37 Data Democratization
We are joined by [Clark Richey](https://www.linkedin.com/in/clark-richey-6b737/), Chief Technology Officer at FactGem to talk about how organizations handle data, how it might be made useful for a wider audience, and about their experience in this field.<br><br>### Resources and Shoutouts:<br>- [FactGem](http://factgem.com/){:target="_blank"}<br>- [Data Architecture Summit (Chicago)](https://das2019.dataversity.net/){:target="_blank"}<br>- [O'Reilly Strata Data Conference (New York)](https://conferences.oreilly.com/strata){:target="_blank"}<br><br>##### Music Credit: [Tumbleweeds by Monkey Warhol](http://freemusicarchive.org/music/Monkey_Warhol/Lonely_Hearts_Challenge/Monkey_Warhol_-_Tumbleweeds){:target="_blank"}
Top Keywords
- data 0.013
- business 0.007
- technology 0.006
- neural 0.006
- problem 0.005
- graph 0.005
- neural network 0.005
- database 0.004
- understand 0.004
- engineering 0.004
- peanut 0.004
- peanut butter 0.004
Transcript
Hello. I'm Ryan Cook, and this is Civic Tech Chat, a podcast about the civic technology movement. We seek to harness the power technology has to improve the delivery of public services to people everywhere. Clark, thank you so much for joining us here on Civic Tech Chat this time. Could you give us an introduction and tell us a little bit about what you do?
Yeah. Absolutely. Thank you. So I'm Clark Ritchie. I cofounded and I'm the CTO of Fact Gem. For about seven years ago, we started that, company along with my cofounder, Megan Kwame. Prior to that, I ran public sector sales engineering for MarkLogic, a database company headquartered at Silicon Valley. And really for the majority of my career, I've worked in the Department of Defense and intelligence industry sectors, as a contractor in one form or another, which gave me the ability to really work on some of the largest and hardest problems, that exist today. So that was a lot of fun.
And one of the things we like to start with on Civic Tech Chat is this idea of personal why. So what would that be for you? You know, what is the thing that drives you to get out of bed each day and do what you do?
Yeah. It's a great question. So, in my time in defense at Intel, as as I said, I I got to work with some, really hard challenges, lots of data, interesting people. It's also great technologies. And then moving into, more of the product side, working for MarkLogic running sales engineering, again, got to see really new and innovative technologies being rolled out to people. But what struck me was that almost all of these big data technologies are being geared to software engineers. So we're giving software engineers lots of great tools, which is important, certainly. But I always felt like if we work just a little bit harder on the product side, we could start to deliver these capabilities to non engineers. And if we could do that, we could enable businesses to solve problems much faster with a lot less budget because not everybody can afford to have a whole bunch of software engineers on staff. So that's what I'm really interested in is how can we deliver those capabilities quickly directly to businesses.
Your current role as a your chief technology officer, I imagine there is a some bit of path that leads up to getting to a position like that. Could you tell us a little bit about your path and how that kinda led to leading an organization like yours?
Yeah. So, you know, I I started out like, most people. In my case, like, I came directly out of the military where I had some experience, sort of as a second or or tertiary responsibility doing some IT. But then, you know, going, as I was finishing up my degree in in software engineering, going starting out at come at the bottom doing websites like most people and then, you know, some server side programming, and in in fairly small teams, again, in in defense at Intel, but then discovering, you know, I had an aptitude for working directly with customers, helping to understand their problems so we could build, better software. And, one of the things I found at least in in software engineering is while it's certainly important to be good at the engineering side, if you can communicate with customers, understand their problems, and then translate that, that's really important. So that enabled me to, take on a lot of team leadership roles on different projects, you know, and so do that in the consulting space for a number of years. And then some opportunities, arose where I was able to go into some presales, engineering, originally with, BEA before it was bought by Oracle, so the original WebLogic platform and all of that. And then over to MarkLogic, where it's been about five years, again, now teaching people to do these types of things, work with technology, but really focus on understanding the problem and and translating that and helping people understand technology. And I I I I think that bridge is really important for people. And as a CTO, yes, I I developed and created a lot of technology that FactGerm runs on, but really so much of that is being able to explain to potential customers, to prospects, to existing customers, to other engineers, and even internally, what is that vision, what is the market need, what are people saying, so that everyone understands, really what's going
on. As a follow on to that, I I imagine one thing you probably experienced is a change in the ratio of time you spend, like, really dug into the code and then shifting over to that communication with customers and whatnot you're talking about there. What what was your experience like going through that transitory period?
