A Datapoint Walks Into a Bar

image

Hi! That’s me giving a talk at the Chaos Communication Congress (also called CCC; also called 33c3 because it was the 33rd CCC and “CCC” consists out of three letters and hackers love abbreviations) at the 27th of December 2016.

I blink a lot. That’s what people do when they’re nervous. People are nervous when something means a lot to them – and the CCC means a lot to me. I love this festival of hacker culture, of arts and weirdness and wonderfulness. Giving a talk there was a huge honour.

This blog post won’t be a transcript of my talk, but an overview of the questions my talk answered and didn’t answer; and the sources for what I spoke about. My talk can be seen here, on the website of ccc-tv and my slides can be found here.


What I talked about

Mother Teresa said “If I look at the mass I will never act. If I look at the one, I will.” I presented ways that make us act when looking at the mass. First, I talked about our problems to see the individuals when creating data visualisations about huge numbers of people, thanks to our two different modes of thinking:

image

However, we still need to address both modi in fields like news or advocacy: The fast, emotional system to motivate people THAT they should do something (e.g. donating), and the slow, analytical system to help them decide HOW they should do it.

image

I then talked about how data vis can do both, since it’s a tool like language, that can speak to both modi. I tried to make clear that the right use of tools depends on the goals: Eg in politics or science, we might exclusively want to address the analytical side in us.

image

In the second half of the talk, I presented ways to talk to the fast, emotional system with data vis. Personally, I see especially much potential for the data vis scene in the category “Show what the data would mean for your experience”.

image

In the end, I pointed out that it can be harmful to just evoke emotions. We also need to be responsible enough to help people act upon them.


Questions I didn’t answer in my talk

  1. Do these methods I presented actually evoke emotions? Anshul Pandey, Enrico Bertini and Jeremy Boy did some first experiments in this direction, and failed to find any effect of data vis on empathy and donations. I’m looking forward to further research.

  2. Is creating empathy with data vis even necessary? Especially in newsrooms, a graphic is often accompanied by an anecdote about an individual. Maybe that combination works well enough? Maybe we want to keep data vis as as very true- and unbiased-looking overview; as a juxtaposition to the anecdotes?

  3. “How can we add re-calculation of big numbers into personal, more personal, more emotional numbers in newsroom tools?” That’s a question that Mirko Lorenz asked as a response to my talk, and it’s an excellent one. I have no answer. The easiest solution is certainly to work with colours and to show more dots than bars – but maybe there are better ways?

I’m sure there are more questions. If you know of some, write me an email (lisacharlotterost@gmail.com) or find me on Twitter (@lisacrost).


Sources

I got interested in the topic “Data Vis and empathy” thanks to a great, great Data Stories episode about Statistical Numbing. The research of Paul Slovic blew me away, so I listened to it for a second time and skimmed big parts of Paul and Scott Slovic’s book “Number and Nerves”. Their website “The Arithmetic of Compassion” gives a great overview of the key concepts. I also gained a lot of understanding into how people can deal with smaller bits of information better than with bigger bits from the paper “Small Wins” by Karl E. Weick (thanks to Jan for the tip!).

To learn more about empathy, I read and listen to Paul Bloom. E.g. his famous article “Against Empathy” and an episode of Sam Harris’ podcast on which Bloom was a guest for two very enjoyable hours (thanks to Enrico for the tip!).

To learn about the state of mind regarding the idea to evoke emotions/empathy with data vis, I found the following articles the most helpful:



Finally, and like always: Thanks to OpenNews for sending me to the CCC, and thanks to everybody who gave me hints about talks, papers, podcasts and articles about that topic.

Here’s the talk to watch right here, right now:



What I Wrote About In 2016

image

2016 was an intense year. In global news, Brussels, Brexit, Trump and Pokémon GO happened. In superduber-local news (meaning, in news regarding me), I spent nine months in Washington, DC as an OpenNews Fellow for NPR, went to more conferences than ever, gave more conference talks than ever, met amazing people and even talked with them, hiked up a 14k mountain, went to the Southern Hemisphere for the first time (I went in Summer! But there was Winter!) and thought about stuff.

