
We typically acquire data with the expressed goal of testing a specific hypothesis. But as we have seen with the gorilla experiment, we are likely to miss other interesting phenomena as soon as we are in a mental mode of hypothesis testing. To account for this, we must consciously adopt a different mindset—one of exploration, where we look at the data from as many angles as possible. In this mode, we take on a sort of playfulness with the data, comparing everything to everything else. We become explorers, building a map of the data as we start out in one direction, switching directions at crossroads and stumbling into unanticipated regions.
More on night science here.
It’s a bit tricky to write a pitch for a talk that’s more than two months away about a project that isn’t quite finished yet, so I had to keep it fairly vague. But I think the approach I took - focusing on what people can learn from the year of work that Miriam and I have put into the project - should hopefully be compelling.
I’ve also started outlining an article for Nightingale on the same subject, which involved a lot of research into the history of sonification. Some of the earliest sonifications were created by earthquake researchers in California, and there was a very strong connection with avant garde musicians in the same area.
My favourite little bit from the research was a liner note on an LP that was released in 1953 which contained sonifications of earthquake data:
It is understood as a condition of sale that Cook Laboratories, Inc., will in no way be responsible for damage this phonograph record may cause to equipment directly or indirectly. For users with wide-range woofers this disclaimer shall be construed to include neighbors as well, dishware and pottery.
Speaking of outlining, I recently discovered Dynalist - a cross-platform outlining app made by the same people as Obsidian. It’s really handy for getting your thoughts down quickly, then dragging to re-arrange them into an order that makes narrative sense. It’s also free for unlimited use (there’s a pro tier with more features). Give it a try!
I’m part of the organizing committee for Outlier, the aforementioned first conference of the Data Visualization Society (I’m recusing myself from any speaker selection discussions because I’m applying to speak, in case you’te wondering), and one of the things we’ve done that I’m most proud of is selling tickets at wildly different price levels.
The standard ticket price is $149. That’s what we think the conference is worth. But there’s also an “income conscious” ticket available - at just $49 - which offers the exact same access but at a lower cost.
Conversely, there’s also a “Support Outlier” level of ticket - at $299 - which we’re aiming at people who’ve had a great year, or folks whose companies are paying for their ticket. Again, it comes with no extra perks except a fuzzy feeling of goodwill.
The payment options are not limited to those three levels. At any ticket level you can add an extra donation, and we’re also giving out free tickets - no questions asked - to anyone who emails us and asks for one.
Why do all this? Because the Data Visualization Society is not a corporation trying to maximise profit. We’re a group of volunteers who want the field to be as broad and diverse as possible - and offering access at a lower (and higher!) cost is the easiest way to achieve that while still meeting our budget.
If you want to join us then you can get tickets here at whatever price point suits you. If your company has an interest in dataviz then consider sponsoring Outlier. And if you’d like to speak (in any language!) then here’s the application form.
I’ve been invited to give a workshop at an upcoming dataviz conference (more details when it’s formally announced). The workshop will be about creating data art, aimed at dataviz folks who are beginner-level coders.
I’ve not given a virtual workshop before, so I’ve been overpreparing a little - figuring out exactly what I want people to learn and how I’m going to teach it.
For that I’ll be tapping into my experiences of taking a bunch of workshops this year. But I’d also like to ask you. If you’ve taken a workshop online this year (especially if it was a code-oriented one) then what did you like about it and what didn’t you like? Hit reply and let me know. Any recommendations would be much appreciated.
After spending most of this year getting into making generative art, it was only natural that I be drawn into making generative music too.
In some ways, generative music is simpler than generative art. You have much less going on at the same time. In other ways it’s more complex though - there are loads of parameters in sound to worry about. Luckily, I’ve spent a big chunk of this year looking at sound parameters as part of my work on Loud Numbers, so I’ve already got a lot of the hard work done there.
I spent a little time over the weekend composing a first proof-of-concept style piece in Sonic Pi that does a bit of what I want it to do. It’s very simple - just three voices that play in oscillating cycles for as long as you want them to. Here’s a recording of ten minutes of that sequence, which I’ve titled “Early Winter”
My goal is to create ambient music that you spend most of your time ignoring, but then it occasionally drifts into your consciousness and you realise how nice it sounds. I love having that kind of music on while I’m working. So it would be really nice to be able to make my own.
Finally, I’ll leave you with a Christmas gift idea for any nerds in your life.
Nervous System is a generative design studio based in the United States that creates beautiful things out of computer programs that imitate natural phenomena. You can even design your own products on the web, which will then be fabricated and shipped to you.
Have a poke around in their online shop - you’ll find lots to love.