Ron Itelman: Thinking in Networks

Loris Marini - Podcast Host Discovering Data

How can we maximize innovation in the enterprise? Today I learn from Ron Itelman in a voyage from data contracts to metacognition, neuroscience, and more.

Also on :
Stitcher LogoPodcast Addict LogoTune In Logo

Join the list

Join hundreds of practitioners and leaders like you with episode insights straight in your inbox.

You're in! Look out for the next episode in your inbox.
Oops! Something went wrong while submitting the form.

Want to tell your data story?

Loris Marini Headshot

Checkout our brands or sponsors page to see if you are a match. We publish conversations with industry leaders to help data practitioners maximise the impact of their work.

Why this episode

Network effects happen when we remove all silos and align systems and people. As we’ll see today, this is more than just a data challenge. In an attempt to introduce data contracts, we end up exploring the realm of experiences, objective and subjective meaning, metacognition and neuroscience, computer science, and information theory.

We talk about product design, semantics, the problem of conflicting definitions, and the need to bring different people together to collaborate and learn from one another.

Ron's passion is data product management methodologies that enable collaborative human + machine learning networks. Ron helps teams unlock their network potential by aligning software engineering, design, and data science teams to a common vision and roadmap. More from Ron on his website: https://www.synchronicity.ai/

You can follow Ron on LinkedIn.

For Brands

Do you want to showcase your thought leadership with great content and build trust with a global audience of data leaders? We publish conversations with industry leaders to help practitioners create more business outcomes. Explore all the ways to tell your data story here https://www.discoveringdata.com/brands.

For sponsors

Want to help educate the next generation of data leaders? As a sponsor, you get to hang out with the very in the industry. Want to see if you are a match? Apply now: https://www.discoveringdata.com/sponsors

For Guests

Do you enjoy educating an audience? Do you want to help data leaders build indispensable data products? That's awesome! Great episodes start with a clear transformation. Pitch your idea at https://www.discoveringdata.com/guest.

💬 Feedback, ideas, and reviews

Want to help me stir the direction of this show? Want to see this show grow? Get in touch privately or leave me a review with one of the forms at discoveringdata.com/review.

What I learned

Share on :

Want to see the show grow?

Your ideas help us create useful and relevant content. Send a private message or rate the show on Apple Podcast or Spotify!

Episode transcripts

**Loris Marini:** so I'm here with Ron Itelman. Ron, thank you for being with me today. Welcome to the podcast.

**Ron Itelman:** Thank you. I'm really excited to be here. You've had some amazing guests and I'm just like very excited that I'm included in the, your conversations.

**Loris Marini:** Oh, man, are you kidding? This is such a pleasure. I've been following your content on LinkedIn for a while, and I, I just love your style and how bold you are with not thinking twice about, going from a thought to a graphic that engages, that explains a concept that is deep. maybe I'm jumping the gun.

We go slowly here, but I I'm pretty sure this is gonna be a great conversation, . So let's let's jump right into it. why are we here today?

**Ron Itelman:** We're here today because I've seen a lot of people who have incredible talent. They can think out of the box. They can think in the box. They can, they have skills that they've developed for years, and they're frankly not being utilized. entire teams are. Slow down because of often bureaucratic inefficient systems that businesses use.

And we're here to start to think about how to break down silos and get people to collaborate effectively. And it's really about, helping people be successful. And the worst case of this is when you see things aren't working and teams are blaming other people whether it's individuals or blaming other teams, that it's their fault.

And this is, this can be often a toxic situation. I know I've been in some of those. And so I really started to think about are all these frustrations and inefficiencies happening? And if you start to pull on that thread a little bit you begin to see certain principles emerge. And that's what I've been focused on for.

At least five, five or six.

**Loris Marini:** Yeah. And that's, that definitely resonates with me. And I guess with a lot of folks in the audience whether, you are, you're a business leader a data scientist or an analyst, like if you have anything to do with data and you're trying to get some sort of value out of it improve that I'm sure you'll resonate with this problem.

