datamaster summit 2020

How 5G, IoT, and Cloud Shift Data Management Paradigms


Rich Miner

Android Co-Founder and Investor, Google Ventures

The era of mobile, cloud and IoT shifts the way enterprises deal with data, requiring organizations to handle data at scale, eliminating silos, and using it for better business outcomes. In their open conversation Rich Miner, investment partner on the GV team and co-founder of Android, Inc and Andy Palmer, CEO and Co-Founder of Tamr will discuss how the rapid developments in the mobile and IoT rapidly transform the way data is managed by organizations. Rich and Andy will discuss the benefits and challenges in a world that is rapidly transformed.


Speaker 1:
DataMasters Summit 2020 presented by Tamr.

Rich, thanks for joining us at DataMasters. It’s really fantastic to have you here. And for those that don’t know, Rich is absolutely a seminal figure in the start-up ecosystem in Boston as the co-founder of many companies, but probably most notable Android. And then is one of the core members of the GV team as GV was getting started and investor in many companies in Boston and around the world. Rich, it’s really a pleasure to have you here today and really excited to spend some time talking about data and maybe a little bit about Tamr and also anxious to hear stories that you might have about data as it relates to Android in the mobile ecosystem. Thanks for being with us.

Excited to be here Andy.

Take us through the story. I mean, everybody wants to hear the story of Android, but maybe could you just give us a few minutes on how Android got started and how you guys ended up at Google?

Sure. Yeah, it’s actually a story that Eric Schmidt has been telling me to put into a book for a long time because it’s not all that well known. In fact, a lot of people don’t even know that Android started as a start-up.
There were four co-founders, Andy Rubin was the lead co-founder and I had met Andy this… What I did at Google doing a start-up, being acquired by a large company, starting a venture fund, in this case, GV; it’s not the first time I’ve done that. My first start-up Wildfire Communications, a voice-based personal assistant, which we started 30 years ago, a little before its time got acquired by Orange, the European mobile phone company. I started a venture fund for Orange and one of the companies that we invested in was Andy’s company, Danger, and so I got to know Andy in that start-up.
And then when he was thinking of building an OS originally an OS for cameras, I started chatting with him. He knew it really needed to be an OS for phones. We were really early on in building that mobile phone OS and just, we hadn’t even raised any money. We were just out talking to different VC’s and also to some of the large tech companies, because we thought if… We knew that this was going to be the platform of the future, the compute platform of the future, these mobile devices. We thought if that was the case, it’s not only getting venture money, but it’s talking to the large companies that are going to be the drivers of data information flowing to computers in the future.
We were talking to Yahoo and Google, basically saying if we build an open mobile phone platform, could you imagine Yahoo or Google putting your brand on this device as a mobile data device, as well as a communications device? And it was in that meeting, early meeting at Google; Larry Page because he had known Andy from a Davos thing, was standing at the back of the room. And it was when we said, there’s 800 million phones sold every year, which there were back then, now it’s well over a billion, but 800 million phones sold every year. And it would be bad if Microsoft or some other large company owned that platform in a closed way, we thought it should be open. A light bulb went off in Larry’s head and he really pursued the acquisition.
That’s how we went from start-up to being at Google. And we were nine people. We had just the beginnings of an idea. This was way pre-iPhone, in fact, Apple hadn’t even started the iPhone project at Apple yet. It took us three years to build and announce the platform, another year to actually ship the first mobile phone. So it was a long haul and now, thanks to the investment that Google has put in and right place, right time. It’s basically, I don’t think you could argue, it’s the most successful piece of software or operating system, ever.

Really exceptional must be incredible feeling to be a part of something so powerful and transformative and, and so open.

That’s old news, I’m a part of Tamr now, that’s more exciting.

Tell me more about data and data management in the mobile ecosystem. I’m sure you’ve run across these problems for years. Back at Vertica, we did a bunch of… one of our core vertical was Telcos and doing called decal reporting. Data and data management has been a part of the mobile ecosystem forever. Where do you think the most interesting opportunities are for data in the mobile ecosystem?