Here I’ll present the questions I pondered in 2016, and my blog posts that resulted out of this pondering:

1 What is a good (data vis) tool?
2 What implications has the concept of Map vs Territory?
3 How does the US election work?
4 How to communicate feelings?
5 What is the purpose of journalism?
6 Do we all need to fight against world hunger with our skills?
7 What’s the point of visualizing self-tracked data?

Let’ start simple:



1 What is a good (data vis) tool?

I don’t have a Google Analytics Tracker installed on my blog; but I’m 68.3% sure that my very practical posts about data vis were the most visited.

After spending a few hours writing a post about tools for choosing colors in data vis (with a deceptive title that suggests that this post is a manual for choosing colors)… image

…I aimed higher. I decided to recreate the same bubble chart with a few tools that were known and unknown to me. “Can’t be that hard”, I thought. “Won’t take that long”, I thought.

I ended up spending two full weeks recording GIFs, doing tutorials, checking Stack Overflow and GitHub Issues. The result were three posts. In the first post, I tried 12 charting libraries: image

In the second post, I tried 12 data vis apps: image

And the third post, I went high-level and compared the tools with each other (half a year later): image

Do I use different tools now than I did before? No. Was that the goal? No. I’m happy to have a better overview of the field, and I was happy to share that overview with all of you. And the best thing about writing the post was having an excuse to have conversations with great tool makers like Christopher Groskopf or Ben Fry.



2 What implications has the concept of Map vs Territory?

The prize for my most-loved-concept 2016 goes to the idea of map and territory. It’s an old idea. But one I really learned to appreciate this year.

I got so excited about it that I wrote an explainer of the concept early this year: image

…and then came back to it in a talk at NACIS in October: image

It’s an idea that made me think about the way I look at the world and how I interact it. And it definitely influenced my “Mission Statement” you’ll encounter if you’ll read on.

A very practical implication for the concept of “the map is not the territory” were this years’ election (map):



3 How does the US election work?

My brain started caring about the US election when everybody else was already tired of it. But then I couldn’t stop thinking about it. Once I understood the difference between the electoral votes and the popular votes, I was hooked. I wrote a general post about why I think that we should (also) map the popular votes: image

After the election, I added a post with the actual election results in popular votes: image

I didn’t get as much feedback as I hoped I would, but that’s ok. I guess everybody was just too busy making the most amazing election maps ever. I can live with that.



4 How to communicate feelings?

Feelings are hard, man. Also blurry. And necessary. This year I bravely marched into the unknown fields of writing absolute not-criticizable posts about my feels.

I started with explaining “How I feel when I have a conversation”– which was extra tricky, because it was about privacy issues and I don’t know a thing about privacy. image

I continued with talking/gif-ing about “How I Feel When I Learn To Code”image

…and ended the year with a love letter to drawing thoughts and using pen and paper. image

All these topics have been covered by smart people long before me. But I found writing these posts incredible helpful. In the best case, these posts built a compelling narrative for me – and let people compare their experiences with mine (as I’ll explain a bit later in the part “What’s the point of visualizing self-tracked data?”). Win-Win!



5 What is the purpose of journalism?

At the beginning of the year I went to the Responsible Data Forum in NYC. Coming from my “data vis for journalism” bubble, the mostly advocacy-driven attendees opened up a new field for me. One exchange with Steve Lambert during some team work resonated especially strong with me. When I asked him “So what’s the goal of our data vis here? What should it accomplish?”, he said “Change, of course.” I was blown away. And jealous: NGOs and activists have very clear goals; journalism lacks them. What IS the goal of journalism, really? To answer that question, I read lots of mission statements of newspapers, had discussions with smart people like Brian Boyer, Gregor Aisch and Alan Smith, asked Twitter and drew diagrams like the following: image

I also gave a talk at my maybe favorite conference this year, Information+; about how I want journalism to focus on context instead of anecdotes: image

But I found a truly satisfying answer to the question why journalism matters in a slightly related question:



6 Do we all need to fight against world hunger with our skills?

A highlight of this year was discovering Polygraph. Its creator, Matt Daniels is obviously very talented. But something felt wrong about it. In his articles, he drew attention to words that rappers use in their lyrics; instead of to the main issues on this planet (as Hans Rosling or Max Roser would do). I couldn’t understand why Matt didn’t use his amazing talents for the greater good. In an email conversation, he replied to my discomfort with the following words:

I’d argue that every project that I’ve done has had impact – it just depends on how you qualify “impact.” Take the rapper vocab project for example: Teachers have used it as a way to engage students in math/literature, grounded in something kids care about. Numerous people have emailed me that they were inspired to pursue coding, design, and data viz via that project. It’s been used to legitimize hip hop, which is widely marginalized in the realm of intellectual culture.

