Build and Deploy Episode 2: Creepy and Cute IoT with Christine Sunu
Time to read: 41 minutes
Build and Deploy with Liz Moy is a curiosity-driven podcast that explores the lives of people who create things with code. In this episode we dive into a conversation with a creative who makes things with code, hardware, soldering irons, and even, on occasion, faux fur.
Christine Sunu is a maker who designs and builds technology with emotive, human-centered interfaces. (She also happens to be IoT developer engagement manager at Twilio). We talked about Sourd.io, an IoT Sourdough Starter Monitor, which monitors your bread’s temperature, humidity, and rise, so you know how it’s growing and when it needs to be fed.
Her project was featured on The Verge and Mashable, uses Twilio Narrowband and can be easily attached to the top of a washable canning jar. We also talked about ways people can get started with their own IoT projects and you can get some more ideas from her latest blog post.
Livecoder Dan Gorelick made all the music you hear in this episode, using TidalCycles. It’s all open source and available on Github. You may have also seen him perform on SIGNAL TV with a guest who appears on the next episode of the podcast.
- Christine’s first DIY “video game” [01:25]
- Writing image processing algorithms for levitating paramecium [08:29]
- Choosing to go to Particle and explore hardware [11:27]
- “What does technology do to us as people?” [13:37]
- How she knows when a project is ready to share [18:16]
- What is it that people who make technology do to make us feel attached? [24:15]
- “Yeast are just like tiny livestock” [30:09]
- What happened when Christine set up the sourdough tracker [34:02]
- Trying different MQTT brokers [35:45]
- Working with Narrowband and when it might be a good option for your project [40:00]
- Christine’s 3D printer of choice [45:42]
- How to get started with hardware [48:45]
Liz Moy: Welcome to the second episode of Build and Deploy with Liz Moy, where we talk to curious creative coders about the projects that they're working on. When I landed in San Francisco for my Twilio orientation, one of the first people I met was Christine Sunu. She had previously run the developer community at Particle, one of the pioneers and leaders in the Internet of things space, and she worked on weird open source robots at BuzzFeed. Christine was bringing her IoT brilliance and community building experience to Twilio. Intelligent, curious and kind, I was so drawn to learning from her. It's always scary stepping into a room full of folks you don't know, but Christine immediately made me feel at home.
I wanted to bring her on the show to ask her about a very quarantine relevant project that she made, which uses hardware to help you hack your sourdough starter. And I have to say, I was a little shocked when I found a TED talk of warm, empathetic Christine killing one of her cute furry robotic creations on stage.
Christine Sunu [01:01]: This little guy, what we feel when we see him, that's some deep programming. I'm really uncomfortable watching him be in pain. And I think it's a good question to ask, whether we're ever going to be able to see something that appears alive and feel nothing, to watch it struggle and feel nothing, to watch it die and feel nothing. Watch the full TED talk here.
Liz Moy [01:25]: The thing is, Christine wants people to feel living emotions about non-alive objects, whether it's an app or a hardware product, or robot pet. I'm thrilled to share this conversation with her today about that element of her work as well as about the famous breadbot.
Christine, you're a hero of mine, for real. I seriously am so thrilled to speak with you today. What is the first thing that you can ever remember building?
Christine Sunu [01:25]: So, when I was very young, I went to an arcade, and there were, of course, you had these side scrolling games, like the Ninja Turtles game and the X-Men game and you would even have these side scrolls where you push the joystick one way and your character would move and you could push it up to have the character jump. And in my, I think, four-year-old brain, I just thought, "Okay, that's awesome." But I don't think that we have anything that would allow me to construct that at home. What we did have was cardboard, string, and tape. And so, I attempted to create a complex Marionette system controlled by a stick, like a joystick like interface, using cardboard, string and tape and it was such a disaster.
I remember there being a cardboard box that was the frame, that was supposed to be like when you have an arcade machine. And I remember creating and cutting out these little characters that I then put, I tied like pieces of string or yarn to them and the string was very small, and it was very tangled. And I had attempted to attach that with really no plan to a stick that went into the frame. And it got to be a problem immediately because all of the strings tangled and I remember sitting at the kitchen table watching my parents, both of them at once bent over this Gordian knot tangled string, at the ends of which were hastily cut figures that I had made in my slapdash attempt to recreate an arcade machine from cardboard. And as they were working so hard on untangling it, I remember thinking to myself, "This is not worth it." I think that's one of my earliest memories of trying to build something, that actually is trying to build something that was also in tech.
Liz Moy [03:45]: I do think that speaks to really your brilliance at a young age, because you didn't really understand how it was happening, but you were determined to create it somehow and the analog way in which you went about it was basically the way that your mind made sense of how to make something and I definitely wonder - I'm very curious about if that's kind of informed how you've arrived at the work that you do now - which really bridges those interactions between what we experience as humans and how we interact with computers and trying to kind of bridge that gap with empathy and emotion. And would you mind talking a little bit about that part of your work?
Christine Sunu [04:28]: Yeah. I am very interested in human elements. I'm interested in how the tools that we create and the things we interact with digitally affect us as social and emotional creatures. I think that that's something that a lot of people think about now, much more than they did before, but it's something that we still lack typically when we begin designing interfaces, especially hardware, I think, which is a shame because a lot of times hardware is the space that has the most engagement, I think with some of our emotional centers. That digital still feels quite intangible, but a lot of the interactivity that we get through hardware and the physical objects that sit in our space have a stronger emotional resonance with us.