So I love that you talk about the system because a system is a beautiful word. I think it's. It encompasses everything there is, that's required to achieve functionality. So it's not just the technology, it's not just the people. It's not just, the communication is literally everything.

It's the holistic view on on data. And you talk about under utilized skills, that's a huge problem. And I think it'd be interesting to see how much of that actually contributed to the silent quitting phenomenon that we are seeing. And a lot of people just going, it's enough.

I'm gonna just do something else. I don't wanna be involved in data anymore because it's just too much frustration. I don't see value created as a result of my work. And not everybody is okay with that. Some people can do it, some people cannot. And bureaucracy and efficient systems are definitely killing, the ability that of data teams to achieve some sort of velocity and prove their value.

Let's, dive into each one of this and we can take the conversation any, anywhere you want. Because I think this is a complex topic, so there's no right way to start this. So I'm gonna flip it. I'll leave it to you. Where should we start?

**Ron Itelman:** I wanna focus more on what you said because as you were speaking, if you think about there's a real cost, not just to the people on a personal level who are not. Having their talents fully utilized. On a personal level, someone's not getting job satisfaction cause they've got all these bureaucratic systems. They're not able to do what the type of modeling or research they're interested in.

There's an organizational cost. At its worst, you can have layoff. And, that affects the, all of your teammates. And the impact of, losing jobs impacts negatively families. And then there's also societal costs where if you're a biopharma company and your data is siloed, which was shocking for me to learn about how siloed data can be in the bio biopharm industry.

Then drugs aren't being created as quickly and efficiently as possible to help reduce suffering, right? To help cure disease. think this problem is much bigger than simply bureaucracy. if we can figure out how to let companies leverage data better, they will help create a better world.

And this is really what the world I wanna live in, is I want companies to be able to maximize innovation and create wonderful new products and make wonderful new discoveries. And unfortunately, very talented people are not able to unlock that value they have because there's very little methodologies to bring together the business and the data and the software engineering and the design teams.

the rate of technology innovation is, exponentially growing fast, but the rate of organizational design innovation is not keeping up. And that's really what think is needed to help accelerate


**Loris Marini:** in.

So again, the system view, so it's not just design, it's not just engineering. It's not just security. It's not just marketing. It's all of them. And we need to find a way to, to bring all these people literally on the same page, which is not something that, teams love and team leaders love to say, but it's hard to put in practice.

And there's probably a billion reasons why that is. we could explore that if you want. Or

**Ron Itelman:** I'll give you a billion reasons right now. I'll give you a literally a billion reasons right now on a project I worked on, okay? There was a billion dollar a year revenue product I worked on.

**Loris Marini:** Yeah.

**Ron Itelman:** And they had something I'm guessing anywhere from 60 to a hundred people working this project.

A competitor of theirs had this feature as related to AI and in the education space.

**Loris Marini:** Yeah.

**Ron Itelman:** And they wanted to have it, and they called this one feature by one phrase, and all the product managers sat and agreed, we are going to, in, in two quarters, we're going to do this and we're gonna have a competitive product.

And everyone's excited and the company, puts in $10 million budget, 60 people and go agile tooth weeks. No one took the time to make sure that word meant the same thing to the different teams. So all the teams. Agreed that they were gonna do that one word and they were all working towards it, but that one word meant different things to the different teams.

And so they were all going like this. And then guess what happens in six months later, you find out nothing's lining up, nothing's integrated, and now everyone's, pissed off, upset, scared blaming each other and there's millions wasted. And the executives are like, What? What's going on with the project?

**Loris Marini:** Yep.

**Ron Itelman:** So this concept of how do we architect, how do we build, how do we design? There needs to be a formal process to literally get people on the same page. And this is a product management problem, I believe. And. That's the world. That's exciting to me because it seems like a wide open field.

This I don't think this has done well, so I wanna dive more into it, but I could talk, forever,

**Loris Marini:** yeah. No. I love this because interested in the systems that power, that level of collaboration and and how many levels there are in the, what you call the pyramid of knowledge or the concept plane. I'm looking at one of your graphics that really stuck with me because, there's a usual data, information, knowledge, the diy pyramid.