For me, that was mind blowing. He doesn’t just state that “impact” is a very, very blurry concept (lots of concepts are). He also implies that you can’t always foresee what the positive consequences of your work will be – but you can trust people to use your work for their goals. These goals might not be your top priorities (“engage students in math”), but they are THEIR first priority. It means that people can use and interpret you work in different ways. I can see something in a different way than the artist intended, and it can still be extremely valuable for me. And such an artwork could have a huge impact on me.

This epiphany made me more comfortable about my own (not-big-problems-like-hunger-solving) work. And I saw it as a good-enough purpose for journalism: Journalists try to educate people who are in different positions and will use the same information differently.

It also resulted in the first draft of a “Mission Statement” for my work. In my daily blog, I wrote down what kind of things I want to create:

  • Something that resonates with me and makes me feel something, like music.
  • Something that’s beautiful and sophisticated.
  • Something that makes me wonder.
  • Something that makes me feel astonished.
  • Something that shows the map of the world I have.
  • Something that sets my map in context to other maps out there.
  • Something that sets my map in context to the territory out there.
  • Something that shows me that maps are maps and not the territory.
  • Something that shows me the world from a different perspective.
  • Something that challenges what I know about the world.

These goals don’t want to be fully achieved, but always aimed for. Doing this was my main intentions for my two election map posts, and for a little visual experiment that asked: Which Cities Are On Similar Latitudes? image

Turns out, challenging beliefs is one of my favorites hobbies. I’ll try to come up with more ideas that will do so in the future.



7 What’s the point of visualizing self-tracked data?

People who have read my blog exactly a year ago might remember my issues with Self Tracking and its visualisation. “Self-tracking results in irrelevant visualisations which give a fake comfort of meaning, but don’t help neither the tracked person nor the viewer”, I wrote a bit bitter. As somebody who has done some quantified selfing myself, I wanted to see a point in it, but just couldn’t.

Until I met Nicolas Felton at Resonate in Belgrade. Finally I found the perfect guy to ask. His most appealing argument in our conversation was that readers can compare their lives with the life of the self-tracker. That mirror-effect works similar to a fiction piece – but instead of asking yourself “Would I have been brave enough to destroy that spaceship from within?”, you ask “Do I have more or less social contact than this guy? Am I exploring my city less? Do I exercise more?”. I was so happy with that answer that I drew another little diagram: image

And it made me feel better about so many projects I encountered in 2016.



Last, and also least?

I don’t think that every post I wrote is the best post I’ve ever written, but I’m still glad I wrote them. Here are the posts that truly deserve to be mentioned at the very bottom of this Year In Review article:

An explanation of Information Theory and its definition of “information”, applied to American vs. German small talk: image

“How I organize my files”. A friend asked me to write about that. I’m looking forward to look back at it in 20 years: image

At the end of October, I gave an experimental session I titled “Let’s Ask All Our Embarrassing Data Questions”. A few days later, I wrote about what went well and what didn’t: image

———

And that was it! 2016. Thanks to all the amazing people I met at all the conferences I’ve visited (Responsible Data Forum, Tapestry, NICAR, International Journalism Festival, Resonate, Data Journalism Unconference, Information+, Netzwerk Recherche, SRCCON, Hacks/Hackers Media Party Buenos Aires, NACIS, MozFest). Thanks to everyone who gave me an opportunity to give a talk or to talk through ideas. Thanks to everyone who retweeted my work and/or told me what they got out of it; it made my day every single time. Thanks to everyone who pushed me and/or the data vis field forward in 2016. And thanks to everyone who read an article or two on this blog.

Let’s do everything even better in 2017! If you want me to cover anything specific on this blog, write me an email (lisacharlotterost@gmail.com) or find me on Twitter (@lisacrost).



Drawing Thoughts – A Love Letter

image First sketch for my election graphics.

