Melody Meckfessel v2: Access, Explore, Save-as | Turn the Lens #23

Cold Open

So I'll just count us down and we will go.

All right. In three, two. 

‍

Jeff Frick:

Hey, welcome back, everybody. Jeff, Frick here coming to you from the home studio for another episode of Turn the Lens. And we're excited for this next episode. This guest has been involved with software productivity at the human point of view for years and years and years, driving huge development teams at Google that she left and now she's doing her own thing. We're excited to welcome and probably from San Francisco, I'm guessing through the magic of the Internet. She's Melody Meckfessel, CEO and Co-founder of Observable. Melody. Great to see you.

Melody Meckfessel: 

Thanks so much for having me. It's great to see you, too Jeff.

Jeff Frick:

Absolutely. So I was looking the last time we talked. Hard to believe it's been almost like two years. How time flies. A lot of a lot of developments of course, in this crazy space on which we choose to play. I just want to get you kind of your observations in terms of how things have changed over the last couple of years. Where are you at with Observable, give us a kind of a quick update on what's happening in dataviz

Melody Meckfessel 

Yeah, it's I think there's a lot of ties back to what we talked about before and we've just seen an acceleration really of demand from software developers and data analysts and data scientists and just really teams and organizations trying to use dataviz faster, make it easier for folks to ask questions, get answers and move quickly. More dataviz forms more great open source options out there. Observable plot being one of them built on top of D3. Pretty widely used data visualization library. So that's been an incredible development in the last two years of being able to create, you know, charts and graphs in minutes instead of having to spend maybe longer time. The other thing that I would highlight is that the modern data stack continues to evolve, but practically customers are trying to find ways. Companies are trying to find ways, like make sense and do a lot of blocking and tackling right with what's happened in in terms of complexity of people's data stacks. How do you navigate that? How do you have visibility into differences and validation and and metrics definitions that maybe you have in one layer of abstraction of your data stack? How is, how are you comparing that to other definitions that are evolving in other parts of your stack? So it's been really fun to work with companies and teams and Observable to do not just like really important blocking and tackling, but bring more speed and ease to how people are doing that. Things like lightweight data apps, you know, we're always going to have dashboards, right? We're always going to need to look at data in this like visual way. But how do you have things that are like really lightweight and interactive and maybe lasts for like a month and then you're on to the next one? Like it's kind of fun. It's like modern app development, but with data.

Jeff Frick

So let's jump into some of the specifics because everybody wants to be more data centric. We're trying to get we're trying to get more people using data and their decision making processes, not just at the data scientist level. And one of the interesting things you've talked about before is kind of this interactivity with the data. And you just asked you just talked about in your answer what's key to, you know, kind of questions and answers and this concept of exploration versus kind of knowing what you want exactly. And then, you know, writing it and getting the answer. But more this interactivity that changes over time as you get more information, and I wonder if you can speak to the power of that and how you see that play out within the community, because that's a very different, it's a very different kind of engagement model.

Melody Meckfessel 

Yeah, that's a great prompt because there are a lot of people out there that do important data work in spreadsheets, it's tabular data, you get to filter, you get to add formulas, you get to add logic, right? And actually business logic through apps, script and and other forms of, of coding and spreadsheets. But there's a ton of folks out there that also know sequel and know how to query their data, their live production data sources. So we have these modes that we're kind of used to working with and take for granted. And what we're seeing is that when you can bridge those, when you can have an interface which is similar to working with tabular data in a spreadsheet, but opt out to sequel, to do something custom, you can move very fast working with your production data. So you have these different modes that we're used to working in, and take for granted, right? And what we're seeing is when you can bridge those, you can have an interface that is similar to working with tabular data in a spreadsheet, but opt out to sql to do something custom. You can move very fast working with your production data. So you have these different modes, these contexts that people are used to working in. I don't remember the last time you were in a spreadsheet just but like.

Jeff Frick

A few minutes ago

Melody Meckfessel 

I remember it was like this morning we were like, I was looking at spreadsheet data. I was also looking at our production data. So, you know, I think these are very relevant ways that people explore and you have a question, you iterate and there's something new. There's like an 'A-ha' that pops up out of that and it doesn't have to be super complex. We're not talking about visualizations that are on, you know, major media news outlet we're talking about. I want to understand like are sales spiking in this particular reason? I can play around with the data and I can look at the products maybe that are that are selling faster than others. Meanwhile, all these very practical questions that people have. So I would say like bringing those spreadsheet like tabular ways of looking at data, fast filtering, search, really fast search, and then having this option to opt out right sequel and then come back in to that spreadsheet like interface that really has lit up people just being able to move very quickly when they're exploring and analyzing their data.