Well, I mean, it’s a very broad term and this is when you start talking about data in any one discipline, you realize, wait, I mean, if you talk about mobile, like every one of these mobile phones is emitting signals back to the cell network. Literally, you can start with data just in terms of how you optimize the network. I mean, carriers started doing this a long time ago, just looking at each signal, each signal connection where things drop. How do we optimize our network? Oh, turns out if you look at the data, you can actually see where you don’t have good network coverage, because you don’t have to have people calling you saying, people can’t hear me, you’re just seeing calls dropping and handoff.
There’s data in the network and transmission network of these devices that can be totally, real-time used to gather, optimize and tune. There’s the biometric data that every user has on this device that as you start to see, can be used for health and biometrics and knowing, if somebody is elderly and they have one of these in their pocket, you can know when they fall, number of falls somebody is taking as they’re elderly, as an onsite of dementia and other illnesses, payment, mobile payment in the mobile space and all of the information that you have from that.
And that also then starts to say, well, there’s privacy issues and how do you do this? And of course with some of the biometric data that I was talking about, Google and others have started to pioneer ways of keeping some of those fingerprints and information on the device, but improving models for machine learning that then talk back to the network in ways that you can get signaled, but without transmitting a lot of personal data off of the devices, but, it’s endless.
So, you want to ask a more specific question?

Well, it sort of takes us to Tamr. At Tamr, we believe there’s so much data there that you can’t possibly do it all, organize it all with human beings and you have to use the machine. Tell us more about your inspiration to invest in Tamr early on, literally the first investor. What inspired you? As Tamrs’ grown, how have you seen it evolve?

Yeah, so starting with that example of… I was probably only 10 or 15% into talking about all of the data streams and information that’s being acquired by these devices. And by the way, some of that information is incredibly fine grain. Like if I want to know, this is my WHOOP device on my wrist, which also talks to my phone. It’s constantly measuring every beat of my heart and every interval between every beat of my heart. And so, it just goes to speak of, there’s just fire hoses of data everywhere. Then, as you said, way more than anybody can sift through and then the real interesting stuff happens is not only can’t you sift through it, you can’t see patterns and you can’t see correlations. And it’s the correlations between all of these different data streams that let you start to see, not just patterns in that stream, but broader patterns that start to be indicative of major shifts or things you should be making decisions around.
And to do that, you need to, not only capture these extremes and be able to store them, you need to be able to figure out what of that information you need to store in long-term versus pull signal out of it, and then you don’t need to store. And then you really need to understand how to do the unification of that data and tagging of it in such a way that when you want to start correlating it to other bits, you can see things that are similar or the same, know that they’re the same, so that you can start, again, [inaudible 00:09:04] drawing conclusions or being able to really ask questions of the data.
And to me, that was the light bulb that went off in my head when we started talking. I think there were a couple of things. One, there’s just way too much data that every company is dealing with to be able to handle it. But that’s like saying, you’re the CFO and I’ve got too many pennies flowing into my company to be able to handle it. That’s not an acceptable answer for the CFO. They have to collect those pennies and know where to invest, store them, leverage them.
A CIO can’t say I’ve got too much data flowing, and I can’t deal with it. That’s valuable information for the company and the answer is no, you have to be able to organize it, clean it up, unify it and then ideally be able to ask questions. The executives, the board, senior leadership needs to be able to ask questions of that data and get insights. And those insights can only come if you have some sort of way of cleaning up unifying and what I… caught onto me Borg defying that data. The Borg in Star Trek would unify and assimilate, and to me, that’s the key thing, is unifying that data in a way that the company can ask information of it and ask questions of it.

Well, I think my marketing people are going to have hard time with Borg as a [inaudible 00:10:34], but I like it.

We had the Droid as our first phone with… That was really leapt to success with Android. And that was, they managed to spin that into a positive campaign.

Any Star Trek reference is good one in my mind. One of the things that I know we connected on really early as we were starting Tamr was when we met with the team at Google Knowledge Graph, and they had very similar patterns inside of Knowledge Graph that they were using to crawl and unify tabular data on the web to be integrated into Knowledge Graph.
And over the last seven years, it’s been amazing to work with Google and, thanks to you and everybody on the GV team for bringing us in. But now it seems like some of these behavioral issues around data, what Google [inaudible 00:11:27], or how other companies can and should [inaudible 00:11:30] treat data as an asset.
Could you talk a little bit about how people at Google interact with data and how data is a part of the culture and data as an asset, is a part of the core culture at Google? Because I think a lot of our customers who our traditional businesses really struggled with how to change human behavior, to be a bit more like Google.