That was something I started thinking about a lot when I started doing hardware things, but I guess, you're definitely right, that it's something that gets informed from what my earliest experiences were in building. When again I think about identity and like what it is that you do, if what you do is building and what you do is not specifically code development or engineering. When I was a kid, I did a lot of writing and the reason was not because I liked to write or I liked to imagine, although those things were true. The reason was that it was the fastest way to make a thing that felt complete. That still remains. My toolset was different, but it's just been that all of my grounding things were in soft craft and things which in 1992, you would say, is appropriate for a little girl to do, right? It's like sewing, sculpture, painting, drawing, music, art, writing. These are all things where you were lauded for doing them if you were like a small like five-year-old girl in like the early '90s.
And so, it's that toolset that informed my thinking like the toolset that is associated much more strongly with human emotion and expression that informed my palate when I started working with technology. My first experiences writing code were actually, I think I was about 13 or 14. And I was really interested in how to make video games again. I mean, I actually don't know sometimes when I think about this why I don't just build video games. But I was highly curious about how I could make these game-like interfaces. And I ended up with a book on C++ and it was one of those like, learn C++ in 24 hours. And my dad wanted to give me a goal. He thought like, "Oh, I can give her a goal of something to do that would engage her, so she really feels like she can build the next step." And he said to me, "Why don't you try to build an algorithm that will guess what number that I'm thinking of?" And so it would be like asking different questions and then based on the answers if you get to the right number."
And me as somebody who while I can math or not, like my interest and my center, I don't center around math. It wasn't at that time something that I thought of as a deep part of myself and it wasn't something that drove interest for me. So of course, I thought about it for a minute and then I wrote him a program. And the program said, "What number are you thinking of?" And then it asked three random questions and then it spat that number back out. So, I was not... and then pretty soon after that, I lost interest in writing code. And I pursued a bunch of other stuff, a bunch of other ways that you could create and I didn't really work with code again until I was in college.
Christine Sunu [08:29]: And I was working in a lab where we were levitating paramecium in gigantic magnetic fields because when you levitate them using these big fields, they lose their sense of gravity. There's this interesting thing where paramecium is a single-celled organism and yet it somehow knows which way is up despite all of the currents around it. That doesn't even make sense. How does that happen? So, the question was about like, somehow they know gravity, if you change gravity, what occurs? There's a couple of options for this. One of these is hypothetically you could take it to space, which I think they also did, if I remember it correctly, but the option for staying on Earth was levitating water droplets in big magnetic fields. But you had to see what direction the paramecium were going and kind of how they acted in these fields.
And so, I ended up having to write image processing algorithms. And I think, what was it we're doing? It was image processing algorithms using MATLAB. And so, that was really, for me the time when it clicked, when I felt like I was using code as a tool to do a thing that I needed and that it was achievable. And that's also driven a lot of things for me when I build like, if it's something that I need to do, I'll learn whatever I need to do it. And if it's optional or if somebody else has asked me to do it, it's harder for me to get motivated, but it was a really cool time.
Then after that, I continued to do code in the context of research. I worked in a neurogenetics lab where we were doing next generation sequencing, which at that time really was actually next generation and not previous generation. That was a huge amount of data and we had to figure out how to take these like countless reads of like DNA. This information that was coming out of the sequencer and assemble it and then also organize it, pare down which mutations were actually of interest and then pass those off to the researchers in a format that they would be okay with looking at it.
And so then, I was using code for that purpose and the whole time as I'm writing in, I think that was a lot of shell scripting, a lot of SQL, a lot of Python, because Python and strings, but I was getting really curious about what is interesting with hardware and what we could do with hardware, because I was at the same time, actually setting up the wet lab aspects of this. So, a lot of the machines that I worked with were, well, they were machines. They were running their own code. They were interfacing with our computer systems and they were giving us this data. And like that was... yeah, the accessibility to the world that I felt through writing code that fit into the system made me feel really empowered to start thinking about what I could do in hardware.
Christine Sunu [11:27]: And then I went to medical school instead of pursuing that. And so, I was in med school for a while. I missed building things. I left. And when I left, when I decided I was going to leave, I saw that there was an opening at a company where their whole goal, as stated on that particular job's website at that time, was to make it accessible, to make it easy for individuals who didn't necessarily feel well-versed in hardware to do hardware and to do hardware in the context of the Internet of things. So that was Particle. I worked at Particle for a while and that was where I really learned to do a lot in hardware, and where I started getting really active building things and building different like weird internet hacks that were often whimsical, sometimes were a demo of what we needed. I typically went into near parody of the interactions we have on the internet, things like somebody tweets at you and so now, this thing is going to happen, like things that are very typical now.
Yeah, so that was where I started doing hardware and my considerations coming into hardware were both from the context of me having started building things in what people would call soft craft, where people would call like more emotional like spaces, more art spaces. And from me having just come out of medicine, where while that side of science is not necessarily associated with or lauded for its human interaction, the training that you receive during the first couple years in medical school is very focused on people. And it's very focused on "how do you interact with patients", "put them at ease?" "What is the emotional state of a person dealing with disease at every stage of that illness?" It's a lot of focus on people. And so that was my context coming into the world of technology, and coming into the world during a time when IoT was the favorite buzzword of every VC.
Christine Sunu [13:37]: So, all the questions that I asked once I felt empowered enough to ask questions and build things under that context, of course, it took like, a year or so of adjustment before I knew like I was building anything interesting that once I started asking myself questions. All the questions were about, "what does technology do to us as people? What does technology do to us emotionally? What are ways that you can underscore different feelings you get from technology or better contextualized feelings that you have as you engage with technology? What is the meaning of a vibrating phone or a vibrating stuffed animal or vibrating anything? What is the meaning of a dashboard setup versus a button setup or a dial versus a notification?". They have a lot of associations and meanings that go with different digital interfaces, different physical interfaces. And once I started asking these questions and playing with this, with different hardware aspects, just expanded my vocabulary a lot in terms of the building stuff of it all.