I love drawing my ideas. Using pen and paper is a very big part of my work as a data vis designer. I doodle, I sketch, I make notes. It always comes very natural to me. Intuitively, I grasp for a close-by paper and pen multiple times a day. It’s something I don’t think rationally about.

That’s what I want to do here: To think about it. Why does drawing thoughts feel so good? I found some reasons, and I’d love to hear yours. This is my love letter for using pen and paper:


It structures my thoughts. Paper is not a passive butler who just saves my thoughts as they are. What paper wants from me is a structured version – a version that CAN be drawn down in the first place. I need to bring my blurry, tangled ideas into shapes and forms that are separate from each other. I need to categorize, label, organize. This process feeds back: It changes my thoughts, which then result in a new drawing.

image

It doesn’t distract the creator. White, unruled paper invites me to endless possibilities. The unlimited potential of the sheet of paper in front of me is magic. Everything is an option. When I’m sketching out a new project, the pen lets me ignore technical difficulties. What separates my blank sheet of paper from ingenuity is not technology, but skills and smarts. I don’t need to open StackOverflow to know how to draw. There is no technological constraint to overcome; no bug to fix; no tutorial to watch to be able to draw with a pen. I don’t get distracted from the tool and the technical problems I have with it. I focus on the thoughts I want to bring to paper. I ask “What is think-able?” before asking “What is do-able?”

It doesn’t distract the viewer. Everyone knows how to use a pen, so everybody can relate. Nobody asks “How did you do this?” when I show them a drawing, because they know the answer: “Practice, and my imagination.” When someone asks me the same about some code I wrote, the answer is complex: “There is this new library out there, and that lets you do x while still be able to do y, and then I imported it into z where I did this little hack which let’s you…” Looking with somebody at code ends up in conversations about technology. Looking at drawings ends up in conversations about the content.

image

It gives me an overview. Boy, I looooove overviews. Overviews are the best. Overviews let me compare, prioritize and decide between elements. To see all my thoughts on one piece of paper at once gives me a map. This map can show me empty spots that still need to be thought about. This map can show me my goal, and it can show me how to get there.

It lets me stay high-level. Designers need to work in two modi: The high-level mode asks: “How do I organize the content?” The low-level mode asks: “How do I make it look good?”. The high-level mode requires common sense, thinking about user experience, storytelling and structuring skills (or, as Juani Ruiz Echazú puts it, “setting up a skeleton”). The low-level mode requires a defined taste for everything visual and insights into how it gets perceived: colors, shapes, images, layout, type. Most tools put me immediately in the low-level mode: Adobe Illustrator, Photoshop, CSS. Pen and paper is one of the rare tools that forces me to think high-level. I can’t decide for a typeface here; everybody will understand that my hand-writing is a placeholder for that still-to-be-made decision. Everybody will understand that color will be added. Lines will be straightened. Shapes will be resized according to the actual data. And when I say “everybody”, I include myself. Pen and paper don’t ask: How will elements look like? Pen and paper ask: What elements will be where?



————

I often use drawing to lay out blog posts or presentations I will give. Here are some sketches and links to the final posts:

image Draft for my talk at Information+: Less News, More Context. I draw slides next to each other when I want to think through alternatives for the same slide.

image Draft for my How I Feel When I Have A Conversation post. It was originally just about Facebook. I chose a more general view for the final blog post.

image Draft for my How I Feel When I Learn To Code-GIF. I did some more searching sketches beforehand, so this draft is already very similar to the final piece.


————

I haven’t read Dear Data, Drawing a Hypothesis or Infographics Designers’ Sketchbooks yet, but I probably should. Have a look at them if you’re interested in the topic.



The US Election 2016 in Popular Votes

image

Let’s map the unmapped voters. Let’s not show the winner of each state, but the percentage of people and voters that voted for a specific candidate: The popular vote. I talked and wrote about the reasons to do so. The following graphics and maps are dedicated to the popular vote of the US Election 2016 (you know, the one last week. The one where Trump won).

Let’s start with an explanation of the electoral college:

image

If we know the percentage of people who voted for the winning candidate in each state, we can look at the votes that we normally don’t find on a map:

image

This map prove once again that it was a close race. There were lots of blue votes in red states – and lots of red votes in blue states.