I think certainly, Google data… both data organization, organize the world’s information. The organization of that data is one thing, but Google’s also been really good since day one at using data to answer important questions. Those are two separate things, but they do come together, from Google’s standpoint, early on, big table… The investment in large storage infrastructure, to just realize that the amount of data that we were going to start needing to accumulate and organize was going to dwarf anybody’s understanding of what that was before.
If that’s your mission day one to say, I’m going to organize the world’s information. You have to step back and when Larry and Sergei would say something like that, they were serious about it. They would want serious answers about it. So figuring out just the storage infrastructure for doing that and then the way of setting up structures for relationships.
Now, I would say for the most part, Google built itself on data relationships that were much simpler than most of your companies are dealing with. We have web links, the content that they were pointing to, the concepts that were in the knowledge graph that related to those things, that kind of semantic structure was pretty simple in comparison to a lot of the business decisions a lot of companies need to make when just starting to get to the worlds of parts and suppliers and transactions and customers. And as soon as you start going a few levels down, those problems get a lot more complicated.
So with Google organizing was big, large amounts of data was big and then getting analytics off of that data. And again, Google was very early on with our Google analytics and being able to not only use that ourselves, make them available for other people too in their web systems for measurement, but measuring everything and being able to get and record the data, to be able to measure everything as part of its decision-making was sort of key.
So again, I think that brings me back to the things that I saw on Tamr. Now, again, the thing that was with Google was you could automate most of what Google was doing because it was Greenfield.
We were acquiring and bringing that data in. We knew where it was coming from and we could organize it. I think what most people hit as a challenge when they say, well, let’s just do what Google did is they realize their data sources aren’t as neat and orderly and clean.
They have legacy of history associated with them. They’re coming from the result of an acquisition of three different companies or whatever. And that’s where I think the brilliance of Tamr and getting that line right, of pre-Tamr people. Like I said, you either need 300 people organizing the data and then we’ll just make 300 people 3000. At the end it just doesn’t scale with humans or we’ll just use machine learning and we’ll just have the machine organize the data. And course that has huge blind spots as well. So, getting that balance right that Tamr did of realizing, Oh, wait, we can have the machine organized 90% of the data, 90, even 96% of the data. But that last 4% is key. And so that’s where we’re going to be really smart about how we engage people to help get that last 4% figured out of, is this field the same as that field?
Is BT blood type? I don’t know, no one really knows. It looks like it could be blood data so we’re going to give it some confidence, but it’s just going to take, five seconds for somebody to click a yes button. And now that data is much more clean, you know it with certainty. And so, in the terms of building a master of where truth is, you’ve got the signal of the best machine learning combined with the disambiguation of people, done as lightweight as possible. Now to me that was huge.

That’s awesome. Well said. I hate to do this, but I want to go back to some of your earlier start-up experiences. Can you tell us about Wildfire as a company and you guys were way ahead of the game with regards to voice assistance and how do you view Wildfire now in the context of all of the Google and Amazon and all the Apple products that are out there to do voice? What’s your view on that now?