So, I ended up in the open lab for Journalism Technology in the Arts, which was a program that BuzzFeed was running at that time. I was sponsored by GE to be a fellow there, which basically meant that GE paid me to hang out at BuzzFeed and build weird internet connected stuff. That was really fun. It also gave me time to really dive deep and explore some of these concepts about when the physical interface of hardware changes, what does that mean for us as people? And what does it mean if you have... what does like a color mean in hardware? What does the sensibility of a soft object mean in comparison to what we see on our phones? If I tried to mix cues from what you expect, if I give you something that's simultaneously cute and unnerving, how do people react to that? If I go really like down the rabbit hole into cute, how do people react to that? I spent a lot of time thinking about that.
And that gave rise to some of the things that I am more seen for on the internet like the Fur Worm, which was an experiment, like what are the minimums for appearing as though something is alive? The starfish cat, which is about mixing cues between something that's really adorable and something that could be horrifying, and how do people react to that? After I played around for a while, I felt like I really needed to put my effort into thinking about what would be therapeutic for people instead of just kind of running different experiments about what would be weird or funny.
And so, I started developing these haptic plushies like these different haptic systems that would just make you feel better or relieve stress while you're holding them. The questions were all around "how do you make something that takes the vocabulary we have from everyday life and the living of us being creatures who need other creatures, the natural state of humans? And take those, take the vocabulary from that and put it into technology to help us feel better about stress in our lives, about things that are going on? How can that help us sleep? How can that help motivate us?" Questions like that. Still working on those sorts of things today.
Liz Moy [17:09]: I was hoping that your Stanley little haptic robot that purrs and breathes would be available during quarantine since I was alone for all of quarantine, but I just had to get a cat instead, which he's been a pretty good substitute. But I did come across a couple of things that you just mentioned when I was looking stuff up that you'd built. I saw the video where you... well, I don't know if I should spoil it for people in case they want to watch.
There's a video of a TED talk that Christine gives where she has the furry worm and something very tragic occurs and you know the thing isn't alive, but you feel like it is and it's so jarring because you cognitively know that it's not alive, but you really emotively feel like it is. What is it like to kind of come across those sort of findings and experiences, like what does that tell you with the things that you're making?
Christine Sunu [18:16]: I know that something that I make is ready when it gives me a feeling, because if I made it and they wrote all the code and I decided the velocity that it was going to move and the acceleration profile that it was going to use and I'm still emotionally affected when I see it move or struggle or reach or purr, then I know that it works. If it works for me, the creator then it's going to work for you, someone who's never touched it before. The Fur Worm was an interesting experience. I don't think I've ever told the whole story behind how I came up with that and what really happened there.
I'd been invited to work on this talk. The first couple of things that I wrote, I just didn't like. I just couldn't, I couldn't figure out how to tell a story. This was also I think the first time that I was really doing a talk in front of a larger, more professional techie feeling audience where I was talking primarily about myself and it was going to be everybody looking at me, and that made me highly uncomfortable. I was really nervous about that, that thought. And I talked to the guy I was dating at that time, who's now my husband, and he is a technologist who also thinks about these kinds of ideas and has played with this stuff for a long time. And he started telling me about this puppet and he had made a really simple puppet that was three servos. It was remotely controlled and he would wiggle the controls or like hand it to somebody else to puppet it and it would be this very convincing movement and people would get attached to this like basically, three servos in a fuzzy sock.
And I thought about, "Okay, so what if you were to automate this?" And then we were talking about that and he said, "I wanted to incorporate that one day into an exhibit, where you would come in, and you would pet it, and you would interact with it, and then I would kill it in front of you." He's going to be mad that I told you this story and framed him in that tone. But his thing that he had talked about was, he said, "And then, I'm going to stomp on it and then it would be destroyed and I just want people to, like I want to see how people react to that." And I was like, "That seems like too much."
Liz Moy [20:50]: You're like, "Babe, tone it down."
Christine Sunu [20:52]: Yeah. I was like, "That seems like kind of a lot." And he was like, "I mean, you should do that on stage." And I'm like, "No, man, I can't do that on stage. You want me to show up there like with an axe, and just like cut the head off of this, like that sounds terrible. People are going to be so mad." But the more I thought about it as you often do with art, the more I thought to myself, like "This is something where, like what is the message around that concept? What is the message around taking something simple that people have gotten attached to in technology and destroying?" And it was actually a message that I had been thinking about a lot of that time. This was during the rise of the Amazon Echo as well, so a lot of people are going to attach their Echoes, buying Echoes for their grandma like that sort of thing. And I was reading a lot about how we're affected emotionally by our tech devices.
I don't like to shock people, especially at that time I was very shy, and I didn't... I was very recalcitrant to shock people. But I also deeply felt like it was necessary in the kind of thing that he had described. This idea that you would take an electronic thing and you would destroy it in front of someone. That felt like the level of significance that would make people remember how much the interfaces are affecting them, and how important it was for us to understand that and to take caution and protections against it. This was during my year of research at BuzzFeed, and I was just walking around and like thinking, "we're not noticing, we're not really". This is just getting folded into everyday life, like how emotionally affected we are, by interface. And unlike medicine, there is no Hippocratic oath.