That can be pretty dramatic. There are times even now where I might not get to write any code for a week, and I I start to get a little jittery, honestly. But it's important and it's necessary. And I think sometimes, particularly engineers, you feel like if you're not writing code, you're not contributing, and and that's not true. If you're still helping with marketing or you're talking to customers, you're making sales calls, or you're, even just spending time kind of staring at your whiteboard thinking about what does the future look like ninety days out, that's still contributing. It's just a different way of contributing than what a lot of people are used to, and and you can't count lines of code or numbers of bugs fixed or numbers of tickets closed, but you're still contributing as part of the team. And, it's something I still struggle with sometimes, and you and you need a change of pace I find. I I still love to write a lot of code, but it it can certainly be a bit of a challenge.
Is there any media, whether it's a podcast, print material, video, or some other such thing that you found has been especially informative or inspirational to you as you've worked in your practice?
Yeah. I mean, mean, generally speaking, just because of my sort of my schedule and the way I do things, I I like to listen to podcasts. I like to go to conferences, though, honestly, still. And and for me, the reason is, again, as as a CTO and someone who's juggling code and and and a 100 other things, it can be difficult to find time during a regular day to read a blog or even listen to a podcast. But if I can block out two days, or sometimes I'm lucky, like, four days to say I'm going to a conference, I'm gonna be immersed there and listen to people, that brings about a certain focus. Now I know that's that's hard in certain organizations and certain people, their cost factors and things. So in those situations, I recommend trying even just to block out days. So sometimes there's a lunch and learn where organizations will block out, you know, forty five minutes or an hour, but actually trying to schedule those times, put them on your calendar and block them and say, hey. For this hour, I'm gonna go and read about this interesting thing or I'm going to, you know, watch this tutorial because otherwise, it's just too hard to schedule and something that appears more pressing, because it's tactically urgent, comes in. But strategically, you've gotta keep on top of these things.
And speaking of conferences, have there been any you've been to that kinda stick out in your mind as, like, you had a particularly positive experience?
So just recently, I was, attending and speaking at the, data architecture summit in Graphoreum conference in Chicago. That was really interesting for me because it it brought two big communities together, data architects and and graph. There's a big overlap there. So I saw some really interesting things come out of that. Generally speaking, I've attended the last couple of O'Reilly Strata Data Conferences in New York. Those are always interesting. You get a wide variety of of speakers and topics there, and I think in particular, the in this the non promotional industry keynotes are always really interesting. You get really deep thinkers in a variety of of areas not necessarily obviously related to computer science, but are really working in the field of data and analytics and, I enjoy a lot of that because I think it it it gets you shifting get gets your thought shifted away from just algorithms implementation to larger concepts.
Shifting now to our our main topic of this episode, this idea that I believe you alluded to in your answer about personal why. It's idea of trying to democratize the way data can be used as a tool. Could you give us an idea of the the high level of what that concept is to you, that idea of democratizing data?
Yeah. So, in in a lot of ways and and I don't, you know, coming out of, you know, IT, I don't in any way mean this in in a negative way, but our IT staff organizations have become essentially the high priests of our data. So all of our organizations and our business are are run by data, but we don't really get direct access to it. As a business, we we explain sort of the problem we wanna solve or the questions we wanna ask. And people in IT, they they listen and they take notes and they go away for a while and and they do something kind of magical, and they come back and they deliver something. And then and I've been in so many meetings. There's always sort of this mode of pause, and everyone kinda looks at each other and sort of shrugs, like, yeah. I guess that's sort of what we ask, and they go and they play with the deliver deliverable for a while. And that question is not quite right, and you iterate on it. And I think the biggest challenge there again is there are different worlds and different vocabularies. Right? So when a business describes a problem, typically, they go to a whiteboard, they draw some circles and lines, and they say, here's our problem, and here's how we think about things we wanna get from a to b. But in the translation of that, not due to any malicious intent or anything, but due to the tools that as technologists we've traditionally been using, they don't map directly to that. There's a there's a cognitive disconnect and a and a gap and so the things get translated, to this again, this this jargon is really only spoken by these high priests of data and then we present this back and go, I took your picture and I made this really big diagram and I cover the table with it. Is this what you meant? And no one in that room can validate that. So if he shakes the head. So I we wanna remove that gap. We're gonna say, look, if you can draw that picture, this is how you as the people who understand the business and understand the data, think about it, then that can be the model. And if you can do that, right, if we can separate that gap between how the business reasons about the data flows and how it actually lives in our systems, you're reducing friction, between the teams, between the actual people using the software, and then you can get more done. You're still certainly gonna need IT support. If you wanna go to a petabyte of of data, for example, that's gonna require an infrastructure support. But you're really enabling the business to get more connected to it, which is going to enable IT to spend more time doing other things, helping in in larger projects, more complicated things. And you're gonna build, I think, a a closer bond between just people and the data, but those two teams as well. So, again, really just letting the people who rely on the data have more say in how that actually works and and really be able to establish those clear lines of communication between both parties.