It's interesting because you have some you make a clear diff difference between the concept plane and the knowledge plane, the experience plane intelligence, and the fractal plane.

Guide me through these layers? How do you see them? Why do we need to make it, such a difference between these layers and what. And ideally I'd love to explore with you examples and then we can brainstorm because I have some stories in my, in the back of my mind as well, of when things went wrong at each level and, how does it feel, How do you know you have a knowledge plane a crack or a silo?

The knowledge plane level? How do you know if you have it at the experience level? Cause I think it's interesting to understand what can we do about it, right?

**Ron Itelman:** Yeah. start before answer is to say that everyone has different meanings of the word information, knowledge, et cetera, et cetera. People should do whatever works best for them. I can only say what I've learned having worked with teams of PhDs in learning science. These are learning scientists who study how people learn in education.

their background is in psychometrics, and so I learned this kind of thinking from them. And I've also learned from the machine learning I try and ground my phrases in mathematics, physics, or academic terminology and from psychometrics machine, computer science, right? So information I'm trying to borrow from Claude Shannon's communication theory, which is basically if you strip away all of the meaning and you're just measuring the amount of bits of information you need to resolve communication that gives you a universal and generalizable way to connect any communication, right?

And that's really important. Let's say in education when you. All these different content that you give to students, and there's no way to normalize it. The only way I found that you can connect equalize everything is by measuring the amount of bits, right? So how many bits of information did one piece of content affect someone versus another?

And when you look at it from a kind of a computer science approach, that also opens up other areas of thinking, like what's called Kolmogoroff complexity, which we're not, It's too advanced and deep, and I can barely explain it, but it's, that goes in the direction of a reinforcement learning where you're what is the minimum amount of communication needed in order to successfully program something, right?

so information is our foundation. and it's devoid of any meaning. It's just purely like, how do we measure the signals right? Of, communication. so data is also very interesting because in this one people on LinkedIn, the semantic people, library information systems, like they really struggled with my definitions, right? Like really ch like really were challenged by it. But let me explain what I mean.

Data by itself in my mod, in the model that, that this is proposing is purely a means to capture information it can be a piece of paper and it can be in a database. But it's not describing the meaning, right? So let me give you an example. If I write down an I imagine alien symbols, right?

So these are alien symbols. A, b, this has no meaning to you, but it's still data, it's still information that's recorded, right? So information that's communicated is music data is like a painting, it's something frozen in time, right?

those are the of distinctions the next level up, which is the concept plane, right? That's where we start ascribing meaning with metadata, right? what are these measures and metrics? And you can have many dimensions to describe a single data point. You can have, 50 dimensions of metadata for one field of data.

That's when we get into the world of concepts

**Loris Marini:** For example, if I draw a house on a piece of paper there's information because there's ink on a paper, there is data because it's, there's medium and mean that data is there and not a piece of paper I can convey to you. But there's also a concept that forms in my mind when I see it.

And if I go to someone that has never seen a house take a child, that has never drawn a house on a piece of paper, the first time that they see, they go Oh, what's that? So you have to explain, Oh, that's a house. So they start creating a concept in their brain of what a house is.

Is that metadata?

**Ron Itelman:** Yes. And we're getting into the wonderful world of experiences now because you're getting a picture in your mind of the house, you are, your brain is generating that, right? And that gets with the language and everything else. And these are experiences we have. So there's this weird thing, right? So I'm looking at you, you look at me you look to me, you're outside, You're my monitor.

But actually this entire experience is being generated in my. Your experience of me is generated in this your skull. It's not outside of you, it's inside of you. But our experience is, it's outside, right? And the concept of this house and this child has never seen a house. It's being transferred from my brain through my words, through a, a camcorder and audio.

It's being converted. And it goes through to the other side to you, and you're able to understand it because there's some normalization cuz we're human and we have similar experiences where we can understand each other, right?