Where none of us design interfaces thinking like "This is going to affect someone emotionally in the future," in a way that might make it impossible for them to get off of this service or in a way that might make it so that they'll pay more to have their Roomba repaired than they will to get it replaced. This is a real thing that happens. People will pay more to repair their Roomba, even though it would be cheaper to replace it because they feel attached. And this is an inherent thing that happens with physical interfaces. And in a time when increasingly people were talking about personal robots, it felt important to contribute to that dialogue in a way that people would hear.
So I decided to do it. I remember talking to Richard about it and going through all of these risks, because like it's a really different world for him. Like my husband, the man who's now my husband, he's a white guy in tech and he has a degree from MIT and he was working as an artist in L.A. Like he is all of these things that the context of him killing a robot in front of you seems to me, I thought like that seems cool in a way that I will never be able to pull off and also like, I'm uncomfortable of. But I also, I think I wanted to know what would happen, I wanted to know if I could do it, I wanted to know if I could create this thing that I would program end-to-end that would actually convince me, I wanted to know how simple I could make the code, I wanted to research the minimums.
Christine Sunu [24:15] This is a very short conversation that inspired me to make an extremely in-depth project that I then brought on stage at the talk as sort of this culmination of thinking about this stuff and doing this like research where I spent the whole talk, just talking about like what are the minimums for making a thing seem alive? And this applies to both hardware and software and like people need other people. That's our instinct. We like to be around things that are alive. We have emotional attachments to things that are alive. What is it that people who make technology do to make us feel attached and what are the strategies here?
And so I outlined it and I explained exactly what you do. I explained how it was implemented in this and other projects they did. The whole time I was holding this little worm in my hand. And it was something where I explained that when a creature appears to have purpose, if it's consistent enough, then we think that it has purpose. And if it's somewhat random in its behavior and it has a little unpredictability, then we don't think it's robotic, we think that it also must have emotions or it must be as random as we are.
So the critter that I had made, if you pressed a button, if you squeezed it, it started to struggle. And if you squeezed it for longer, it struggled harder. And then eventually, it started to squeak. This is devastating for me, by the way, when I was programming it, but when we see that as people. We think that it has a purpose, its purpose is that it wants to live. It wants something. It has the purpose like we do, has a want like we do. Their randomness that I programmed and made it so that we don't think it's a robot, we don't think it's an automaton. We think that it has variation, just like we do. We think it's thinking.
And at the end of the talk, after I explained all this, I literally remember standing up there with a slide behind me that had all the code that was making you think that it was feeling an emotion, you could fit it on a single slide. And at the end of the talk, I snapped the worm in half. I had actually created a special part of the spine that would break without breaking my work, so it was designed to be broken, which is also tragic from any given perspective, but it broke in half on stage. In the recording of my talk, I do that and then there's applause as I finish the talk. That is definitely edited. Because I remember everybody when I cut the worm in half and having just like stunned to silence and gasping.
Later on, somebody from the audience came and told me, so she was a cognitive scientist, and she said, "I'm in cognitive science. I know all the things you were doing, I understood it, and I still cried when you broke the worm in half." She cried. I almost didn't do it at all because I was walking around during intermission and showing people the worm and they were saying things to me, like it reminds me of my dog, so I almost didn't do it at all, but-
Liz Moy [27:20]: You’re like “Should I traumatize this crowd?”
Christine Sunu [27:26]: It was the idea of doing a little trauma in order to contextualize the larger dangers that we're putting ourselves in everyday.
Liz Moy [27:36]: No. I'm so glad that you talked about it, because I found it fascinating and as you were talking. It was kind of reminding me a little bit about, did you ever read Snow Crash? So, there's like those little dog robots in Snow Crash and they attack people. But then like there are a couple of chapters that are written from the perspective of the dog robot, and you realize that it's like thinking things like a dog would think. And it's like, I bawled through the chapters because I was like, "This is horrifying." And it's really probably not that far off into our own future of what's possible in terms of the ways that we are maybe not always being as mindful as you are about creating these beings really, these like, these machines, but also these beings that in some ways make our lives a lot easier. But also add these different layers of complexity because of how they integrate into our own lives.
I feel like when we have more data about something, it allows us to understand it better to some degree and even though a sourdough starter is alive to a certain extent, maybe not sentient, I don't know. Some of these sourdough starter stories might be sentient, but I feel like having more data and information about something really allows us to have a different connection with it and have more understanding of it. So, that's just like a thought that I had about your project. But I would love for you to tell me how you came up with the idea for the sourdough tracker.
Christine Sunu [29:13]: So, pretty soon, in the beginning of 2020, we had the Coronavirus hit and a lot of people started doing more from home, a lot of people started baking. Baking became such a huge thing. It was such a huge comfort point for people. It was something where they felt... people seem to feel so strongly, even if they had never made bread ever before that making bread would bring them a level of normalcy that they were losing day-by-day, which was really beautiful and interesting. I think, also there's obviously other factors here and obviously, flour is very cheap. You can make bread with flour and water. There's a sense of safety in the frugality of the creation of your own bread, but it's-
Liz Moy [30:07]: And the carbo loading as a stress relief technique.
Christine Sunu [30:09]: Yeah, it also tastes delicious. Yeah, it has that carb comfort. There's a lot of reasons why people liked bread, but it became one of those things where people were suddenly and emotionally attached to a new thing. And that thing for many was sourdough starters for the first time. I think sourdough starters are amazing. Yeast are just like tiny livestock and so you have this tiny little flock of yeast that lives in this bread culture and you just make bread over and over again. And if you've been keeping sourdough starters for a long time, you can get a sense of when they're healthy, you can get a sense of what kind of bread they're going to make and how it's going to taste even before you make the bread.