Jeff Frick

Right. And it just really speaks to the way that we learn and discover, right? Because as you continue to get more data and your hypothesis maybe changes, or maybe you just find a different alleyway that you want to that you want to run down and explore that you didn't necessarily know is that interesting? Another thing that you've talked about before is kind of the the analysis of the data versus the presentation of the data and the fact that, you know, the people that are engaged with that data and the output of that data are a lot more than just the data scientists. There's product managers, there's marketing people, they're sales executives, there's executives in the company. And that before there was always this kind of break between, do the analysis, you know, kick out a beautiful PowerPoint and then present it to the team where now you guys are finding ways to bring more people into that iterative process so that everyone can see and manipulate and explore the data all kind of in one single place.

Melody Meckfessel 

Yeah, I, I think it's important to acknowledge anyone, whatever your tool that you're working in, when you're working with data, it can be really chaotic, right? As we just talked about, you don't, you have these questions, your questions of all of your answers evolve. That process is it's chaotic, it's messy, It's iterative, right? So when people are coming together and they're working quickly between this kind of data analysis and manipulation with a visualization and can loop through pretty quickly, you kind of get to the 'A-ha' faster. But then like that 'A-ha' is just, it's not just for you and maybe a colleague or two to talk about. You want to share that right. You want to get input. And so the fact that you can add interactivity to that dataviz and have the exec or the, you know, the product manager or the marketing manager then iterating with you on that dataviz, and asking questions together really opens up bringing the skills that your organization has into that process. And it's, you know, collaboration is table stakes Jeff, like what what we're trying to do is deepen the conversation that happens around data, that happens when people work together, but it happens when people work together and they have the tools that's aligned with their experience. So we see a lot of examples of small teams iterating on data quickly, building that dataviz and then sharing it in their organization, but sharing in a way that lets people explore the data through that visualization. And they do that through search and through sliders and through prompts and everyone's part of it, right? I think we're, we don't live in a world anymore where any sort of business decision maker, line of business owner is not in the weeds with their data. Right. Like the urgency in the market, the importance of looking at the importance of looking at the freshest data, the freshest set of analysis is, yeah, that's just part of the world that we're in. And I think we want a way for all of the skills in an organization to be brought to bear on some of these important questions and the insights that companies have around their data.

‍

Jeff Frick

It's great because it's the ability to double click right is to double click and explore a particular area of interest that you want to to go down that maybe the original person didn't think was of that interest. I think it's it's such a different model in getting a summary report of things that happened in the past versus, you know, kind of all the data rolled that that's happening now that you can that you can make decisions in a very different time frame.

Melody Meckfessel 

Yeah, I mean there's this there's this word transparency. You know, like I think it's one of these big words. It's like collaboration right, it's one of these big words to say. I think what it means when people like when people are looking at the data together, when they are looking at the analysis together, when they are looking and exploring the dataviz together, that's a depth that doesn't happen when you're looking at like a pre-canned chart or a pre, like a template into a pre-canned chart with lots of limits around it. Because when people come together in this way where they're able to interact through the visualization, they keep going, right? They keep going. And that's like that's the depth that businesses need to get to to have that kind of strategic advantage. So yeah, I think that's what we're seeing in the market. You know, in two years, much more urgency around everyone in your organization being able to do this data work or do some dimension of it that applies to the expertise that they're bringing into the company.

Jeff Frick

Right. Or at least some literacy around it. Right. Even if. At least to start thinking in data centric way, start to ask data centric questions, even if you're not necessarily, you know, starting that way. The other piece I want to explore a little bit and I don't know if it's the open source component per se or were just kind of the way you guys have architected, but this concept of 'Save As', which is my word on, you know, smart people talk about forking code and this and that. And to me it's to the layman, it's the same as where you don't have to start with a blank piece of paper. I mean, the world is so much easier. You're not looking at a blank when you get started. So to be able to just start with something, to start with a model and then fork it and take it in a different direction and then leverage the community, I think that's another really powerful piece of your of your story.