**Loris Marini:** yes.

**Ron Itelman:** So the structure of meaning, even though it's happening inside of us, is preserved somehow and transferred through bits and data, and this is this wonderful weird world of concepts and experiences, right? And in computational psychometrics,

we have to. Predict what is that other person's model of a house? Because we can't get in that person's brain. So we have to be able to model other people's models of concepts. is like magic to me. I can't believe I get to work What's knowledge? Okay. Knowledge is when there's a problem to solve, and then we have to take our concepts and operationalize them in the experience plane to solve a problem. in education, we call those learning.

**Loris Marini:** Yep.

**Ron Itelman:** student has to learn to do the Pythagorean Theorem. The goal is to learn the Pythagorean theorem.

You can't know if they actually know it. The only way you can know it is to test them. They call it assessments, then how do you assess it and all that? They have all these mathematical models and the goal is to influence them, to push them from less knowledge to more knowledge. And

and my work is focused on the business. How do we give knowledge to the employee for the employee to solve a problem, I'm

**Loris Marini:** right, right, right.

**Ron Itelman:** Now, intelligence, right? Is go, is going to have to do with, we have the human intelligence, but then we also can have the intelligence of a system.

What is the best knowledge we can give the employer, the student to solve a problem at the most efficient rate. And you have to be able to model all of this stuff. It gets pretty complicated but for intelligence, what we wanna do is we wanna be able to understand the situation. This is very similar to a reinforcement We have an agent in an environment,

and we want to predict what is the knowledge that agent needs in order to succeed in their

objective. Okay, fractal plane is the most beautiful and wonderful, and also the hardest to explain, but I'll try and do my best here. there's a couple things to think about, right? There's, we talked about language, right? Like this one word, meaning all these different things to all these different teams, right?

In that company that I worked for, that they had one word, they all agreed that they were gonna build that one feature, but they all had different definitions of that feature, So they went in different directions, right?

**Loris Marini:** Yeah.

**Ron Itelman:** We'll call that the meso plane, right? So you have micro meso, macro, right? The meso plane, the human plane, they all had the same word with different meanings. But then when you go down deeper into each Groups, they would take that concept into code, into the microplane, right? then the business, as an organization would use these features to, to make decisions and product decisions. And what I'm trying to do probably terribly in this example, is that system was completely chaotic.

They weren't in alignment on the meso plane. And then if you look at the software engineering code and the data scientist code, they're using different languages, different data structures. So what I'm focused on is how do you create alignment in, it's like horizontally across the different teams, but then vertically for each team, even the code level to be align.

And then at the macro level the business strategy should be aligned. And then you have this wonderful thing called collaborative learning networks, where you can even have businesses that can share information and can learn, is fractal thinking. It's, you can zoom in and zoom out just, like I said, the word house is going from inside my brain to bits and then to you and into your brain, and then to back.

these concepts, we're trying to align them in, horizontal ways, in vertical ways and at multiple levels. And so the best word I have to describe that is fractal, right? Cause in a

fractal, you can zoom it, you can

zoom in,

**Loris Marini:** And you'll see the same structure repeating over and over.

think about.

**Ron Itelman:** One project I worked on was for government, right? When we had diversity, equity, and

inclusion, we were seeing that government and foundations are giving money to all these different non-profit organizations, but they had no way to measure whether they were being successful.

And if you think about this as a giant network, you have nonprofits, governments, and foundations. If they agree to the same standards, to the same structures of knowledge, then they can all collaborate together, right? That's the ultimate form of this kind of thinking, is collaborative learning networks, which I don't know if my lifetime I'll get to work on, I.

**Loris Marini:** This

completely, another level I think, and that's what we need to do more, I believe, to tell the story of why this matters because we are not used to think in, in those terms. There are no many common real world experiences that we have, in our repertoire, in our portfolio of experiences in our brain to think about this stuff and actually understand it at the knowledge, experience, intelligence, and fractal plane to get better.

we need to get a lot better, I think, as storytelling as an industry to explain why, I was having a conversation with a dear friend of mine leading, he, he leads the r and d division of a big multinational food company.

And when I was explaining, why they, and all this focus on knowledge and information and he goes yeah, but in the end, all this stuff is gonna take time and my team is already exhausted. We, we achieve capacity. I can't ask people to do more work if I if you ask me to collaborate from my domain, my department with another one, because data is a shared asset and it benefits everyone.