And I was certainly before 2020, I was still this kind of, I think the way I've described it to others is like a cat lady for sourdough bread starter. One of the first presents that my husband got me for a Christmas present was an assortment of sourdough starters from around the world. He went on eBay. He found sourdough starters from all over. He like got one from Egypt. He got one from Italy. He got one that was like a really old San Francisco classic sourdough bread starter. So, I still have a lot of these in the freezer, because they keep for a long time.
There's a lot of things that are, I think in many ways. Sourdough starter is the perfect storm of a perfect pet for a science or tech person. It can be databased, it is an organism that you can record, it is very hard to kill, and it has saved points. If you like the way your sourdough starter is tasting at that time, you can dehydrate a part of it and put it in the freezer and then revive it at any time. This is all an extremely compelling proposition and certainly was to me when I first started being interested in sourdough starters, when I was a college freshman who was too irresponsible to have any kind of bread. Pro tip, do not try to make a sourdough starter in your college dorm. It will not turn out well for you or your roommate.
But yeah, I think it was really interesting to see people suddenly becoming interested in sourdough bread. And so it of course, naturally followed that if we had a way to record data associated with something that you were suddenly emotionally invested in, that would compound the comfort that people felt about this object, and it would compound the feeling people were having. So, I’d had a lot of preexisting designs in my head for how you make a sourdough bread monitor. Despite the fact that if you've been making sourdough bread for a long time and you're making sourdough bread for a long time, you don't really need these things.
Like if you're somebody where you just end up building hardware stuff, you've thought about it already. You've thought about, but "what if I took data on it? And like what if I wanted to know how high it was rising and like what the rise and fall was? And what would that look like?" And what would I need to put it in. And so, I had already thought about like, "Okay, it needs to be on the top of the jar, it needs to breathe in this way, like these are sensors that might work." In the end, I didn't actually really use it, giving myself a choice of sensors, because I had just barely started working for Twilio and I had received a Narrowband developer kit that I was playing with. And so, that was already set up and that already had a temperature and humidity sensor and it already had a distance sensor. So, I decided to just put that on the jar and see what happened.
Christine Sunu [34:02]: And what happened was I actually, I got surprisingly good data from it. From a distance, it was like I did not think I would get good data because what that does is it kind of does almost an echolocation thing, like it pings and then it listens. And so I thought in an echo jar, that's not going to work but it actually gave like a kind of good data. And what ended up happening was I could see the rise correlate with the humidity, which was cool, because you could see the jar fog up and then you would see that the humidity had risen, so I built this as a demo. And then I noticed that a lot of the people who were writing about it were sort of like these very like startuppie people who were being really excited. And so, I thought it would be kind of amusing or like good for them if the page was also something that was within a category that they were used to seeing.
So, I bought the domain sourd.io, like sourdough, S-O-U-R-D-dot-I-O and put what looked like a very startuppie page on it with these pictures of the device. And talking about how it's going to monitor your bread and having really Buzzword-y sounding things about it. And then also explaining what the Twilio Narrowband kit does and like what Twilio Narrowband is because like that is my job also. And that was something that people really liked. It was funny to watch that go around. People really like data and people really like bread, and people really liked data and bread at that specific moment in time.
Liz Moy [35:45]: Yeah, it's so great. I love the site, because it kind of reminds me almost of The Onion or something where it's like done so well that it could be like you could just start a company off of that, but it's just so clever. I had a lot of questions about the tech for you because I am not super ingrained in the IoT or the hardware world. One of the ones that I had was, okay, so in the setup, you mentioned that folks can choose an MQTT broker of their choice. So, is MQTT like the HTTP of the IoT world or what is that exactly?
Christine Sunu [36:30]: It's a communication protocol, like HTTP sort of that is, with like some obvious differences. It's commonly used, but not necessarily the standard. There's a lot of, you can actually typically just do an HTTP request also from a connected device, and many do, too. But the system that we had made for Narrowband use MQTT. There's a couple of advantages to using MQTT over HTTP, but it is also widely debated, and so I'm not going to get it too much because it's going to deeply depend on your use case. But suffice it to say that it's a protocol that can be popular among internet connected devices and it is a protocol that there are several existing services for because of that popularity and use case.
So, I think that I had put out a couple, one of them was the Adafruit IO one, one of them was from ThingSpeak, and then one was an open broker that we sometimes suggest in the Twilio docs called HiveMQ. So, these all are very different from each other. The ThingSpeak and Adafruit I wanted to include, because it gives you the ability to immediately graph what you're seeing come in. And I think that if you can save yourself any work in the frontend after you have spent all of this time doing the hardware, the firmware and the backend, then like save yourself the trouble.
So, I quite liked the ThingSpeak broker actually. I remember I started using that. I used that second because I used the Adafruit one first, which was also good. It was like such a nit to pick. Like the Adafruit one in the line graph, it doesn't have dot on the line graph whereas the ThingSpeak one is a dot line and so then it was easier for me to see some of the data, like that's like such a personal preference thing. Both of them work. They have slightly different ways that you communicate with the broker and so that is what is included in the library of the build.
Liz Moy [38:36]: So, the other question that I had from the build that you did and from the GitHub Repo was that the sensors looked like they were using a library from something called Seeed Studio?
Christine Sunu [38:49]: Yes.
Liz Moy [38:51]: Was that a choice that you made or is that kind of a standard library that a lot of folks use working with the sensors?
