The week of July 5, 2015

Rachel Binx: Making sense (and art) of big data

By Selena Larson

By day, Rachel Binx is a data scientist for NASA. By night, she’s finding new and creative ways to transform data into beautiful pieces of art.

At NASA’s Jet Propulsion Laboratory, Binx builds data visualization tools for rovers, satellites, and other space technology. When not working with space robots, she creates projects like Monochōme (personalized map-themed clothing), the physical GIF-generator Gifpop, and map-based jewelry with Meshu.

I had the pleasure of listening to Binx describe her work as both a data scientist and an artist at the Gray Area arts and technology festival in San Francisco earlier this summer. I followed up to better understand how space robots operate, what makes data beautiful, and why people might be edging away from obsessively checking in on apps and social media.

Tell me about your work with NASA.

I’m currently working at the Jet Propulsion Laboratory (JPL). I’m in the human computer interaction research group. We are basically like the designers at JPL. The project I’m doing is about mission operations: the people who are monitoring the spacecraft and making decisions on if the spacecraft is healthy, and if something goes wrong, [figuring out] what went wrong.

Right now the data is stored in these different formats, and it’s hard to compare the different types of data that are coming down off the spacecraft. If something goes wrong, it’s a lot of looking in different places.

What we are doing is trying to make this one Web program that would make it a lot easier, so you could cross-reference everything that is happening and also save your investigations for future use. It’s really hard to explain without explaining how the spacecraft works.

How does the spacecraft work?

There are two types of data that are the main data formats coming down. One is basically numbers that are being throttled to every 30 seconds. An example of this would be the temperature of a component or the voltage on a motor. It’s a stream of numbers, and those get aggregated into channels of data. You can also encode data. If it’s the send data mode, maybe that’s a string of zeros; when it switches to receive data mode, that’s up to one. There are a couple different ways people use those to send data, but they are numbers.

The second is called EVRs [Event Verification Records]. They are text statements of things that are happening. It is logging activity, like if a command sequence is starting or ending, or the diagnostic information about the different components. EVRs are important for knowing what’s going on in the spacecraft. If something goes wrong, that’s where you see it. There are many different categories of EVRs, but the ones you would look for as an operator are Warning Low, Warning High, and Fatal.

“For me, a smaller collection of points or input is more meaningful. When you think about what are the most important places you’ve been in your life, you probably come up with a handful of places.”

Fatal hardly ever happens. Warning High is very, very rare. One of the orbiters I was working on, they were up to 30,000 EVRs per day. It was a ton of information.

A lot of them were known bugs on the spacecraft, like, “Oh, yeah, that happened again.” But your job is to read through all those and find out: “I haven’t seen this one before; I have to go investigate what happened.” What you get from that are text statements that give you an idea of what happened, but you mostly get a timestamp of where something went wrong.

So then once you have that timestamp, you go back to your channel, and you start looking at plots. Alright, I need to look at the temperature plot, oh it spiked, so what was happening in the command sequence? If we switch modes to sending and receiving data, do the plots line up? The tool is a way to quickly find the EVRs that you need to look at and compare them to the different channels you need to pull up.

There are hundreds if not thousands of different channels on the spacecraft, so you as the operator for a particular subsystem probably have 150 that you care about or are under your command. There are probably 10 to 20 of them that you ever need to look at on a daily basis.

How many spacecrafts are you working on at any given time?

My group gets funding on a mission-specific basis. We started this project working with SMAC, which is Soil Moisture Active Passive. It’s a very new [satellite]; it just launched a couple months ago. Now I’m currently working with Cassini out of Saturn and the Curiosity Rover on Mars.

It’s so much fun. There are so many people that work on that one.

What has the experience been like? Did you always think you would do data analysis for spacecrafts? How did that happen?

This opportunity came up recently; they contacted me about it last December.

Before that I was doing freelancing for a couple years and also working on data visualization and building tools—sort of the same family of work but definitely not anything space-related.

Will this technology be used for rovers and satellites, or is it something that can be used for rocket launches?

We are trying to build it so that it will be applicable to any mission.

Between the different missions there are similar data types. EVRs are recent. Cassini doesn’t have them, but it is 20 years old. More recent missions could use the recent data format. It’s more of a tool for when it’s in regular operating mode, so I don’t know if a launch would be a good situation for it. But once up in the skies, our tool would be good.

IMG_0240 (1)

I was looking at your projects like Monochōme, Gifpop, and Meshu that use data visualization in a very artistic and creative way. How do you come up with these ideas? It gives people a tangible way of understanding the data.

I’ve done a fair amount of travel in my life, and it was an important personal goal. I did freelancing for several years, and I tried to have a lifestyle that allowed me to travel fairly often.