What's in it for me? , I was there trying to explain what's in it for you Is that if the organization is connected at these different layers, then things happen much faster. And so you get less frustration, perhaps even more funding, you can prove value for a new initiative much quicker. And so you, what you get back is that you go back home with less headaches and you can do more, and maybe you get promoted even faster.

But, it feels like a promise. Far in the future for a lot of people that don't, they're not used to think in terms of strategy and long term planning. The managers, the middle management in particular, I they're promoted based on their ability to get shit done, right?

They need to respect a schedule, they have to make things happen. It's all about the timing. And so that's another, I think, gap that, that sometimes we need to bridge because yes, we need people on doing the work, the line workers being data scientists, knowledge workers, or literally factory workers.

To appreciate and think about this stuff. We need the C level folks to to sponsor these initiatives, but we also need the folks in between, right? It has to be an organic, like at all levels of the organization. We should try and reach a consensus on what that concept plan looks like for dead information and knowledge and I think that's what I'm really interested about, that challenge And , we, it's gonna be, it's gonna be an interesting decade.

**Ron Itelman:** Let's zoom in on that cuz I think you hit the nail on the head This is in my mind, if not the number one problem. It's certainly the top problems. Which is why I think it's important we have a methodology a product management methodology that we can give to the C levels that they can tell their management, especially middle management, to follow this playbook.

But I wanna tell you a true life story about their, a gigantic multinational company. I was working with one of their leaders, middle management on a data project, and the metadata team, which was completely underfunded and, was crying for help and in resources wasn't able to keep up with a request.

And I said, Look, if you wanna do this project, like we need to involve the metadata team. And the manager said, my KPIs are that I do X, Y, and z. I'm not gonna sit around and wait for the metadata te team, I'm gonna do what my performance reviews based on. And and that's how I keep my budget basically.

that's how I get my promotion. And it was very clear that this was harmful, literally harmful to the success of they were spending hundreds of millions of dollars on this digital transformation that this kind of thinking was harmful to everyone else's success. Cause the middle manager was like, I'm only gonna focus on what my bonuses and my performance review is, right?


**Loris Marini:** And

**Ron Itelman:** we

**Loris Marini:** can't blame them for that. That's natural thinking. We are animals after all right.

**Ron Itelman:** Yeah.

**Loris Marini:** to

**Ron Itelman:** So we need to have the incentive structure, Fundamentally looked at by the executives and to think in networks, right? Everything I described about the fractal, the highest point in the pyramid, the fractals require a network approach. people don't think in fractals.

People don't think in networks, right? So this is a, real challenge, what you're highlighting. It's probably the most difficult challenge that I'm not worried about the technology. I'm not even worried that if I had a budget and a team and an executive buy-in, that we can build these kinds of systems.

It's the, almost impossibility of getting people to think like this because the minute I talk about networks and fractals, people are just like, I just wanna have my search take three seconds and not 30 seconds my search results when I'm trying to find a file, or I just want this app to not crash, or, I wanna bring my cloud costs down.

They don't necessarily wanna think about redesigning, their, the way their systems operate.


**Loris Marini:** It takes hugely strong leadership to continuously communicate the why. This is something that you would want to go to, to move towards. And and also reinforce those incentives you're talking about, which, yes, we need to have clear metrics to measure the performance of someone at work because, lack of clear parameters leads to people feeling demotivated, not clear whether they're doing a good job or not.

It feels to me we are not organizing teams and organizations in general, like the Strat, the topology of an organization. It's still very much siloed. when you are on board and when you are hired for a job as a marketeer or as a salesperson.

You get to hang with the sales people or the marketing people. You don't talk to the engineers. A market here in a, at a dinner table with a bunch of engineers, , it's not a pleasant experience because the engineers talk.

They're engineering lingo. Can we imagine a future where when you join a company, you don't join a team? Yes, maybe you spend more time within sales, but really what you do is you browse around different teams and there's structures and incentives. That the stimulate you to do that.

