Eyeo 2018: Conferencing in the age of the Internet

Holding my plate, I spotted a participant eating alone. “Thanks for saving me a seat”, I said as I sat with him. As we engaged in a lively conversation about online groceries, what’s recyclable and, of course, our jobs and the conference, four people sat next to us looking either at their phone or their computer. 

All the talks from Eyeo Festival 2018 will be available online. Why come here and not engage?

This is a tech conference unlike many others as questions of ethics, bias, inclusion and impact are brought up in a large proportion of sessions (one exception that stood out for me was that of David Ha on neural networks).

Setting the tone, Manoush Zomorodi’s keynote shed a light on the ways technology can worsen or enable our worst human traits. Our mobile phones can turn into time-suck and not surprisingly her projects found a segment of the population with an appetite for relief. One could sense agreement in the crowd, but maybe it was guilt. 

Eyeo is a special moment in time. It’s one of the most hyped conferences in several fields and it sells out quickly. Yet, so many of us stand in the middle of this rare mix of people, looking at our phones. Having conversations with people at home. Keeping strangers at a safe distance on Twitter. Watching or interacting mildly with acquaintances on Facebook. Plunging back into work on Slack, as if our brains didn’t remain there afterwards.

I asked one participant what she thought of the previous talk that we had both attended: “I zoned out. I was thinking about work…” she confessed.

As with the joke about the aliens, confused about who’s the master between the dog or the human picking up the poop, an observer could wonder who’s in control: technology or the user.

The irony of course is that that power was the underlying question of so many talks: Who holds the power in tech? How is it yielded? To the benefit of who?

Jane Friedhoff turned the power fantasies on their head in her games, defining the audience not as a benevolent majority gracefully willing to empathize, but as the members of the opressed group that needs catharsis. To redistribute power, Matt Mitchell teaches cryptography to African American populations accustomed to be monitored. Meredith Whittaker conveyed a sense of urgency about the biases in the data we feed to our new overlords of artificial intelligence. 

In the description of their talk, Dynamicland lay it plainly: “Increasingly, working on a computer isolates us more than it connects us”. But it doesn’t have to be like this, so they take the technology out of the computers and into the real world. No more screens: let the humans share the space with technology. And it works. 

The Eyeo organizers have gone to great lengths to make the experience of attending more real and less virtual. The workshops on Monday were very interactive, forcing the participants to get to know one another from the beginning. The delightful personalized button designed by Giorgia Lupi based on our answers to a survey were playful conversation starters. The program of the conference printed on the back of our name tags gave us one less excuse to pull out our phones to check the schedule and then slip into email or social media.

The Eyeo app created a private space for the participants to connect, away from the chaos of Twitter. There was no live-tweeting function, thank goodness, so the speakers were given the full 45 minutes to communicate one on one with each member of their audience before we have our opinions distracted and shaped by the perceptions of others. Both sides deserved this time.

The app was also the place to organize meetups of people with shared interests, to find the show and tell sessions of the participants, or to organize spontaneous dinner plans. All ways to ricochet on the virtual and back into the real world.

Efforts to reach out to strangers were richly rewarded. I was energized by the enthusiasm of Hannah for data visualization, seemingly unaware of her talent as we discussed her recent piece. How can I not be short on time when discussing with Kim, an architect who now teaches board game design? It was exciting to see the possibilities in Jamie’s project of moving to New York City. Did I even scratch the surface with Anni, a coding artist from Mexico who briefly lived in South Korea? How much more could I have learned from Brian’s experience of leading a dataviz team? I could go on.

I walked up to Amanda Cox, Casey Reas, Giorgia Lupi and Stephanie Posavec because they give me inspiration and energy and I wanted to gather some of it live but also to give back just a little by thanking them in person. I’m taking the memory of our interactions home with me.

So many speakers have challenged us to think about how the tech we create distributes power amongst owners and users. We also need to think as users about how much power we yield to tech. As I looked around at people standing side by side, staring at their phones, I asked myself: What will it take?

The upward climb of Max Verstappen in Monaco

It doesn't matter if you kick a ball, ride a horse, dance with skates or race down a hill: all sports tell the same stories. The underdog. The comeback. The dynasty. The near-miss. The feuds. The rivalry. To get interested in any sport, get to know its stories.

One such story is the young prodigy. A new competitor, often young, who comes with tremendous talent. In Formula 1 racing, it is the recent story of Max Verstappen. The son of a former Formula 1 driver (a whole other story), he was the youngest F1 driver in history, at seventeen and a half years old. I pass the mic to Wikipedia:

He is also the youngest driver to lead a lap during a Formula One Grand Prix, youngest driver to set the fastest lap during a Formula One Grand Prix, youngest driver to score points, youngest driver to secure a podium and youngest Formula One Grand Prix winner in history.
— Wikipedia

He's generating a lot of excitement, needless to say. His aggressive style of driving is responsible both for his successes and failures and it will be interesting to see if he manages to adjust.