Christine Sunu [38:59]: Seeed Studio produces the Grove Ecosystem. Grove connectors are often seen on boards, but it's basically a system where you can plug in the sensor through this very specific connector and then there's a bunch of libraries for them, so you can add a library. The board that we have - the Twilio Narrowband board - is very similar to and uses a lot of the libraries for an existing board by Seeed Studio and it's and it works with the Grove Connector Ecosystem, it works with the Grove Ecosystem. So that's why the libraries are from Seeed.
Liz Moy [39:40]: Cool. That's really fascinating. What differences have you noticed in working with the Narrowband kit in particular, as opposed to that might be like a super broad question, but just as opposed to other devices that you work with? Like what are some specific things that you've encountered working with Narrowband?
Christine Sunu [40:00]: So, when you work with a Narrowband connection, the idea is that you open the connection and then it just stays open, so it might like take energy to power it up, but then like once it's open, it doesn't take as much power to keep it open and to just send data occasionally. It's really meant to be used in cases where you're not sending really big data and you're looking for a device to probably be always on and you're probably not moving the device around a lot, because it's cellular, right? So if the Narrowband kit turns on or the Narrowband device turns on, it connects, the connection is open. If I then move that across town, then it has to connect again, which takes more power and is a whole other thing.
So, that's like one of the big decision making points in their event is like you want to make sure that your applications are right for it. A lot of times we think about Narrowband as being a connectivity tool for what you would call like massive IoT, like a lot of sensors in the field that occasionally send data, the data is not time sensitive, but we want to know a lot and we want that information like cheaply and it should just be like these static devices that are just giving me data all the time. So, it's like smart city stuff often gets referenced I think with Narrowband.
In the decision making for this, the thought was, "Well, you're probably just going to leave it on your kitchen counter. And it needs to keep the connection open and other than that, it should be good." Now, it doesn't connect you for a while when you move it to the fridge, so that's a problem. But there's software fallbacks that you could build, which I did not know, but one lets you save that data and then send it later. But the thing that I noticed also with Narrowband is that you do want to keep that connection open.
So, in the case of the Twilio Narrowband kit, you do have to ping, you have to send some data on a certain cadence in order to let it know, in order for the carrier to know you're still there. Please don't quote me on that exactly. It would get unhappy if I didn't send data within a certain like requisite number of seconds. I think it might have been like 20, 30 seconds. I'm trying to remember. But anyway, if you are building the device and there is a problem, check if you have changed the rate of communication.
Liz Moy [42:36]: Do not take your sourdough starter on a road trip with you.
Christine Sunu [42:40]: Also that. Do not take it on a road trip. I mean, you could. It would just lose connectivity and then connect again to the deficit of your battery.
Liz Moy [42:49]: I love everything about the project, but I especially love how nice it looks. It just snaps on to the top of the jar and it's so cute. I would love to hear about how you decided to design that, that part of it and also the process of I think, you used a 3D printer for that?
Christine Sunu [43:11]: It's a problem I have. I have a huge problem where I try to make everything sort of cute or like an animal. It's an instinct I actually have to fight. I used to have to fight a lot when we were doing more consulting work for rapid prototyping, because it's not what everybody wants. But I think a lot about enclosures and how it looks because of everything we've talked about already. I like to think about how it's going to sit on your counter, and it's going to be there and especially when it's representing a living creature, it's cool for it to kind of look like a pet. So, this was one where it was actually appropriate for it to look like a cute animal.
When I started designing the enclosure, I usually start on paper. I just draw a bunch of sketches of like what do I think it might look like and what might be cute to have. And so, there are a couple that I drew, like I try to, and in the cases when I have a board like this, I have to work within the constraints of the board, right? So, I knew that the board was a certain size and it was going to have a certain size. I knew I wanted to put it on the top of the jar because then you could remove the bottom of the jar to make it easy to clean. I knew that it was going to be raised in a certain amount, a certain height off of the top of the jar because I had stacked all the components previously and kind of measured it with my calipers, so I think about like what's the minimum height that it would have to be and so, there's also the antenna, so I had to think about like where does the antenna go.
So, I drew a couple of hypothetical designs around like how do you make it feel like it's a pet while having this height off the top of the jar and being able to incorporate this antenna somehow. And the one I landed on was this one where it's kind of this face and it has little, it has like something reminiscent of a face and it has like eyes and then it appears to have like a little hat where the antenna sticks out. This was the design that was the easiest to make. We were worried about getting enough filament because a lot of people were running low on filament.
And it was during a time where people were scarcity buying and during a time where people were doing a lot of printing. I was also trying to do a thing where I knew that it wouldn't take that many tries for me to get it right and I would be able to serially print things to enclose it, I would be able to make an enclosure that would not require a lot of iteration. I didn't make it too complicated, so that's also why it's just sort of this object that sits on top, and it has the little face, but it's not overkill on the cute.
Liz Moy [45:42]: What kind of 3D printer do you use in your shop?
Christine Sunu [45:46]: So, we have a Creality 10S that we have in a coat closet. The reason for the coat closet is partially noise, but mostly temperature control. So, a 3D printer is affected by a lot of conditions, but one of them is certainly that it's going to print more consistently if there's less temperature variation because what it does is it melts the filament through this extruder and then the filament cools as it goes around and pattern. And so, if you change the outside temperature, it might cool differently.
And it sounds like it should be like a nothing variation, but like actually, when I've printed things in the past, using this closet, I'll sometimes open the closet with a really like strong motion and like blow all of the hot air out and I can see like a line later from where it changed the ambient temperature around the printer. So, we use a Creality 10S, which is already quite a quiet printer as printers go and it's a good high-resolution printer and we keep it in a coat closet because then the temperatures are really consistent.
