Turing Award and CTO of Hopara, Mike Stonebraker, and Hopara CEO, Gant Redmon, join Ryan Chacon to discuss data visualization. Mike gives a detailed explanation of data visualization and its role in the IoT industry, along with insights into its current landscape. Mike and Gant also speak on common challenges and solving these issues for companies working in the space before giving their thoughts on what they look forward to in the space in the near future.
About Mike and Thomas
Mike Stonebraker is the founder and CTO of Hopara. He won the 2014 Turing Awards and has worked at 10 commercial startups. Gant Redmon is the CEO of Hopara with a strong record of growing and maximizing asset value.
Hopara is the next-gen viz app creator designed at MIT CSAIL specifically for big/real-time data, IoT, Digital Twins, and ESG. Their apps make it easier for any end user to engage with data in a dynamic, visually rich environment that requires no special skills or training. Hopara gives data stakeholders the power to access and understand comprehensive data stores in context as easily as surfing the web.
Key Questions and Topics from this Episode:
(06:40) What is data visualization?
– [Ryan] Hello, everyone and welcome to an episode of the IoT for All Podcast. I’m Ryan Chacon and we have a very special episode for you today. We have two guests, Mike Stonebraker, the founder and CTO of Hopara and Gant Redmon, the CEO of Hopara. They are a next gen data visualization app creator company. And the reason this episode is so special is because, Mike Stonebraker is actually a Turing award winner, which is basically the Nobel Prize for computing. Very, very cool, kind of, conversation we get to have today focused on what data visualization is, why it’s important to delivering ROI in IoT and what the future of the space looks like. So, really think you’re gonna get a lot of value outta this, but, please, be sure to like this video, subscribe to our channel, and hit that bell icon, so you have the latest episode as soon as they’re out. But, other than that, enjoy the episode. Welcome, Mike and Gant, to the IoT for All Podcast. Thanks for being here this week.
– [Gant] Thank you, Ryan.
– [Ryan] Yeah, it’s great to have both of you. Looking forward to this conversation. Let’s kick it off by having you give a quick introduction about yourself and the company for our audience to have a little context.
– [Gant] After you, sir.
– [Mike] Okay. I’m Mike Stonebraker. I’m on the faculty at MIT. I guess my main claim to fame is I won the Turing award in 2014 for my contributions to databases. If you’ve used Postgres, I built that. If you use Vertica, I built the prototype that turned into that. If you use SciDB, ditto. And so I do databases, but I’ve had a long interest in information discovery. I’m a huge fan of Google Maps and, as you probably all know, it’s a drill down pan and zoom paradigm where you can get from a picture of the world to a plot map on your home street in 11, or sorry, in 21 clicks. And so I’ve had a long, long interest in information discovery, since I do information.
– [Ryan] Fantastic.
– [Gant] Absolutely, and because of all of that, Mike, I thought it would be a great bet to join you. So I’m CEO of Hopara and have been so for about the last year and really enjoying our journey.
– [Ryan] Tell me a little bit more about Hopara and, kind of, what you’re all focused on, the role you played in the space, that kind of thing.
– [Gantt] I’ll take it, sure. So, Hopara, and this is really part of, Mike’s journey, is also the journey that brings us to Hopara. I mean, you heard, Mike has created the databases that’re used all over the world. Then he really got into the normalization of data. How do you make all these different databases work together? But, now you have your data, you have it organized how do you see it, how do you use it? How do you take this to the next level of visualization to really utilize that data that you spent all this time getting together and organizing. That is what Hopara does, that’s the solution now of taking what is, what we think is, an antiquated way of looking at your data and taking it to a new level.
– [Mike] So, I think-
– [Ryan] Mike. Yeah.
– [Mike] So I think I’d like to just tell a little vignette about one of Hopara’s customers. They are a… They build sensors, they’re in the IoT space. They build vibration sensors that you put on machines. And I didn’t know, a while ago, that you put a vibration sensor on a machine, because it starts to vibrate before it fails. And, so, that allows you to do preventive stuff or order a new one in advance. So, anyway, one of our customers is an IoT vibration sensor company in Brazil and they, of course, sell their sensors. And one of their customers is a very large manufacturing company in Brazil, which has 58 factories across Brazil and they have thousands of these vibration sensors. And IBBX, of course, has a traditional dashboard where you can display data about what’s happening with your sensors. You can have the average number of bad values and they have a machine learning system that predicts when a machine is going to fail if it’s started vibrating. And it’s, sort of, a standard tableau spot fire, kind of, information visualization system. And this large customer, the 58 factory guys in Brazil, they said, “Well, that’s all really nice and we like that, but we don’t get the big picture. We don’t, you know, we can’t see what’s happening in our 58 factories.”. So, IBBX partnered their traditional dashboard with our information discovery drill down system where you can start with a map of Brazil with 58 dots on it, representing the factories. They’re color coded, green, red, or yellow. If there’s one that’s red, you can drill into it, get a picture of the factory, figure out where the sensors are that are, you know, out of bounds and drill into specific machines. And, so, it’s a Google map style drill down system and you can… It’s very much like Google Maps on steroids. You can use maps, you can use floor plans, you can use canvases that represent other things, and you can drop any, kind of, data on these canvases. And, so, it’s a great information discovery system that compliments a traditional dashboard. And, so, in our opinion, the thing we’d like to convey to your readers is that they probably have the standard, kind of, traditional dashboard right now. And they should augment it with a sophisticated information discovery system such as the one we have. ‘Cause an eyeball on complicated data is an extremely valuable tool that complements, sort of, the standard presentations.