At the most prestigious grand prix of the season, in Monaco, he made one such mistake and crashed his car during qualifications, meaning that he had to start from the back of the grid. This is very bad news in Monaco where passing is notoriously difficult because it takes place in the middle of a city. 

This is what makes the performance of Max Verstappen so interesting. He started in 20th position and ended in 9th, meaning that he scored points towards championship, which is very important for a team and a career.

The official graph of the race prepared by the Fédération internationale de l'automobile, overseeing F1, does not really show this, even if it shows all the data (as with many bad graphs). Verstappen is the electric blue line, but its progression gets lost in the traffic.

 Lap Chart by FIA

Lap Chart by FIA

It seemed like removing some information irrelevant to the story at hand would help to make it more visible. For instance, we are here interested in a single driver. Also, we are only comparing the start and finish positions.  It seemed like a slopegraph would do the job just well, as it conveys the idea of ranking clearly, on a vertical axis. I used the colours of Verstappen's racing team, Red Bull, because they are bright and attractive and that this graph is about him.

Picture1.png

I worked quite a bit on the title because it was going to give the spin to the graph, to reveal what's interesting about it. The result is a title that both conveys the scale of the challenge and the boldness of Verstappen. Then my subtitle is more descriptive, getting at the point.

I did not write the positions of any other driver. The line for Verstappen is very contrasting, in dynamic red. It passes over several lines and its angle is steeper than any other line. It conveys a very simple message: he started in the back and made it unusually high in the ranks. Also, the horizontal lines of the top 6 racing drivers suggest that it's a race where the starting order tends to determine the finishing order.

It turned out that Cole Nussbaumer chose slopegraphs for her #swdchallenge a few days later, so that's my submission.

The convoluted making of a simple chart

Finding the right concept is my favourite part of doing data visualisation. It depends on understanding the topic, the message and the audience and emerges through discussions with the client. Getting to that aha moment where it finally makes sense is a delight. So this is an example of when it happened not once but twice. Or perhaps only once.

In the present case, as sometimes happen, the client approached me a little late in the process: the data is all collected, analyzed, visualized even and the blog post is written. That's when they realized that their graphs didn't look the part and they called me.

The topic is the investment needs of developing countries to achieve the Sustainable Development Goals. They are huge and they are not met. And it will be very hard to meet them, even in the rosiest scenarios. This is their original graph. The numbers represent trillions of US dollars per year.

Figure 1 original.png

After a quick look at the chart and a discussion with the client, I thought that the best metaphor would be an area, a container, representing the needs that needs to be filled with investments. I suggested a waffle chart and they agreed, so I developed the concept below.

 
Figure 1 waffle chart.png
 

Even though this is my final draft, after much iteration, it didn't quite work as well as I expected. It's not unusual: graphs need to be tested with real data to know if they work. In this case, the idea of a container was more or less lost and it looked more like some compilation of several categories. The labeling was awkward too and I had to use two very different colours to clearly differentiate the labels of "current" and "higher" rate of private sector contribution.

Before even going back to the client, I knew I had to develop another concept. I thought that a simple stacked column chart  would convey better the idea of a container getting filled, but leaving a gap at the end. Here is my first attempt.

 
Figure 1 stacked 1.png
 

I started at the bottom because that's how containers fill up. It created a bit of awkwardness on the right-hand side as the "current" and "higher" rate scenarios are seen above their title of "Potential private sector contribution". Also, I had to use a lot of arrows and lines to explain the different categories, which is not a good sign. Since we had discussed that a sizeable proportion of their visitors arrive on (small) mobile devices, I decided to flip the chart upside-down so that the reader scrolling on a phone would see the headline first and then the current situation, followed by the likely and less likely scenario, ending with the gap. 

 
Figure 1 stacked 2.png
 

It seems like a better ordering, especially as the column description (top) can act as a title and the awkwardness on the right-hand side is reduced. It didn't resolve the multiple explanatory lines, but it was my best option.

So I sent it to the client. And the client didn't like it. They found it harder to understand than their original cascade chart. This sent me back to the thinking board. Ditching my original instinct that a container was the best metaphor, I looked much more closely to the cascade chart to see what was wrong with it. 

Figure 1 original comments.png

It became clearer why it wasn't working.

  1. The direction of the investments changes: “current investments” starts from the top, then ”private sector investments” starts from the bottom. This switch in convention makes the chart harder to understand than it should and the final gap ends up in the middle, rather than at the bottom like the current gap. 
  2. The main data point of the graph is implied and not shown, namely that there will still be an investment gap after the private sector investments, even at the higher rate. 
  3. A gap is a negative number and should be shown below the line, not above, with investments going up, not down.
  4. It was also missing units.