Liz Moy [46:53]: I did want to ask if there were any bugs that you ran into or any sort of like problems that came up along the way that you didn't expect?
Christine Sunu [47:00]: Yeah, I mean, hardware is hard. Well, one of the things, this build was for me an experiment in some ways to just see which if any parts of the sensors actually would reflect good data and so, one of the things that I didn't expect, actually I kind of expected this, let's be real. I didn't know how useful the temperature was going to be. I didn't know if I was going to see any variation in the temperature. And given the existing libraries for it and I hadn't done a lot of research for this specific sensor, like there's probably better, there's a lot of like better, more accurate temperature sensors that you can use that you probably would see variation of, but I just wasn't sure what this one. And so it was interesting, because what I didn't expect was to see the higher variation in humidity and so, that was really cool.
In terms of debugging, there's always stuff. If I've had more than a month's distance from the project, I've had amnesia about everything that was hard about it and that's why I have a cadence of about like a month between building projects on the same stack where I'll say, "Oh, yeah, but I definitely used like such and such to build that other thing, like that wasn't that bad." And then I'll go back and there'll of course be 10 million things I forgot that I hated.
In this case, like I remember, the process as being pretty smooth. I had already done the setup for it, though. So, you have to follow the setup that Twilio outlines for you in a specific area of the docs, which I think is now much easier to find. I like to solve the problem and then document the easiest path to set up people for success in the future. So, that was the other thing, like any bugs that I ran into were pretty solved I think by the time that I put that post out.
Liz Moy [48:45]: So, you had a tweet recently, where you were saying that hardware is hard and that the more barriers to entry that are created for people, it makes it that much more difficult for someone to get started. I could definitely relate to that and you and I have talked about this a little bit. But a lot of the experiences that I've had with hardware were in settings where everything was kind of set up for you, you follow a very specific process, which can be great when you're learning because you can get something to work and there's that reward of getting something done and having it run. But for someone who wants to venture into that world and wants to get started kind of building something on their own, like how do you do that? How does that begin?
Christine Sunu [49:40]: Yeah, hardware is hard. That is something that people say all the time, particularly people who have spent any time in hardware, but the highs are high and the lows are low and there's more things that can go wrong. And it's hard to figure out where to start or what the vocabulary is and you're worried about saying the wrong thing and you don't know who to ask or who to talk to, or what to Google.
I got a little lucky with a lot of my intro to hardware being at Particle because one of the goals of the company was almost to make it easy for people who already did web stuff or who were more designy to be able to build a prototype. And so, I was in an environment where my feedback as somebody who is just getting started and my questions were given priority and they were totally accepted. Where like if I had a question about like why does it look like that or why does it sound like that or what do I Google to try to build this thing? People were just immediately with me. They were like, "Cool. Yeah, you do it like this." And everyone was really friendly. And I think that's actually the way it is in like a number of hardware communities, but it doesn't feel that way.
When you look at the landscape of and I think this is the anxiety that we all have, in almost, doing almost any new activity, like you just don't know. It's just nothing feels friendly when something's new. And I got really lucky with my intro to hardware and where I ended up. I mentioned about knowing what to Google since there's so much information out there and you just have to know where to start. And I always say like, "I will tell you what to Google if you have a question like you can ask me, like send, like tweet at me, send me an email like I will help you find out what to Google," because that was such a blocker for me.
But the other part is it's just hard. It's hard to stay motivated. It's hard to find the time. It's expensive when something that you've been working on breaks, it's just like feel so bad and a lot of things can break. And I used to try to make it feel like hardware was easy. When I did this series on the Verge, the idea was to make it feel like a cooking show and to make it feel achievable. And like I think that you can give people outlines where it walks through step-by-step kind of like what you're talking about where does feel easy, and it does feel achievable. But when you release people to do the thing on their own, it's much more of a complicated question. So, I used to say like "Hardware is easy, you can totally do it." Now, what I say is, "Hardware is not easy, but it's so much less hard than so many of the other things you've done." Yeah.
Liz Moy [52:24]: Are you ready for this or that? I will read off two options and you have to tell us which one you choose and why. And they are somewhat tailored towards you.
Christine Sunu [52:40]: Uh-oh.
Tabs or spaces?
Christine Sunu [52:45]: Tabs. People are going to be mad that I made that sound about spaces. I always tabs, I can't — tabs.
Fry's Electronics or Adafruit?
Christine Sunu [52:59]: That's hard. I have to say Fry's mostly because I love to go into the physical space of Fry's and because I would be very, very sad if either of them disappeared, but I think that currently Fry's is in more danger of disappearing. And so, I'm going to say Fry's. It's very close to my heart and I’m feeling emotions about it right now.
Liz Moy [53:24]: I used to live really close to the Fry's or not live, work really close to the Fry's in Burbank and some of my co-workers would like go there over lunch break and stuff. So, it's definitely a Southern California beloved staple.
Christine Sunu [53:38]: Adafruit is where I order from when I'm prepared and Fry's is what saves me when I haven't been prepared enough. So, that's why Fry's if I had to pick.
Raspberry Pi or Arduino.
Christine Sunu [53:58]: So, depends on what you're doing actually. For me, I generally use Arduino based things because mostly what I do is really small. I don't typically do things that require the entire computer, power of an entire computer, so I end up using microcontrollers, just typically more Arduino stuff, but I'm actually more into Raspberry Pi. I always ask myself that question. I'm like, "Will you finally use Raspberry Pi today?" And then I don't feel the feeling that I need to feel in order to get over that activation energy barrier, but maybe I'll do it tomorrow. We'll see.