Melody Meckfessel 

Yeah. I mean, you and I both know like open source, it is, the model has changed how we think about all of software development and open source has been built in to Observable in two really important ways. You raised the first one, which is this idea of 'Save As' or make a copy, right. And then get going with your own ideas or your data set shifting in a certain way. And, and then maybe, or maybe you fork and you make suggestions back to the author, right? That's the that's the open source style of contribution that's possible when working with data visualizations now, especially on Observable, we see the majority of community members and our new customers start with these save-as, right? Like find a great example a live Viz that you want to then make your own. And that way like, yeah, no one should ever expect to start from a blank slate. We want to make it easy for folks to get started. The other path that I would say that's really important principle for me working in software development with, you know, teams large and small, is that we always want people to use what makes them productive, right? And there are open source libraries that are out there that are just really incredible to get started from. And so we've integrated with a lot of those open source libraries, the two most important visualization ones that we integrate with are, of course, D3, which allows you to create very complex, bespoke data visualizations. And last year we launched Observable plot, which is to build charts, you know, in minutes and, and really provide that full range. The bookends of like, use the right tool for the right job and have the things on platform that are going to make you productive and open source is just it is the way that that's done and that's what we followed in Observable.

Jeff Frick

Well, that was a great Segway because I was just going to say, let's shift gears and talk about one of your other favorite topics, developer productivity. So thank you so I know that's something that you've been focused on for a long, long time. I remember Dave Rosen at a Google show talking about early Google days and we just data we need to automate this stuff. You know, the rate of growth is not sustainable with hiring people. So as you think about developer productivity, what are some of the latest things that you're doing to help people, whether you think you had a great line too, before, each one to remove their toil or remove the mundane, you know, let people get to work and do what they're doing. So what are some of the latest things that you're seeing in terms of productivity as we try to get more and more people in this in this world?

Melody Meckfessel 

I think part of it starts with education. So of course, I want to talk about automation and tools in a second. I think there's so much opportunity when you are community based and you kind of talk about it as like finding your data people, finding your data peeps. When you're in community, you really have an opportunity to learn and teach each other and having great tutorials, having great courses. We just launched an intro to dataviz course on the platform. I think this is something that community members should expect. I'm working on a particular tasks I'm working on this dataviz, this chart in my, you know, in my day job. And I'm also trying to grow my skills and more and more people are coming to dataviz and data analysis every day. So how do we help wherever you are in your journey, your expectation should be that you have this like ongoing education that's part of the community. We have community members that are launching workshops and sharing how tos to work with like the latest and greatest, you know, piece of data infrastructure that's out there on the Web. So that understanding and grounding in ongoing education, the importance of it, data visualization itself is in its infancy. New forms are going to be created, new data analysis techniques that empower, that enables those forms. That's all going to be going on, right? So I think that's something that's really an important expectation of the community. And, you know, I think the other thing that I would say is that in what we just spoke about before, this idea of save as or always having a starting point, right, more folks are creating templates. So if you think about different data questions in, say, retail or like think about supply chain in the last couple of years, creating templates that can be that structural frame for people to get started from. And then the last thing that I would say is when you're working in the environment, at least an observable, it is really this combination of a notebook and like integrated development environment. So you think about we just, you know, you should have autocomplete, you should have the function prompt there when you're when you're searching, you should also have access to put really easy code snippets or starting points as you're working. And that can come through either like a dropdown UI, like I'm going to pick this component and it's going to be here and maybe I manipulate it and change it or just use it out of the box. So I think there's like the big and small, right, of like start with the code snippet or have the fully formed example for you. And then finally, I would just say this idea of importing is like, you know, it's like magic, like reusing code in a way that had been created in the community directly for the job that you have on hand  means that you focus on like the the actual piece of work that you're doing. You focus on the specific task that you have at hand, because you've got you've got imports, you've got code snippets, you've got templates, and you've got working examples. And then if you get stuck, you can go ask someone a question on the form, you can go out to the Internet to get the answer to your question, or you can take a pause and go do a deep dive on a particular chart type or piece of code that's going to help you in creating that current dataviz or the next one that you're going to create on your next job. So I don't know that kind of like that's how I think about what's happening right now, and that's how I think about like my expectation as a software developer in the dataviz space or a data analyst is that I should have all those. Like it shouldn't be unique to a particular role.

Jeff Frick

Yeah, it's it's so interesting that, you know, it should be so much easier and there's so much automation. And yet I know another big theme that you like to get behind is we need to make it even easier. Right? And you've talked about training and, you know, there's the great the great resignation followed by the great layoffs. And now I think it's kind of the great reshuffle in this idea that certifications and people, you know, have kind of a different education path now to learn specific skills during the night or during their day job to to, you know, get into fields like day visualization and I know that's a big passion of you, and I've been posting a lot lately of teachers getting into technology and lots of different people. So when you, you know, how are you kind of getting behind that mission? Because I know it's super important for you to make tech more accessible to a much broader population than just the smartest engineers coming out of the top schools.