So it's almost like community experience within the company, but that also means how do you allocate budgets and how do you promote people? Without connectivity, nothing happens. So a lot of, in data management we talk and standards, interoperability between systems because without those standards, without those contracts, without those agreements in between systems, we can't share knowledge and we always gonna get, keep getting stuck in the realm of the silos. How do we get out of that run?

**Ron Itelman:** Yeah. So the first thing is I think we need to give people a playbook, right? A data product management methodology, a playbook that will guide them through how to connect systems. And thinking and creativity. Okay. That's one thing. The other is let's dive into the data contracts. So going back to the lowest level, we have information.

So information is, you have a sender and a receiver. And what the signal sent in between that's our, that's our information, right? Where information is a description of that signal, right? A data contract in the simplest form should say the sender and receiver are going to agree to how to interpret those concepts, right?

And so the way I would approach it, the way I would recommend is you can do it with a simple Excel file. If you're gonna send an API and it's gonna have five properties in that call. You need to have a a unique URL or unique identifier so that each one of those properties has meta information and that reference to that meta information must be sent with the information, right?

You have to transmit it. And that is the only way I know that you can guarantee that the sender and the receiver are on the same page. And that's because you have a third party reference frame. Now, the beautiful and wonderful part of this story is what inspired me for this kind of thinking was the Italian architect in the Renaissance.

I'm gonna miss pronounce his name. I think it's called Bruno. Brunelleschi. He is one of the creators of modern architecture, specifically perspective line drawings. He created the perspective lines.

why was it such a catalyst of innovation in architecture in the Renaissance, the reason is he could create a diagram and then give it to anyone. And any of the construction workers could deduce exactly where someone was to go.

It got everybody on the same page. And then because you knew where everything was gonna go, you could plan for the future, you could build things and you had a guarantee, a contract that you were safe in your decision when we talk about concepts, because they're invisible, It's much harder to imagine these perspective lines, but that's exactly what a data contract in my mind must do.

At least in this framework, a data contract must be the equivalent of an architectural perspective line, but for concepts, right? It's a third party reference that anybody can use and is guaranteed a third party reference reference frame for that information.

**Loris Marini:** I love it. back to the pyramid, right? The information data concept plan, I was thinking an example to crystallize it. I a hundred gigabytes hard drive, right? And I fill it up with zeros, right? I just, I pick one, it is zero one.

I just write 1, 1 1, 1 a trillion times until I fill out a hundred gigabytes. I've got a hundred gigabytes of data. But really I have one bit of information. Because it's super redundant.

**Ron Itelman:** Yes.

**Loris Marini:** imagine for a second that we agree that one really means house, right?

The word house we encode it. We say one means house,

right? So I've got house house 3 million times, but really only one time. So now the concept is one, right? We've got one bit, one concept, one word, really not one word. How, So what I'm trying to get to is , you don't, you might have a hundred gigabytes of data, but there is only one concept

**Ron Itelman:** Yes.


**Loris Marini:** So you don't need big data, you don't need Thera bytes. You need to worry about the concept that you extrapolate from those beats, because that's what ultimately will give you an insight to act, right? So just is that, does that making sense?

**Ron Itelman:** Yes. And why I'm reacting so much is, do you remember when I said the word Kolmogoroff


**Loris Marini:** Yes,

**Ron Itelman:** The shortest amount of information required to successfully execute a program is, that's exactly it. All of that trillion zeros. The minimal information you needed is, was one bit to achieve that. Instead of reviewing people's performance, reviewing systems performance, what is the minimal amount of information needed to solve objectives, That's the fundamental thing we're going towards. the least amount of information the more efficient your system, right?

So when I go into these clients right now, and I see these incredible inefficiency. in these businesses, words mean different things. People, have no metadata. It's mind boggling. Like it's so obviously


**Loris Marini:** That has copied all over the place. copies of


**Ron Itelman:** things are lost in emails, right? Again, it's a, simple stupid thing. when we say this word, does everyone agree to what this word means? There's no technology in that, but the, that's tens of millions of dollars

wasted because of that simple thing.