Sure. And just to disambiguate that one, because I’m a human can do it. Wildfire is a common term and there been other start-ups Google in fact, acquired a start-up called Wild Heart, which isn’t the one we’re talking about.
The one we’re talking about Bill Warner, who also founded a start-up in the Boston area called Avid Technology, which built the first video editing company in which I helped Bill start Avid. He just… When mobile phones first started coming out, and this was in the late eighties. Bill had one in his car and he was just worried he was going to get into an accident because talking on the phone and driving and whatever, and he just had this idea, can’t you use voice to organize and navigate and organize information. And that was the core premise of what Bill had with Wildfire.
And we started as you do in start-ups. And what does that mean? What can we actually build? What would you do? So we incorporated in 1990, so 30 years ago and with this idea that we could build a voice-based assistant. One of the co-founders, Tony Lovell, there were four co-founders. Tony was just a visionary for user interface.
We were talking about, we should build a voice control and information management system. And Tony was like, “No, we should build a secretary.” You could say the word secretary back then. He was like, “We should just build a virtual person sort of like Mabel that, you crank the phone and say, Hey, Mabel, I need to talk to Doc Brown.” Mabel would know that Doc Brown was a calf over at the Smith’s house and she’d just connect you directly to the Smith’s house.
We need both intelligence and ease of control with voice built into it. And so Tony brought Wildfire to life in the personality of the product And it’s so funny hearing and interacting with voice-based assistants today because we have what felt like continuous speech though, it was a bit of a trick. It wasn’t really continuous, but it felt like natural dialogue. We had humor and jokes if you called… Wildfire was built into the phone network and she was just that, a voice assistant to help you with your communications need. I could say, “Call Andy Palmer.” If you called me and left me a message, I’d dial in and she’d say, you’d have three new messages. The first is from Andy Palmer. I could say, “What’s it say? Throw it away, give them a call.”
You could navigate to what Bill had in his original vision. You could navigate your voice communications in the management of your messages and contacts, all by voice. And so throw it away and give them a call. Next item, create a context, schedule a meeting. If you called her after 10:00 PM at night, she’d yawn a little bit. You said to her that you were depressed, she’d say, “You’re depressed. I live in a box.”
We had, just like assistants of today, we had a bit of a sense of humor. But we were potentially 20 years too early. I mean, there are other start-ups that hit, but the user interface… Wildfires’ user interface today could sit right next to Google home or an Amazon Alexa and feel just as fresh and fun and interactive as those assistants with a much more limited surface area, because we just did this one vertical problem and crushed it.
But anytime you say, “Hey, Google” to wake up a voice-based assistant, that’s a patent that I have from 1994.

Amazing, amazing. So, this idea of the design principles that you guys used, had integrity to them that’s carried through since there’s… The current systems look and feel very similar to Wildfire. Can you talk to me a little bit about design and I know design is really important to you and it’s always been important and becoming more important and central to everything we do in technology? Can you tell me a little bit more about why design is important to you?

Yeah, sure. Yeah. And I could do a riff on aesthetic design as well, but I think you’re talking about systems design, right?


I’m a platform person. I really dislike when people build something one-off when you could build a platform to solve that and have that solution be one of what becomes many solutions to a problem. And we built this in the case of Wildfire. Even back then, we were very careful to extract all the elements of the platform. It was a voice assistant for communications, but it was really easy for us to prototype weather and sports and other kinds of apps, because it had that platform nature to it.
Android, of course, the whole idea was platform. And from day one, it wasn’t meant to be a platform to build a voice dialer and an SMS app. It was meant to be a platform to build whatever… we didn’t know. And who would have thought an app like maps would have been the thing that took off?
And building any platform, I think key to some of the questions you need to ask is, what is the vision like? What’s the vision and what’s this thing for? Okay, so you’re building a phone and you want to have a dialer. You need to, at least, even if you can’t imagine maps is going to be the killer app, you need to imagine what does a portfolio of things that are leveraging this platform, look at the future. And what are the usage patterns five years out, three years out? And the reason you want to think about that is, it starts to inform you on what kind of APIs you produce, what kind of scale you might need to be able to respond to? Where there could be potential bottlenecks that you need to worry about, that you want to avoid.
And so, it’s key to come up with the right abstractions and those abstractions are abstractions of data and data encapsulations. Those abstractions are for functional elements. Those abstractions are for services that you want to expose and provide at higher levels. And again, those abstractions are to imagine, not just how to deliver on this functionality that we’re talking about today, but where the functionality is going?
And I have arguments with even my engineering peers and definitely arguments with product people, because I have this assertion whenever I hear the term technical debt, there’s always this assumption, some engineer messed up and some engineer didn’t design this thing for what we needed to be doing today. And I think that’s unfair to engineers. I think most engineers try and build the best thing they can build.
What I usually find out is, it’s a failure of requirements. It’s that it wasn’t that the engineer didn’t build something right. He was told to build exactly this, and you’ve got this much time, and you’ve got this much memory and whatever, and they built the best thing they could do for that. Then some product person comes in and says, “Hey, we need to start doing that, too.” And you’re like, “Dude, if you told me that when I was building…” Anyways. So, thinking ahead and getting the product and engineers to work together, to think ahead is important, in that whole process.