Melody Meckfessel 

Oh, absolutely. Thank you for saying that. It is super important to me personally, but also to what we're trying to do it Observable. And that's a big part of the educational courses and workshops and tutorials and live stream events that we're doing all the time. We currently have a couple of folks in the team that we're ex-professors, they love to teach and they love to build, and I think a lot of folks out there feel that way. Like when I learn something, how do I pay it back? How do I share what I know? And then I think in terms of career progression, there's all sorts of paths that I've seen of designers becoming data developers and data developers, also having to split their responsibilities to do part analysis work, collaborations that data scientists are doing with analysts and developers. So these boundaries like titles and roles are blurring and they're blurring more and more every day. And as part of that, you as a person, you still want to demonstrate and show through your work. So we see people sharing what they create on the platform as a public reference or sort of resume, so to speak, about what they focus on and where their skills are and and have actually, you know, gotten jobs out of that public display of their work. I think the other part of it is like that ongoing certification and you brought that up, especially in an industry that's moving so quickly to be able to publicly say like, I am skilled, you know, I have these competencies, you can go look at other folks also that have demonstrated these skills and the quality and depth of the certifications that are out there are really you know, they're incredible. They're getting better every day. So I think this kind of operating in the public has a lot of value for me, as even in the early stage, being able to show publicly the skills that I have and what I'm capable of and then continue on that journey as far as you can possibly go. So yeah, we're offering certifications. We're really interested in doing collaborations with other educational institutions and actually where we're Observable today is used in really kind of significant portion of educational institutions, university programs, even high school programs Jeff, are using data visualization, elementary schools. It's pretty wild to see how if you have access, you know, the tools to do this work, how much you can teach for folks that are interested in learning.

Jeff Frick

Right, Right. I mean, it's so much there. I mean, we'll get so much more innovation through the democratic, you know, democratization of these tools. Right. More people, you know, experimenting, asking more questions with access to the data and access to the tools are going to find new things and have new discovery. I think it's so important. And then the other thing, right, is the job, too, is talking to somebody of their day who's kids looking at schools for certain topics. And like the jobs 20 years from now don't even exist, you know, a lot of them. So, you know, this concept of of continuous learning and then also, you know, providing skills training for roles that are open that there just aren't enough people. I mean, they're just they're just not there. So, you know, hire for attitude, train for skills because even today's skills are going to start getting stale tomorrow. So, I mean, you've really got to have this kind of continuous learning attitude and certainly an open source ethos and sharing in a community is a big supporting piece of the success of that.

Melody Meckfessel 

Yeah, I absolutely agree. Absolutely agree.

Jeff Frick

So let's shift gears a little bit and talk about all the buzz right now, which is LLM large language models. You were you've been involved with big data sets for a long, long time. As you just kind of sit back and hear the buzz. What what kind of goes through your head in terms of kind of what's the potential, what people need to kind of settle down for? What makes you excited about, you know, kind of this change that's happened in the last six months, at least in the noise level, if not necessarily in the technology. But I think there's been some pretty significant technology breakthroughs as well.

Melody Meckfessel 

Yeah, I think it's been incredible. And I, you know, in the theme of how do you help people continue to increase their productivity, I think there's a big lift that is coming from the assistance from all of these tools, the assistance, the level, the quality of suggestions, I think will continue to improve. I think the the thing that I keep coming back to, though, is that it's an assist. I think if people are viewing it as a complete replacement, we still have more work to do. So, you know, in in the developer productivity space, in the data space, we're going to continue to see integration of, yeah, just how rapidly things were evolving with LLM and you think about the AI assist kind of apps and tools that are coming into play. I mean there's and more every day that are being released and I think that gets at. You know, there's a lot of technology and a lot of work that we do that we should have this assistance when we've had it for a long time in like code editors and code tools. But it hasn't been at the level of kind of scope and speed that we're seeing today. And so I am really optimistic and also interested to see how folks that are coming new into technology use these to assist them. I think there's always going to be a case to go deeper and learn. And when I have in the past been in situations where, you know, I'm just I'm just copying and pasting like the craft and the depth of the problem solving starts to like, go away and then and then kind of what creative space are we in right now with data and with technology? Because so much of that is about the business problem that companies, you know, the companies, the questions that people have and the business problems that they have and that that's evolving quickly. So, yeah, I'm really curious to see where it goes. I think that there's a lot of benefit that can come with it. I, I don't think it completely removes the learning that needs to happen for people that are working with data. You know, if you think about everyone in the world having some data literacy just like we learn to read and write and communicate, there's an aspect of that that you bring your own learning and your own perspective to, and then you have the tools and the assistance and the automation to do that. So I think we're going to see a lot of growth in the space.