**Loris Marini:** There's an important plan here. Where we barely touch is the experience plan. We made that connection with data information concept, but then there's knowledge, the ability of use concepts to solve problems. So for the right concepts, in the right contexts at the right time.

But then there's the experience playing and that's really where design, a lot of the design people, they worry about user experience. They experience is baked in the way they think in their framework because they worry about. What is the experience of the user gonna be like, which has nothing to do, or, it's only remotely related to the feature.

How do you build the thing? You can have something that conceptually it's phenomenal but the experience is really bad. Everyone feels and experiences reality in a different way because of the, billion interactions we had since the moment we were born.

So somehow someone, will have to bring this perception, not necessarily on the exact same plane, because diversity is an asset. We do want people to think in a different way, but there has to be some sort of minimum , common denominator so that we can have a conversation and mean things in the same way.

Are you aware of any company that starts, that is starting to understand this and have specific people, for these types of roles? Your job is to talk to marketing sales, make sure that they. Speak the same language.

**Ron Itelman:** was working with these amazing data scientists. This is that billion dollar project.

These absolutely incredibly talented people. And they were trying to predict this computational psychometrics. At what point in, let's say, solving the PHA in theorem a student would get stuck. So you're predicting what someone else's knowledge is. And they had incredible accuracy here and, but they still had some significant problems.

And I'm working with , the data scientist, and I'm just coming from the personalized learning and design team. he said to me like, how stuck is, And I'm like what if we just ask the student when they're stuck, push this button and then you'll know exactly where they're stuck.

And he was like, Oh yeah, , right? So just imagine they had an entire team of PhDs trying to salt predict, and there's a lot of value to that. But just adding a button for a student to say, I'm stuck here, would give them information that they had no capability. They had no way to influence the design of the experience All they needed was to have a conversation with a designer.

And then on the other side, the designers they don't know what, linear regression is or whatever. They get totally intimidated and by all this stuff. Or it's just too much. Or they're excited and they just have a learning curve.

And this is what I'm talking about, bringing people together to have conversations. This is why a product, a data product management methodology, right? It's about bringing people together to communicate. Cuz that's how we move forward. That's how we push innovation. It's by enabling creativity and communication with people.

That's the value, right?

**Loris Marini:** no, absolutely. I think this is essential atomic data contracts, right? Like we talk a lot about data contracts now.

The only way you can make a decentralized system work is by agreeing on things and making sure that these individual actors stuck to one another.

So we call them con contracts or some sort of, maybe a different word, but that's what we mean in the microservices world. Now, in data, some, Zak said, Why don't we use the same idea? And so the whole idea of data mesh came through and now contract said a fundamental part of that layer I wish I could put that drawing into the podcast then transmitted by audio, because we've got an experience plan. This is your stuff. Two people talking and saying, we agree on a world, and the title is Atomic Data Contracts. And then we can go down into, how do you actually do it? But. What's hard is what you put into the contract. And also how do people interpret the contract and are able to reconstruct the concept plan and.

Full back into alignment when they meas, when they become not aligned anymore with the concept. Because that happens all the, especially when you move fast, So we have to have a system where dynamically we have to be okay with the fact that we're gonna fall off alignment constantly, because things happen and people change their minds.

We hire people leave, right? But have a system that's automatically heals this cracks and so that people can very quickly with a couple links and two minutes, I think currently, go back into alignment to go, Oh, okay yeah. That's what we're talking about. We're talking That we're talking about customers?

**Ron Itelman:** Yeah. this goes back to the topic of centralized versus decentralized. And the approach that I would recommend people take is not to do this as an either or, but as a hybrid. I do think there's a tremendous amount of value from the centralized approach that the li the library information science, taxonomists and stuff take is no question.

It is extremely valuable. There's also tremendous value in decentralized software development where people are coming up with their own terminology and their own phrases and language, and that's a lot of creativity and freedom and removal of bottlenecks. And so the question becomes, I feel like a lot of the paradigms we've seen is like, like for example, data modeling first, right?

It's like very controlled. And then, and I don't know I just bought the data mesh book. I just started reading. But on the data my side, you have the decentralized domain control, right? And you have this concept of computational governance, which is interesting, but it seems to be leading more towards decentralized.