I decided that maybe the concept was right, but the execution was wrong. So I toyed with the cascade graph, trying to change the order of the categories, showing the gaps as empty boxes, tweaking the labels, etc. Unfortunately for this post, I didn't keep all of my iterations, but there were many. It was hard to make it work. One of the challenges is the two scenarios for the private sector contribution. The higher rate ($1,8 trillion) comes on top of the current rate ($0.9 trillion), so it can be an addition (+0.9) or its own number. Here is my proposal.

Figure 1 final.png

This addressed the issues I had seen on the original cascade.

  1. The investment needs go down, the investments go up.
  2. The categories are now at the top, which may work better on mobile.
  3. The gaps are visible, differentiated and labeled.
  4. The unit is included, only once.

This version clicked much better with me. No more awkward labeling and lines to explain how it works. While the columns could have been the same colour, I used different colours to differentiate the categories and join the two scenarios. It feels more intuitive. This iteration also made me realize that we had not yet chosen a title that guides the reader so I suggested one.

I have a principle that a good graph should suggest what should be, hence what is wrong or unusual with the current situation. In this case, the dotted lines of the gaps do this job. They show that the bars should be reaching the top, but don't. This gap should lead the conversation and is now unmistakable, while it was only implied in the original.

This time, the client liked it right away. "looks great", "like the new title", "nice!"

What do you think?

5 tips for visualizing concepts

What does banking look like, in one image? How about evaluation? And training? 

That was the easy part. Now, what about the need to improve? Or to speed up and simplify a process? And finally, innovative financing models?

Designing information requires to think of visuals for concepts that may or may not have a strong image associated with them. These icons will come up in brochures, infographics, presentations and websites. Being able to find just the right image that will anchor the concept in the reader's mind is probably more art than science, so I'm sharing my own thought process here with examples from a recent product.

The Independent Evaluation Office of the Global Environment Facility has recently completed a massive evaluation — in fact, tens of evaluations — that are crucial to the future of the GEF. The reports have been published and the IEO wanted to get the word out so they approached us to develop an infographic (pdf) that would present their 10 conclusions, as a teaser inviting their audience to have a look at the full report.

Click to see the full PDF.


1. Use simple shapes

The fundamental principle of design "keep it simple" applies here. An icon does not have to be complicated and in fact, it should not be. Very simple, common shapes can often do wonder in context, as the eye recognize the reality that they represent, as if guided by gestaltism. 

Almost all of my icons are made of simple, often geometric shapes drawn in Adobe InDesign or Microsoft PowerPoint.

In this example, it's nothing but lines and a circle. The key idea is that the GEF has been a catalyst for the agencies that they fund to develop better safeguards. By establishing this new criteria, the GEF has started a reaction across the agencies, much like the domino effect illustrated. No need to draw actual dominoes with their dot patterns, nor to draw a sphere with complex shadows. It will not convey the idea any more efficiently.

Dominoes.png

Here, three concepts — scaling up, replication and market transformation — are different but have in common the idea of massive growth found in the conclusion. It is sufficient to use the arrows and hint at a line chart with a logarithmic growth, without resorting to all the elements of a chart or perhaps finding what is the true growth pattern of the three concepts.

Scaling up arrows.png

2. Leave out entire ideas 

It's common to have to illustrate a complex idea with several propositions. It might be tempting to find a visual hint for each of them, but that would require your reader to decode all those signs, in a way that is much less efficient than reading the sentence.

One idea is generally sufficient: it's an icon, not a rebus. It may not show everything, but it's enough as an entry point or a memory marker.

The IEO had found a lot of useful knowledge generated and captured by the GEF on its project, but also that it was not always accessible. Simply illustrating the latter idea with a vault is sufficient. No need to illustrate the concept of knowledge at the same time. Having, say, a book in the vault would only force the reader to decipher two abstract concepts and their relationship. It's too much responsibility for a single icon.

In this example, we have taken out all substance to illustrate solely the generic concept of "size matters". In fact, we even evacuated one of the two ideas in the conclusion (in-house expertise). The ladder is too short for the wall, a metaphor for the GEF financing that needs to be sufficient to overcome the market barriers that it addresses.

Size matters.png

3. Find the essence of an idea

A visual is meant to reveal the essence of a concept. This may be one of the most difficult aspect of drawing an icon because it requires a true understanding of the content and message. It's worth it though because relying simply on the words used in the concept may miss the point entirely. Also, finding how to visualize the essence is often the most satisfying moment of icon design.

In this example, the conclusion was a number, which might be tempting to illustrate with a graph, even a pie chart as a single percentage carries over well to such a graph. But the key idea is that the gender analysis seems to be found seldom and at random in the documentation of new projects. The GEF does not appear to have been as systematic as it should at this stage of the project development. So the visual takes this idea to illustrate task completion with a checkmark, while the missing tasks receded with an X on a light green background. This way, success looks haphazard, and not like a process that was suspended, as it would if we had put the seven checkmarks together at the beginning.