– [Ryan] Fantastic. Yeah, I appreciate that kind of context. Kind of puts it all full circle to understand exactly how you work with customers and, kinda, what they’re doing. So, that’s great. Thank you for sharing that. What I wanted to do is I wanna shift over to talking a little bit high level about data visualization in the IoT space. And if you could just, kind of, kick it off, either one of you, whoever wants to take it, just high level for somebody who doesn’t really understand what data visualization actually is. If you were to describe it in a very, kind of, quick short, you know, high level way, how would you describe it to somebody?
– [Mike] Well, I’d like to distinguish visualization from information discovery and presentation. Visualization is about rendering teapots and that’s not what any of your readers or any of our customers do. They are drowning in realtime data. In the IoT space, there’s a data deluge coming at you, because your readers are in the process of sensor tagging everything of material significance to report their state or location in real time. So there’s this tsunami that is drowning you and you either pass it to analysis systems, which produce you tables of numbers and that’s valuable, in certain circumstances, or you pass it to an information discovery system that can allow you to see the high level and then drill into your data. Just for example, a while ago we had a prototype with a large hospital here in New England and, as you may know, hospitals have a horrible time with infections and on the average, across the country, if you are a surgery patient, there’s a 1% chance you’re gonna get a nasty infection. And if you start asking doctors, you know, where these infections come from, they say, “Well, I don’t know, they might come from contaminated instruments. they might come from somebody being infected using, you know, a specific machine and then it wasn’t sufficiently cleaned and you used the same machine et cetera, et cetera, et cetera.”. So, they have a record of every single patient at this particular hospital and where they were geographically in the hospital over time. That’s a gigantic amount of information. So, they said, “What we would like to see is render the infected patients on a floor plan of our hospital so that if we can drill into hotspot rooms we can maybe get some insight.”. So we did exactly that and we showed it to the head of infection diseases at this hospital. They looked at it and said, “Wow, the hot rooms are right across from the nurses station. And so the nurses are clearly putting infected patients nearby so they can keep an eye on them.”. And that’s an insight you would never get by looking at a mountain of data. So, basically, these systems are incredibly valuable when you don’t know what you’re looking for and you want insight, you know, your question is tell me something interesting. And if you know the question to ask, well, go ask it with your favorite query or analysis tool. But, it’s often as not, you don’t know. So, that’s where our product excels is when your query is, I’m drowning in data tell me something interesting.
– [Ryan] Fantastic. Okay, great. And how would you, you know, for our audience who maybe is trying to understand exactly what this, the data visualization, kinda, landscape currently looks like, how would you, kind of, describe that? Like, how has it been done? You know, how is the landscape, kind of, currently set up, and what is, kind of, the approach to doing data visualization the right way? Or what’s the things should you be, kind of, thinking about?
– [Mike] Okay, so, first of all, I’d like you to distinguish two cases. One is your readers all have, you know, IoT data. So, case number one is you have 10 sensors or you’re tagging ducks, you know, in a marsh and you don’t really care about real time and you don’t really care about scale. So, if you have a small problem the answer is do it any any way you want to. But, if you wanna do it at scale then it becomes important to worry about your data and worry about how you’re gonna look at it. So, my interest is at scale. So, at scale, the first thing you absolutely should do is put your data in a database. ‘Cause lots of people don’t. And if you don’t put your data in the database then chances are you lose all the semantics of what the data means. One of my favorite bugaboos is people who put all their data in files and then encode the metadata in the name of the file. So, you know, a file name is recording x, y, z 10, 11, 12, 31, which means that you, you know you have to parse the file name to have any idea what the data means. And, so, if you put your data in a database system it forces you to put the metadata there. And that just means that downstream systems will love you, ’cause they can read your data and, so, at scale, put your data in a database system and then you’ve gotta use a viz system that scales. So, my favorite example is, let’s suppose you wanna look at the population of the United States. There’s 330 million of us and you want to put a dot at the geographic location where everybody lives. Well, that will paint the screen black in downtown DC and it will really paint the screen black, you know, in New York City in Manhattan. So, you need a system that will allow you to do drill down so that you don’t just display the finest granularity and it just paints the screen black. So, in my opinion, a drill down system that supports real time and scales is the key things that you need. As far as exactly how your data looks, there’s just an enormous number of personal preferences here. As, say, I particularly like looking at things Google map style, but as, say, Gant or you may wanna look at it differently. So, you need a system that can display your data in a wide variety of ways at scale in real time. And, that’s a pretty rare kind of thing. So, I think, you know, to me the answer is decide whether you wanna look at your data in real time. If you don’t care that the display can be stale. If you can read yesterday’s data it makes life really easy. But, my interest is the people that own the 58 factories in Brazil are not interested in yesterday’s data. And, so, in real time, you’ve gotta have a flexible information system that will display the data the way the user wants to see it and work at scale with some sort of drill down capability.