Well, it’s amazing. To bring it back to Tamr and data stuff, we’re getting this stage in the evolution of enterprise data ecosystems, where they’re starting to embrace this next level of requirements that views the data problem not as, whatever’s required to get any given system up and running and some process automated, but viewing data more holistically across the entire enterprise.
And one of the role models that we like to use when we talk to people about data ops is the Android ecosystem; and how open the Android ecosystem is, and really encourage our customers to think of their data ecosystem as a much more open set of best of breed products, where lots of participants can play and best of breed wins over time. And we really do view the model that people should use, is Android as the primary reference there. And the open ecosystem is a part of what is enabled Android to become the world’s largest, most successful operating system, right?

Oh yeah, no, absolutely. And again, this was the thought day one. People talk a lot about Apple and iOS, but they… It’s still one hardware OEM with one OS and it’s a brilliant OEM, probably one of the most valuable companies in the world and they’re brilliant devices and platforms, but there’s a reason why they only have the 15% of the market worldwide that they have, which again, a lot of people in the US don’t realize that and why Android has the other 85%.
It’s because with that openness and with those partnerships, and by embracing those partnerships, we not only have that plethora of apps on one side, but we have Huawei and Samsung and LG and ASIS and Lenovo and all of these other OEMs that have embraced the platform and are building and adding their own contribution to it. So they get the benefit of the core, but then they can innovate around the edges and they can benefit from a large ecosystem of not just the suppliers on the OEM side, but all the service providers that are providing other services and software bits and pieces that plug into that as well.
I know that’s core to the vision that we have to Tamr is to build that ecosystem of platform and customers, but also other partners in between that are building components and elements and doing integration on top of Tamr as well.

Yeah. That’s exactly the way we see our ecosystem. And it’s a part of, I think, why it’s been so great to have you as a part of the team at Tamr from the beginning, because that open, best of breed ecosystem exudes what we try and do with our customers.
And it’s a bit of a contrast, a lot of these traditional enterprises, oftentimes that were used selling their souls to Oracle or IBM or now Palentir and gave up the responsibility of figuring out how to optimize that for themselves. But building an ecosystem is hard though. It took you guys, decades to do this for Android. We feel like we’re in the very, the first stages of building out a similar ecosystem for data ops. It’s a real challenge, much harder, isn’t it to build an open ecosystem and a best of breed thing?

It is. But, the benefit that we had at Android was it was a conscious decision from day one that that’s how we were going to build the platform and do it with those partnerships. And we started the open handset alliance and we’re doing that. And it’s in the DNA of Tamr from day one as well. Where it’s hard is when it’s sort of artificial, it’s forced because of customers demanding it or it’s like, “Oh, wait, our competitor announced this. We should do that too.” And it doesn’t come naturally. I think it’s a lot… I don’t want to make it sound like it’s not work, but it’s a lot easier when it’s intentional and when it’s intentional from day one.

Yeah. That’s awesome. I know it’s hard for you to talk about the stuff you’re currently doing. I understand you kind of moved, shifted your focus from GV over to another project at Google. Can you tell us anything about that project and what you’re doing?

Yeah, not so much. It does combine a lot of the things that I’ve worked on in the past, but focused on… I’m focused on mostly on education, but also on disrupting how people interact with machines. And I think one of the struggles with ed-tech is that it’s really hard to adopt the technology.
Google with its Google Classroom product has been really good at figuring out how to build something that’s easily integrated into the teacher’s workflow. And, similarly, I just look at how you can integrate into the student’s workflow and really provide them assistance at any moment that they need it. But I can’t say too much more than that.

Well, thanks for sharing. I know for me: a lot of people talk about the potential danger of devices for kids. And it gives me hope that you’re working away on empowering our kids with devices in all these great ways. And that there’s lots of good that can be done with these devices, put in the hands of our children. And I’m sure as you did with Wildfire and Android and so many great companies, including Tamr, if you touch it, I’m sure it’s going to benefit society. Thank you, Rich, and thanks for taking the time today. Really appreciate you spending time with us and appreciate your thoughts and help with Tamr.

I love working with you guys, Andy, and thanks for the chat. I enjoyed it quite a bit.