I imagine the the parties you've described there that they, at times run into situations where it's like, hey. We've decided to start collecting information about a thing, and that information then builds over time. As you mentioned, like, maybe even as large as a petabyte. What sort of problems have you seen in your experience kinda pop up that need to be wrangled as organizations get larger and larger sets of information to evaluate?
Yeah. That that's that's an excellent point. And I think it's it starts half a step back. There's sometimes when, analysts or people from the kinds of business, they have a question that you can't currently ask in the IT system, and they go to IT and say, I you know, I think this might be interesting. I think it has value. It's not gonna change the business, but it's value. And they explain it to IT, and I think it's like, well, because you need us to get involved. You gotta translate your picture and implement it. And that could be three months or so and and a lot that's too much money. It's it's it's it's for another so those things never even happen. Those things just get cut off initially, and that's the problem. Then you should get that sort of medium scale, and you and you can do that. But, again, in a lot of our traditional technologies, the cost of change is high. If you start taking creating a data warehouse, for example, a lot of times that initial cost and effort is three, four, five months just for data modeling and validation because you're forced to look at not just what questions you wanna answer right now, but what might you wanna answer in a year, two years because changing the schema of that data warehouse is expensive and hard and requires engineering and a lot of support. And that just adds cost and time and reduces your flexibility. So anytime you have a technology, in my opinion, where you can start small, demonstrate, return on investment for the business, and then rapidly build upon it and grow that, that's the best possible scenario. Maybe it maybe it doesn't generate anything, fine. You stop early. But that flexibility is just really important to let the IT evolve at the pace of the business.
As folks try to do that or as folks try to answer questions with with data, as you mentioned, what sort of techniques do you typically see folks implement to to try to, like, reach in and and get those nuggets?
Yeah. So in terms of, like, the the, analytics of course, data science is is the hot thing. Everyone wants to be a data scientist, hire a data scientist, data to data scientist, whatever it might be. So in that community again, R is still very big. People are really moving out to tools like Jupyter Notebook, a lot a lot a lot of Python technologies. And those are all great. They're fantastic. If you look at sort of more the cutting edge I mean, real cutting edge was happening at at some larger companies doing research as well as universities. They're doing research on reimplementing older algorithms. And what I mean specifically is, for example, people talk a lot now about, oh, we're gonna we can do some AI machine learning. Well, what kind? Well, maybe we'll build a neural network. So if you think about neural network, you you are you immediately get pictures at least I do of, like, a brain and and the neural pathways. That's actually not what any traditional neural net neural network algorithms look like right now. They're not, graph like that. They're very, you know, either columnar data or tabular data. Why? Because that's the what the technology was many, many, many years ago when those algorithms were developed. And now researchers are realizing, hey. We have the technology to actually represent the data, in a way that looks like that picture of neurons in the brain that looks like a neural network, which is actually what we're trying to model with the algorithm. And so now, like, graph based neural networks are becoming a very big, research point, and and research are finding a lot of, really interesting areas and improvements, by moving in in that area. And I think you're gonna see a lot more of that where data scientists are using these newer, more efficient, more powerful data structures for doing, analytics.
For the listener that might not be quite as aware of the landscape, I I think you mentioned this, like, a graph based neural network. Could could you speak to kinda what the benefit of that kind of, achieving that that sort of model gives us?
Sure. So, graph databases, in particular, property, or labeled property graph databases are fairly new. They were actually invented roughly nine or so years ago by a gentleman named, Emil Ephrem, who's actually the CEO and cofounder of, Neo four j. And, the idea is it's very much like, again, like, if you draw that map of your business on a whiteboard, it's circles and lines connected to each other and think of properties in the circle. So, you know, you might draw a circle that's a person and inside that circle, you write down their first name and their last name and their age, and then you draw a line that says, knows other person. And so once you connect these things, you can really do network analysis. So people so if you think of things like Facebook and so forth where you're trying to understand who knows who and and who's liking what posts and how does that impact things. Well ironically Facebook isn't using graph technology for that. But that's the kind of thing you think about when you when you look at this is social network analysis. The ability to understand how things relate to each other even several degrees of distance away, is really very powerfully enabled by this and it's it's something that's very very hard to do if not impossible in other technologies. So when you're trying to do, artificial intelligence, machine learning, and understand what is affecting my customers? Like, why are they doing this? What are they going to do? Or, I'm in health care, and I'm trying to understand, you know, how can I achieve a positive outcome for my patients or, you know, keep health care costs low? What things are actually affecting that? Those are often secondary, tertiary, or further out effects, so you need to be able to look at that holistic picture. Fraud is another great example of that as well.