Liz Moy [54:35]: Well, that makes sense, so like yeah, it kind of goes back to what you were saying earlier of use the tool that you really need to get whatever the job done, the job is done.
Coarse fur or soft fur?
Christine Sunu [54:50]: Like in what context? Like on an animal, like a live animal or a robot animal?
Liz Moy [54:54]: A robot animal.
Christine Sunu [54:55]: On a robot animal, soft fur. If you are trying to do, actually, I should be more specific. If you're trying to do comfort, soft fur. If you're trying to do realism, coarse fur. Sometimes, the coarse fur comes off as more realistic because we think of it as being like what an animal has, but it also just depends on the quality. They are like so many different like qualities of fur, like aspects of like goat fur and like shininess is also one of them and like certain levels of coarse are one of them. But like we actually have this fur that we use for this giant Capybara robot that is like it's very soft, but it has a certain level of coarseness to it that feels a little bit more realistic to animal, but I think that people respond really well to things that feel like a teddy bear in a lot of ways, so like I try to offer soft fur most of the time.
Liz Moy [55:49]: Wow, I wish that we had time to go into the Capybara, the giant capybara robot.
Christine Sunu [55:55]: It's not that big.
Liz Moy [55:56]: You dropped that like it was nothing. It's like my happy place whenever I feel stressed out. I just close my eyes and I pretend that I'm in like a Japanese hot spring with a bunch of Capybara and some lemons floating around.
Christine Sunu [56:11]: You mean Nagasaki Bio Park? They're not the only ones who do it, but I definitely have watched hours of videos from Nagasaki Bio Park of all of their different Capybaras. They have a lot of videos.
Crawling or rolling?
Christine Sunu [56:27]: For expedience, okay, so I can tell you about this interface I keep meaning to build, but I haven't built yet. Okay, here's the thing about crawling. Crawling is, so when you are asking a device, when you're thinking about a device, like this device is now going to look moot. You have a bunch of issues that come up. "One is like how well does it need to do it? Like does it really need to get to point A, from point A to point B consistently? And is it not getting there going to be seen as like frustration? Is it going to be, like what emotion are you going to feel when it doesn't get there? Are you going to be frustrated or are you going to be like, how fast does it have to go?" There's a lot of questions.
Wheels give you consistency. Wheels also technically make us think more robotic, but what crawling often goes to for a lot of people is like an insect. That's usually not a place you want to go. Really slow crawling sometimes like strandbeest level crawling. People are interested in from like, "This is fascinating," they'll be like, "Now, it's cuddly." It's like crawling, you have to be very careful how you use it. Rolling, you don't have to be as careful. Like, people expect wheels from robots in a lot of ways. And there are ways that you can dress up rolling, so that it feels alive, but you just have to be careful.
So, I would say overall rolling, although there is an interface that I keep wanting to make for it, but not really an interface, not really interfacing with it. There's a setup I really want to do for something that rolls, but it looks like it's crawling, where you just have like a sort of tasseled edge. Like imagine like a fuzzy blob and then it has like a tasseled edge on the bottom of like fuzziness. And as it rolls forward, you just see like the edges go like putt, putt, putt. Like it just kind of like it would be that you would have a tank track and then you would have like a sticking out part of the tank track that like goes around in a circle to like make the edges move.
Liz Moy [58:23]: Yeah. One of the talks that you gave - I was watching and there was a robot that really looked kind of like a spider. And I was like, "whoever came up with that?" I mean, I don't know about that. No offense if you're listening to this podcast.
Early morning or late at night?
Christine Sunu [58:41]: Early morning. I did not realize that I was a morning person until I started living with people who weren't morning people. I have a lot of like moments of clarity in the morning. Lately, what I've been doing is I get up, I drink a glass of water, I stretch and I go outside and I feed the chickens. And that's a really nice time to see like the chickens and the rabbits hanging out together. It's just really, really funny. And then I come back inside and I work on like my robot or whatever I need to work on before I start actual work. And that's really nice. I think that's one of my favorite, it's like my second favorite time of the day.
Liz Moy [59:17]: I am with you. I am an early morning person most of the time. Occasionally, like I don't know. I'll go through the different phases where it changes a little bit, but I find the same thing like I like watching the sunrise. I like just like sort of the solitude when you first wake up and you can just really focus in on stuff.
Soldering iron or hot glue?
Christine Sunu [59:40]: Oh, they're totally different. You've heard me say this, Liz, like I think that solder is metal hot glue. I mean, it's not. It's super not, but I always think about that because I get mad that hot glue was available to me from a very early age, but soldering was not.
Liz Moy [60:01]: Christine, thank you so much for taking the time to be on Build and Deploy today. If people want to find out more about your work, where can they find you?
Christine Sunu [60:12]: I guess you could follow me on Twitter. I'm Christine Sunu on Twitter. I also have a site and a portfolio site. You can hop on to christinesunu.com or christinesunu.info, which is my portfolio. You can check me out on the Twilio blog. I now write things for Twilio. It's really fun.
Thanks for listening and subscribing! Let me know your thoughts in the comments or talk about the show on Twitter @builddeploypod.
From APIs to SDKs to sample apps
API reference documentation, SDKs, helper libraries, quickstarts, and tutorials for your language and platform.
The latest ebooks, industry reports, and webinars
Learn from customer engagement experts to improve your own communication.
Twilio's developer community hub
Best practices, code samples, and inspiration to build communications and digital engagement experiences.