– [Gant] Yeah, in the current state, I think what we’re used to is seeing pie charts, graphs, bar charts. We see it in all kinds of different things and they have their place, but they all look the same. We’ve been looking at the same thing for decades. But, that is going to give you something that is not as valuable to the eye and the brain. There is certainly a place for it you can always use it for lots of good things. We’re looking at what comes after that.
– [Ryan] Okay. Yeah, fantastic way to, kind of, put it. I like that. What are some of the, I guess, common challenges that you come across when you work with organizations that are trying to, kind of, implement data visualization and do it the right way? Like, are there, kind of, common challenges that companies have or that you come across more often than others that, you know, maybe could be avoided if they knew about it or, kind of, could prepare for it? I’d be curious to, kind of, hear, kinda, your experience on that front.
– [Gant] You know, I hate to give you a circular answer. But, I’m gonna. So, yeah, it’s because, same as what we were just talking about, people have become accustomed to something. And then they’re presented to this new way of looking at things. A way where they can navigate their own way through. They can see things with greater insight, they can put things in context. I would say one of the greatest challenges is giving people the ability to visualize what they want to visualize. And that’s almost the greatest challenge of a limitless opportunity is applying it to how best suit oneself. Now, the way we work through that is we give more and more examples, “Hey, try this, try this, try this.”. The other day worked out really well. A customer was looking for solutions. We gave them three, Because, you know that way they could… It was like being a designer of art. It’s like, well, which one of these do you gravitate towards? So, it’s really, if you have an unlimited number of layers and canvases and seeing things it’s really just helping people put those all together in a way that’s most valuable to them.
– [Ryan] How often do they know what they want when you talk to them? Or is it usually, does it take you to, kind of, show what’s possible before they really can align and understand what it is that they truly want versus what they maybe think they want or maybe they don’t know?
– [Mike] Well, first of all, biggest problem number one that we see all the time is messy data. And, so, data that has not been curated so that it makes sense And, so, if you have garbage in the answer is you’re gonna get garbage out. So the first, you know adage is, by goodness, clean up your data. And then the second adage is hardly anybody knows how they would like to look at their data. They said, “I know I wanna look at it, but I’m not sure how.”. And then what Gant just said is the obvious answer. Give them a pallet of 10 things and say, “You can do this, this, this, this, or this.” And they say, “Oh, that looks nice. Show me that.” And so you mock that up and they say, “Oh, that’s nice, but I don’t like this and change that over here.”. So, it’s, basically, it’s interactive design with us helping the customer figure out exactly what he would like to see.
– [Gant] Yeah, it’s the pain conversation. You know, what is your pain? We had a customer a month ago that… We had lots of customers that would have six to a dozen devices in a room. And now we’re talking to a customer that has hundreds. So, we actually had to work with them to how do you visualize that? And we came up with a new thing called proportional zoom. Where when they, you know, the problem is, like Mike mentioned, painting the screen black. There were so many devices that they started to overlap, but we just came up with technology that will allow things to resize and separate. So, you could go down into the individual device and not have that overlap and really understand what was going on in that room. So, yeah, you start with a problem, “Okay, I’m dealing with massive numbers in a single room, how do I visualize that in a way that’s useful?” So you start with a problem then you find the solution through the visualization.
– [Ryan] Absolutely, that’s fantastic. One of the things I wanted to ask you all, kind of, at a high level standpoint as we, kind of, start to wrap up the conversation here is what does the future look like in this space to you all? Like, what are you most excited about? What’s happening that maybe our viewers and our readers are not aware of that they should be, kind of, paying more attention to the data visualization space and as connection to, kinda, the future of the IoT industry?