The space we're talking about, I think, often evokes, some amount of futurism by its nature, especially since I should talk about there's, like, things that are very much on the cutting edge. Is there something in this space that you see as be, like, most exciting or thing that lists the most hope in you, looking forward?
Yeah. You're you're right. It it it is kind of a scary space in some ways. There's lots of talk about, you know, Mark Zuckerberg's up on the hill and what's happening with social media and how is that being used and how does it even affect, you know, our democracy in terms of people and bots influencing other people? So that can be a little scary. I think on the positive side from so the ethical side of you and things, there are now a number of people, a number of of really deep thinkers who are looking at and studying what are the legal and ethical ramifications of a lot of these technologies, and and how should we be thinking about them? You know, what what are the right things, you know, to do? You know, so to paraphrase, Jeff Goldblum, of of course, from, you know, Jurassic Park, you know, so you have to stop and think, you know, just because I can do it, you know, should we do it? And people start to tackle that now, which is which is great. On the technology side, I'm obviously very excited about, the use of graph databases and and as that technology matures. In particular, what excites me there is people are starting to look at taking some of these machine learning concepts and embedding them directly in the database. So instead of having to have data that sits in my database, then I either extract some out, do machine learning, or I have to run complicated algorithms that query the database in place and give me some, result. If I can essentially put some of those algorithms in place into the database so as the database itself updates, those are constantly evolving. In place, I can do much more interesting things at scale. And that is something now that is actually being looked at by a number of companies and and some have started to do some work in that area in in production.
Earlier in our conversation, you you talked a little bit about this idea of that translation layer between folks on kind of the business side of things, folks on the more technical side of things. Can we dig into that, a bit more? Like, in I guess, in an ideal world, what what should that interaction look like from your perspective?
Yeah. I mean, in an ideal world, I'm a big fan of of small integrated teams. Right? So, you know, I I want to see a a team being created to solve a particular problem. You know, not not a a standing team, but, you know, the business comes and says, hey. We wanna look at this problem. Okay. Let's get someone let's get a business analyst from that side who maybe really deeply understands it, and a domain expert. Let's get someone from IT who understands maybe the aspects of how we deal with, our data systems or website, whatever the whatever the technology might be, and and embed them as part of the same team. Right? I think part of our problem and this is true in a lot of ways. So many of our technology problems, I think, aren't technology problems. They're organizational people problems. Right? And as organizations grow, we sell. Well, we have the business department, and over here is the IT department. And there really becomes that gap. If it's serious, you know, we're just gonna have functional teams, and we take people with enough skill sets, and we embed them together so that everyone is just driving toward that same goal, which in theory they are anyway, but we remove those functional barriers. You know, that can can certainly help tremendously. Obviously, that's more of a social engineering way of dealing with it. Again, and and from technology, like anything else, we have to, I think, find, what is the simplest way of talking about the subject? Right? So, again, traditionally, if we were gonna solve a problem together to to to meet your business need and you describe the data to me, I said, well, I'm gonna put this in a relational database, and what I've gotta do, Ryan, is I gotta put this in third normal form, and I think I've lost you. We're not talking about the same thing anymore. And if you start going deep into some business jargon, you know, about, you know, value ratios, like, you you've lost me. So we we've gotta find a a common language that everyone can understand and connect with. And I think a lot of it if we can do so a lot of it visually, that helps tremendously. So be able to see, like, on a whiteboard, these are how things connect. These are the rules about, how the system treats information when it comes in. It it just really helps get everyone on the same page and and reasoning about the same thing.
You yourself, you come at this from, you know, a computer science trained background. How would you say that background has informed and shaped your view and and how you look at these sort of problems?