If we go one step further, we can take non-voters into account: People who are not allowed to vote (because they are under 18 years old, in prison etc.) and people who didn’t want to vote. And this time, we don’t look at the electoral college votes, but at the population. The states in the following graphic are sorted by the share of the population that voted for Trump. We can see that a greater share of people voted for Trump in red states like Texas or Utah than in blue states like Delaware or Oregon:

image

In fact, thanks to the immense population in blue states, there were more Trump voters in Illinois, New York and California than in 12 of the most red states:

image

Do you have comments, better ways to map the popular vote or found mistakes? Let me know on Twitter or write to lisacharlotterost@gmail.com. Thank you!



Let's Ask All Our Embarrassing Data Questions/Review

image

What does scraping actually mean? Why would I need databases and what the heck is MongoDB? What’s the difference between SQL and SQLite? And can I truncate my y-axis or will the data vis community kill me? Even if we’ve worked with data for years, we might have simple questions like that. And now it’s too late to ask. Waaaay too late. Nobody can ever know that you still don’t know what Pandas is. Or why people use PostgreSQL.

Last weekend, hundreds of people came together to celebrate the Mozilla Festival in London, “the world’s leading festival for the open internet movement.” My fellow OpenNews fellows and I attended – that’s me and two of them in Hyde Park:

image

I decided to host a session that would tackle all the knowledge gaps in the sphere of data. The idea: During one hour, all attendees anonymously submit data questions (which we gathered in an Etherpad). Then everybody reads one question and tries to answer it. If she or he needs help, the whole group helps out.

Why I wanted such a session to happen? Because self-taught as many of us are, we have gaps in our knowledge and skills that other people can fill. And we all can fill other people’s knowledge gaps.

The whole session was one big experiment. I’ve never hosted such a question round before. Here is a write-up of what went well, and what could be improved the next time. Let’s start with what didn’t go well, and how to improve it:

image


What didn’t go well

My aim was to close knowledge gaps. That definitely happened. But it happened far more for beginners than for advanced people. During the session, we answered questions that were on hugely different levels. Questions like “What does scraping mean?” as well as “How can I more reliably bin data for choropleth maps?”. The advanced people were the ones answering the basic questions to beginners AND the advanced questions to other advanced people. It might be a good idea to divide beginners and advanced people in different groups.

If I was hosting such a session again, I’d try to get more un-googlable questions. Google got you covered when you ask “What is an API?”, but not if you want to know if “your audience really interacts with your interactive dataviz”. Answering the latter question challenges the advanced people more than answering the former one. Also, un-googlable questions require opinions more than explanations. That’s good, because explaining things is hard. People think for hours how to explain an API best – it’s unlikely that my session attendees and I can come up with perfect explanations in seconds. The result were sometimes confusing answers (which could demotivate to learn more about it), and not enough time to go in-depth and ask clarifying questions.

Third improvement: Better defining the question space. “Data” is an extremely big field: It includes Business Intelligence and stats, maps and interactivity. I appreciate about the field of data visualization that every new project challenges us to learn about new tools, methods, workflows. And most answers made me learn something relevant. But a dashboard person won’t gain anything out of hearing an answer to the question: “Are there any good/clear resources for better understanding of “overpass turbo” queries in OSM?” It would be helpful to at least define the industry for such a session: Will it be about data questions in journalism? Or in science, or in business?

image


What went well

I was positively surprised especially about how open people were with their questions and with what they didn’t know. I tried to create a safe space by
a) talking about how we are all students and teachers at the same time and that nobody can know anything
b) doing a fun short exercise in the beginning where people needed to interact with each other
c) splitting the attendees in two groups to create more intimacy
d) letting people submit questions anonymously
e) stressing it a lot when I didn’t know something.

And it was great to see how quickly people came up with questions. I prepared some questions beforehand, but we didn’t even get to them.

I was also happy with the engagement of the attendees. I hope that everybody got something out of it: The beginners got answers, and the advanced people got the feeling of helping somebody; of explaining things that are relevant for at least one person in the room.


In general, it was definitely an experiment in which I’ve learned a lot. The concept needs improvement, but I look forward to trying it out the next time.

After splitting the attendees into two groups, the great Simon Jockers lead the other group and gave valuable feedback afterwards. Thanks, Simon! And the photo and GIF are by the great Drew Wilson. Thanks, Drew! All questions can be found here.