– [Mike] Well, I think the first thing is that your readers are in the catbird seat, because, as a society, we’re in the process of just sensor tagging everything of material significance. And these sensors are going to largely turn from passive to active. Meaning, we used to put stickers on things and those stickers will turn into active tags so that you can find stuff in real time. And, so, I think, the first thing that’s gonna happen is that that’s going to give everybody a scale problem, ’cause they’re gonna go from hundreds of sensors to millions of sensors. So, I think that is coming in a big way. I think the other thing that’s coming at us is that sensor technology is getting cheaper and cheaper and cheaper. Which is going to accelerate the adoption of sensor tagging everything. Beyond that, I think cheap hardware is gonna help a lot. Right now most visualization systems are 2D, because 3D is too expensive to be interesting to most people. But, I think 3D will get cheaper and cheaper. And, so, we’re looking forward to moving to 3D technology, because if you’ve ever been in, let’s say, there’s a visualization professor of Brown, Andy Van Damme, who’s now quite old, but 20 years ago he had a thing he called The Cave, which was you walked into a room and in all sides it was the, you know, piles of displays. And, so, you could just look at the amount of information you could look at by turning around was a couple of orders of magnitude more than would fit on one screen. And, so, I think, you know, immersive experiences will come over time. Maybe Mark Zuckerberg and his Meta stuff will turn out to work. But, I think that’s in the research phases and it will come to pass over time. So, I think, figuring out how you’re gonna deal with immersion technology and how you’re gonna go with 3D is, sort of, on the horizon. Yeah, I- Sorry, go ahead, Gant.
– [Gant] Oh, no, I was also gonna say, I think the future with that tsunami of data, with all that coming at you. You know, we’re all in business and we all want to be successful individually and our companies. And when we’re gonna get into these projects we do have to show return on investment. And I believe a really big part of that return on your investment is the usability and understandability of the data. So, there’s more and more and more. How do you make that data more usable and understandable to a larger group? Because, I think that is gonna become a major metric for success.
– [Ryan] I completely agree. I think making the data accessible and easily understood and by, you know, by that intended end user is really where the ROI comes in for a lot of these companies. ‘Cause, like, Mike, like you said earlier, about, kind of, that garbage data in garbage data out type of thing, you solve that problem and you start to get good data coming out. It’s then, how do you make it something that can be used? How can it be easily visualized? Whether, like, it’s 2D, 3D, but, you know, we’re gonna be getting data from so many more sources not just how do we process it all, but how do we present it in a way that’s functional and usable and valuable to that person at the end of the day that’s gonna be actually using that data, which, you know, that persona, kind of, varies quite greatly on who it is that’s gonna be interacting with data across different use cases, industries, you name it. And being able to solve that and focus on that for each particular use case is what I’ve seen from conversations in the past. It’s a real key to the value here.
– [Mike] Well, I think a lot of the prospects and customers we talk to are also Tableau users. And they hate the fact that Tableau is, basically, a programmer tool. It is not an end user tool, at least not to most of the prospects we’re talking to. And they really want something that is end user programmable so the end users can tailor displays to what they really wanna look at. And we’re working hard on making an end user accessible system, ’cause I think that’s what will drive the ROI. I mean that’s what will bring the software costs down and deliver an attractive ROI.
– [Ryan] Absolutely, yeah. This has been a fantastic conversation. I really appreciate both of you, kind of, taking the time. For our audience who wants to potentially follow up on this conversation, learn more about the company, reach out, engage in any capacity, what’s the best way that they can go ahead and do that?
– [Gant] our website. Hopara.io Feel free. Yeah?
– [Mike] You better spell that.
– [Ryan] It’ll be plastered everywhere with our promotion, I promise. They’ll know how to spell that one by the time they see it. It’ll be on the title, it’ll be in the description and the write ups, everything. So, but feel free to, if you want to just say it out loud, but-
– [Gant] Yeah, you could always get me, Gant Redmon is gredmon, g-r-e-d-m-o-n @hopara.io, would love to be chatting. Find me on LinkedIn. There’s only two Gant Redmons. It’s me and my dad. I’m the one that’s not 85.
– [Ryan] That’s a unique one. I search for names sometimes on LinkedIn and it’s just, you know, you gotta really- You gotta start putting in title or company names and locations and just to narrow it down. But, that’s very convenient. But, Mike and Gant, thank you guys so much. It’s a pleasure speaking to both of you and I really look forward to getting this out to our audience. It’s, you know, it’s a topic that I think is very important to, kind of, talk further about and you two are, obviously, the experts in it. So, I really appreciate you taking the time to share with our audience.
– [Gant] Thank you, Ryan. You’re doing great work, appreciate it.
– [Ryan] All right, everyone, thanks again for watching that episode of the IoT for All Podcast. If you enjoyed the episode please click the thumbs up button, subscribe to our channel and be sure to hit the bell notifications so you get the latest episodes as soon as they become available. Other than that, thanks again for watching and we’ll see you next time.