So I I've, and and so I I I started doing cosignancy, obviously, many many years ago in in grade school. And I was fortunate enough to, in grade school, be taught to do assembly language. And I talked to it before anyone told me that there were, programs doing that. So I would have you have to put in the codes into memory and things. But what was great about that is I think is a foundation and that and that's, for me, is built on through there is understanding how to take a complex problem and break it down into very small steps. So I've taught computer science at a number of large universities, and particularly I find at the, intro level, that's the biggest barrier for people. And it's not a computer science problem. Right? Because I'll go through an exercise with students and we'll say and it's it's not something that I invented. You'll see this in many places, but, you know, we'll give you a simple task. Like, I want you to explain to me as if I was just a robot who could only do exactly how to go into the kitchen and make a sandwich. And that's really hard for people to do. This you think, well, go into the, you know, go into the kitchen and get the peanut butter. Well, I can't. The fridge isn't open. Oh, no. Oh, take the peanut butter out of the fridge. Okay. Put the peanut butter in the sandwich. There goes a can of peanut butter, you know, on the bread. And, like, no. No. That's not what I meant. Like, well, as humans go, well, I know that's not what you meant, but the computer doesn't know that. And it a lot of that is getting out of our own head. Right? We all of us live inside the context of our own world and what we internally know, and being able to step outside of that to break those problems down for not just computers, but for other people can be really, really hard and very frustrating. And, you know, again, that for me, it's the thing I found most helpful in my career is to be able to, you know, do that. It helpful in writing things. It's helpful in debugging code, and and just talking to other people.
So for someone that is maybe either in the middle of getting over that that hurdle, maybe they've just jumped over that hurdle you mentioned, and they're kinda just getting started out in this field. What advice would you give that person as they're getting going?
So I would I would say, you know, two pieces of of advice. One is, you know, computer science is changing so rapidly. There's so much out there. You can't know it all. There's no way. There's new languages appearing every day. There's new types of programming languages. There's new database technologies. It's a it's a massive amount of things. One thing I see people wanna do is then go, oh, I'm just going to do this narrow thing because this field is too wide and too scary. So we focus on one very specific thing. I think that works in the short run, but you have to have at least a smattering of of knowledge of the other things. So while, for example, you decide what I really love is just is is I have to write interactive web applications. That's fantastic. You know, learn all you can about, your your tool of choice for web pages, whether that's, you know, TypeScript or JavaScript or whatever it might be. You need to know a little bit about databases. You need to know a little bit about network protocols because you're going to get into those conversations. So to take a little bit of time and and things you might learn over in that other area are going to help you think about things a little differently and understand the problem better in your space. So I would say everyone needs a little bit of breath at least, to understand what's happening. And then the second thing I would say is, yeah, communication. Don't don't forget communication. I think it's been something that our universities in the computer science parts have been somewhat negligent about. It's we focus on algorithms and programming, but we neglect, written and verbal communication a lot. And that is a huge part of being successful in this field.
A tradition we have here on on Civic Tech Chat is to leave space, at the end of a conversation here that we have with a guest to allow them to give us an idea of what sort of thoughts we should have as we end our listening to this program. Could you give us an idea of what that would be for you? What sort of concluding thoughts would you have?
Yeah. I think, yeah, a couple of sort of, shorter thoughts. No. One is depending upon your role again, don't be afraid to do things a bit differently. A lot of times in this area, you'll hear people's you'll have a problem and people hear you, oh, the solution to your problem is clearly in this area. This is how we do it. It's in this area. We use this kind of technology. Ask why. A lot of times, the answer becomes, if you dig a little bit deeper is because that's how we do it. Because that's how we did it before. And that's not new to it's not unique to your science. That's you see it in all different types of fields and industries. But be afraid to say be don't be afraid to say, I wanna think of it so differently and try something. I hear what you're saying. I think there's a different way. That's super important. You certainly have to listen to feedback and and and, you know, people more even more experience around you, but trying something different. And, again, I think another area is linked data again. I I think really, while there may never be one type of data storage solves all problems, I don't think that's necessarily realistic. I think we are very much at a crux in the data scene where we are moving away from all the traditional columnar relational type structures into something more. We live in a connected world. Everything we do is influencing everything else. Businesses, health care, finance companies, social media, it affects all of it. It's a connected world. So start looking at those technologies. How can we connect them? How can we move beyond just silos of of data to help our companies understand the broader landscape and and and picture of things?
Look, again, I wanna thank you so much for taking the time out of your day to join us here on Civic Tech Chat and to share with us your insights, your experiences, and and your knowledge there. Thanks, Ryan. It's been a pleasure being on the show. You can follow us on Twitter using the handle at civic tech chat. Visit us on the web at civictech.chat, or subscribe to us for content updates wherever it is you download your podcasts.