Jeff Frick

Yeah. Or even just assuming that it's successful for helping remove toil, right. And setting up basic things to enable you to get to the higher level function of the, the business that you know really well or, you know, kind of the nuance of the data that you want to explore, you know, to be able to get to that starting point faster if that's in fact, what it enables. Again, I think for innovation is just more people having more access to more data and more tools and as Scott Cook says from Intuit, you know, trying more low cost experiments just to discover new things. I think it's really it's pretty, pretty interesting. Yeah. And trying to drive more to this data centricity, which we all need.

Melody Meckfessel 

Well, I think you hit you hit on a really important point, which is when we don't worry about the cost of pursuing an idea. Right. If the cost meaning both like time, the time to explore that idea is really low, really interesting creativity and innovation will come out of that because we will pursue those ideas, right? We won't hold back. We'll take more risks if we can move quickly and kind of cut it off when it doesn't go in the direction that maybe was expected for the outcome. I think that's it. Right. Like that's the benefit of your 'save-as', where if you can just keep pursuing other people's ideas and new concepts and you have this assistance and we don't have a lot of toil like, I mean, we're already seeing it, right? We're already seeing an incredible amount of creativity and innovation happening around the space.

Jeff Frick

Right. And it's barely even got started. I mean, I know, And then, of course, there's all the back side. There's all the back side issues as well in terms of ethics and choices and datasets and bias and a whole host of hornet's nests that are getting kicked here. But ultimately, you know, most people feel pretty strongly that the upside down benefit is pretty high. I just want to on kind of the last topic and that's really kind of dev ops and kind of open source, but really more like Moore's Law was to me more about the thought process than necessarily the math behind microprocessors. But this this attitude of continuous development, right. This get it out, get it, get a minimum viable product out and just this iterative process. And we're seeing it specifically in this OpenAI space where they're even saying they put the models out on purpose so that people can try them and test them and see what happens. So I think the proliferation of this technique is probably much more significant than most people give it credit for. And you've talked about it quite a bit, this kind of dev ops attitude as opposed to necessarily just the DevOps process, which is this really powerful iterative process. You layer on open source with a broad community. Wow. Really moves the needle pretty quickly.

Melody Meckfessel 

It Does. It does.

Jeff Frick

All right, Melody, any final thoughts before I let you go back to playing with really cool charts and beautiful visualization?

Melody Meckfessel 

Yeah, no, I just I, I would say that, you know, just you mentioned DevOps, the idea of tooling being in service of the job that you're trying to get done, the practices, the processes, which is really like how people work together, collaboration and then the culture. And to me data driven culture is, as soon as someone's asking you for like the metric or the dashboard, you're not in a data driven culture. A data driven culture is about curiosity. It's about asking questions. It's about listening and learning from other people's expertise. It's about listening to the data. And to me, what we're trying to do at Observable is really be in service of data driven cultures in that way where it is about exploration, it is about going deeper and it's about the fact that there's so much data. We have to have visualization to do that right. And we need visualization without limits, without the stain in the box. You know, we need to have the tools to use, use the spreadsheet-like interface or write sequel ourselves, or maybe write some JavaScript to work with our colleagues and keep going. I think that's the world we're in, right? Like it's going to keep moving quickly and we as people in all of our respective companies need to move quickly with it, especially as it relates to our data. So I'm really excited about those practices that have been evolving quickly for software development coming to data practices like this, is that this is the generational shift. I think the we're in as it relates to data analysis and data visualization. So I'm really excited to be part of that. I'm really excited to contribute to it and have Observable help, you know, help our community do that as well.

Jeff Frick

I love that summary Data centricity is about exploration and then listening to the data and then doing something about it. That is a terrific summary. Well, thank you as always for all the work that you do and helping make product excuse me, make Devs more productive. Not to mention bringing the visualizations to the world. So thanks.

Melody Meckfessel 

Thanks so much.

Jeff Frick

Thanks for all you do. All right. She's Melody, I'm Jeff. You are watching Turn the Lens. Thanks for watching and listen on the podcast and we'll see you next time. Take care. 

Cold Close

Sweet

Thank you.

‍