This following example is counterintuitive: the idea is limited guidance, yet the icon is a road sign that provides guidance. First, the goal of the icon is to anchor the idea of guidance simply, which the sign does nicely. Also, the road sign is ambiguous: it's a fork in the road without any indication of what direction the traveler should take, which brings us back to the idea of limited guidance.

This is one of the more complex icons in this infographic, but it's also one of my favourites because the concept was so difficult to capture. The IEO found that the GEF captures a lot of information, but not always about what matters most: impact. The GEF can dazzle with numbers about timelines, outputs and the likes, but what about the people they were trying to help? It's this idea that is shown here, with a spotlight on a heap of numbers, next to a person that is out of the spotlight. Alternatively, if the reader only sees someone lighting a heap of numbers and misses the misdirection, it still works.


4. Use abstraction

Icons often illustrate a physical reality, like people, trees and houses. But there is a great wealth of information in abstract shapes. When well-designed, an abstract shape can be as evocative.

Below, the concept was already abstract. Financing models are not just money, they are the structure of a financial transaction. How much interest will the borrower pay? When will the repayment be due? What sort of collateral is accepted? The idea is that the GEF had gone the extra mile to find new ways of lending or granting funds to prompt behaviours that would achieve its environmental objectives. They had been flexible, they had adapted to the circumstances. From this came the idea that their money comes in all shapes and forms. The GEF will make it fit to the particular circumstances of a project. The visual reflects this by placing dollar signs in a combination of odd shapes that fit together.

In the following case, the challenge was that we had already illustrated the concept of scaling up (see upward arrows above), so a new icon was required. After all, we are trying to make the infographic visually engaging. The circles illustrate the idea of progressive growth. By using eccentric circles, we avoid the image of a target that could have formed in the mind of a reader with concentric circles.

Scaling up circles.png

5. Data illustration is ok

Sometimes, it's a number that you need to illustrate. It may be a single number or a small collection. I have read somewhere (Stephen Few perhaps? Or Edward Tufte?) that a pie chart is more data illustration than visualisation, given its simplicity and paucity of information. It's also a very clean graph to show a single percentage because the eye can identify each of the quarters. In the case below, it's immediately clear that the first percentage is a little over three quarters.

Pie charts.png

Icons are not the end all of an infographic. Copy, hierarchy, structure and data visualization are equally and sometimes more important. But the icons serve two main purposes: to lighten the look of a text-heavy document and to anchor an idea. Do not rely on icons to do the work of your text, but use them to attract the eye of the reader and maybe to linger in their minds.

Contact Voilà if you need icons for your infographics and presentations.

When economists get excited about charts

On Twitter, Matt Notowidigbo, Associate Professor of Economics at Northwestern University, has started a very interesting thread.

The mere number of responses (55 in less than a day, as of this writing) is a good example of how much people love charts. They remember, think and talk about them, even if it's not their full-time jobs. Also, none of the charts submitted are visually spectacular and few are even beautiful. Many are in black and white. But their message is what stuck with the audience, Readers are brought in by the insights of the study and the appropriateness of the visualisation method. Here are a few examples.

Let's start with a submission by Notowidigbo, a convincing case of the impact of a technology on the lives of the poor. The change in amplitude is striking enough, and the fact that there are three regions drives the point home.

And here's one that I saw recently on social media and that generated a lot of support in the thread. It's about the unequal impact of parenthood on men and women's earnings and it was published in January 2018.

The following example reinforces that a good graph is not necessarily one that is understood in three seconds, but one that rewards exploration and understanding. To help you a little, you're supposed to be looking at the lag in price changes (the horizontal lines) based on when the news crossed from London to Amsterdam (the vertical lines). I like the creativity of the approach.

Here's a colourful and rich one.  It shows the impact of various weather factors on behaviour, economics, etc. It makes you think about the unexpected impacts that climate change may have on society.

Also, I was surprised to see one that I had the opportunity to edit (not the original) for a report on renewable energy because my client had also found it insightful.

I had retained the whole concept, making only minor changes to the design fo clarity. Now looking at it, I with we could see whether the predictions panned out from 2014 to 2017.

Duck curve FG.png

Verifying predictions brings me to my own contribution to the thread.

What is going on here? Why is it that analysts as a whole exhibit an optimism bias that doesn't seem to abate over time? Is it part of human nature? Or is it something specific to their incentives to overestimate? In any case, the graph makes a convincing showing with simple line charts.

What makes a good graph is first the insight and second the right visualisation. While good aesthetic appearance is part of proper information design — and several of the graphs above would deserve to be improved — it is nearer the end than the beginning of the process.