For me travel turns me on.

I have done a lot of visual explorations of location data for years, and from an algorithmic or visualization perspective they’re pretty simple. For me, a smaller collection of points or input is more meaningful. When you think about what are the most important places you’ve been in your life, you probably come up with a handful of places.

“It’s the view from hindsight that you sometimes realize the importance of the event.”

I struggle with the quantified self thing that is happening. I tried it at one point, and I learned all this stuff about me, but there’s just so much stuff. We have all this data but have nothing to do with it. I guess i can try to come up with a useful analysis—it just never meant that much. A lot of that stuff [I make], you’re having to type things in; for me that’s much more personal exploration of data.

The quantified self is interesting. I feel like as we get more ways to track ourselves through locations or activities, there is just an overwhelming amount of data. At what point does data begin to lose meaning?

I’m not anti-tracking, but I just don’t do it. I do have services running in these apps; it’s mostly a pedometer thing, but I also never look at it.

I take photos when I travel—that’s some sort of tracking, but it’s also for me just another way of exploring the space that I’m in.

I’m such a luddite, working in technology.

That’s an interesting description of yourself. Do you think that helps your mindset that you only focus on the things that are really interesting or important? What kinds of tools do you use for tracking your personal data if anything?

Honestly I don’t. Even the Moves for step-tracking, if I cross 10,000 steps, I say, “Good job, self,” but I don’t keep track of what I do throughout the day or what days I reached my goal.

When we started Meshu, [my cofounder] tracked everything in Foursquare and I didn’t. To the point where we’re at a restaurant together, he would check-in and I would be surprised he was looking at his phone.

When we launched Meshu, I said people should input places [themselves] and he said: “Why would we do that? I have all my data in Foursquare; why wouldn’t I just use that? It would save me so much time.” We launched, and a lot of people wondered who would ever want [Foursquare]?

We did by city, and then by country. If you take a trip and do check-ins, it translates to the places you went to on the trip. But no one uses that. We had two to three people that bought Foursquare jewelry. Of course everyone else was like, “I want to do the place I went on my honeymoon.”

I’m also one of those people who don’t track things all the time but track things that are truly meaningful to me, like the best restaurant I’ve been to or a trip I’m on.

That’s the thing about meaningful experiences: You often don’t know it while it’s happening to you. If you’re engrossed in a first date with a person you’re absolutely smitten with, you’re not checking in with yourself every three minutes to ensure it’s a meaningful experience.

It’s the view from hindsight that you sometimes realize the importance of the event.

“We have all this data but have nothing to do with it.”

Do you think as we are tracking all this information, the meaning and experiences are distilled a little; with technology, are we losing a part of the experience because we are so focused on capturing it?

I do. I remember back when I was in San Francisco, I would go to all these events, and we would talk about this all the time: Are event photographers a good thing or a bad thing? You can take photos with friends and share them, but are people experiencing the event? We’ve had this conversation about photography, but we’re not having so much with collecting data.

There’s a project that just came out recently. It’s a phone designed to only take calls and does not do data. So if you’re going to the beach and someone needs to get in touch with you for an emergency you have this designed for calls only. I would want a phone that doesn’t do anything.

I try not to get sentimental and say that everything was better before my smartphone because no, I have a markedly better time. While traveling I can use data and pull up maps and view certain things on the Internet and not be at the whim of travel agencies or information desks.

Do you think the idea that it’s OK not to overshare will grow as technology grows? Are we going to share and track only the most meaningful bits?

Some people will, some won’t. I think there’s already pushback on this. Some people have a private account on Twitter with 20 people on it and a professional Twitter account where they can’t post when they’re angry.

I think it’s possible with Facebook; more people are getting wary of it or not checking it, so I feel like engagement is going down. There’s definitely a shift and some people are pushing back to get back some privacy and space.


How important is it for you to work on these passion projects, and what would you tell people that are trying to balance their passion projects with their careers?

All of these passion projects grew out of these little personal doodles. For me they have a two-sided force on them. One is the data side the other is the manufacturing. I love custom objects and the idea of building a system that people can make objects that are meaningful and important in their life.

Like Monochōme, I waited for a manufacturer to come along that was offering it with a high enough quality, low enough price point, fast turn around, and ground shipping. Once that happened, I can print things on clothing and I can make my own styles. There are so many ways of analyzing my life through different data formats.

I have this other project, Making Care of Business. It’s an aggregate of interviews with people who are also running their own projects. I thought it was important for people to connect to other people who were in the same situation. When you’re stressed about your manufacturer messing up something, you can talk to people have done this and can share experiences in a valuable way.

Illustration by Max Fleishman