How I'm trying to approach this and how to recommend people to approach this is, you're exactly right. if you stifle people's creativity with too much control of your concepts, they're gonna find ways around it. They're gonna sabotage it or they're gonna abandon it, or they might even try to comply with it.

But as you said, things are gonna go out of sync. It's just, it's going to happen. And so you need a resilient, adaptive, and flexible and easy, and a critical word here is easy. For people to communicate with each other to resolve this. And so that's why I don't think it's one side or the other, and I don't even think it's saying planting a flag in the middle.

I think it's the ability to move back and forth depending on what's needed, is some flex, flexible and adaptive system. That's what I'm researching now actually at work and very excited. We're making good progress on that front as well. Yeah, these are these are very



**Loris Marini:** How do we, how we, how do we know more about your work?

**Ron Itelman:** At Clarkston. We're going to release, I think in one week or the end of the month. Announce this entire the product management methodology. I'm gonna do a little short video, three minute video about it. We're open sourcing what's called the Data Product Studio, which the data product studio, all it does is take these 10 steps and it tries to put them in an open source app.

My job, I feel, is to evangelize this, to spread it, cuz as I said, we want the world to adopt this way of thinking. And if we try and charge for it or make it closed, it's gonna limit it. So that's why we want to make this open. We wanna make it free. And we want to, we believe people will be able to do better things.

The danger of not doing this is you stifle innovation. We want to enable information cuz that's what truly innovative people, teams, and companies do.

**Loris Marini:** Amazing. Yeah, please make sure to send me a link whenever that video is live. we'll, I'll do my best to add the link in the show notes and ensure that everyone that is interested about this stuff has a chance to see and contribute.

I'm certainly gonna be there waiting because we need we, we need this sort of initiatives run. And you certainly inspired me to think in terms of networks and if there's one thing I'm gonna bring home from this is that we need to.

Let go. Props of our fears of, and the exposure that comes with it. When you start collaborating really, truly with someone else from a different domain, that feels scary, right? Cause you don't know would let people understand you. Maybe they have a different language, different interpretation.

Nobody wants to feel isolated. We want to feel connected, but things have to get worse before they get better, right? We need to get through their uncomfortable face of not agreeing and not feeling like we belong to then hopefully reach a better state where we are comfortable.

We know belonging, knowing full well that we can get back to belonging anytime. Because we've got systems that are resilient, that are easy to set up, that allow us to do it. One one thing I want to do to close this off is if you. Where to condense all the many years of experience you have in the field and give 1, 2, 3 core tips.

Whatever the number to head of data lead or a data scientist or anyone really feeling the pain of this lack of connectivity experience plane, in the knowledge What would you tell them?

**Ron Itelman:** best analogy I have is think of synchronicity, specifically. Think of a band. If you're going to go to a concert after a hard day's work, and the band is not playing in the same key and they're not playing in the same time.

It's just gonna sound like noise. Your employees and your organization they're like musicians. The information flow, the data flowing through all your systems is like the music. Focus on getting synchronicity between your teams. Get them to play the same scales and in the same tempo together.

And you will have a far more efficient system to do that, right? They will be able to create and collaborate together, but none of that is possible. , they might be great individually, but it's gonna all be noise until you get them aligned. And that's really the philosophy is I think of an organization in terms of really synchronicity.

**Loris Marini:** I absolutely love it. I'm gonna add one thing. Yes, it's hard and it can feel sometimes really even pointless depending on the state in which, you know you and your organization is. But I am a relentless optimist and hopefully I can transfer a little bit of that optimism to you.

There is a way forward. I don't know exactly what it is. I don't think anyone knows, just goes back to the discovering data idea of the actual name of the podcast. But but we gonna get there , because that's what we do, right? We're problem solvers and let's get uncomfortable. Let's talk to one another.

And figured it out.

**Ron Itelman:** I love it.

Yeah. You continue to have amazing



thank you for


me on.

**Loris Marini:** Thanks Ron. I'll speak soon

**Ron Itelman:** All right. Take


Contact Us

Thanks for your message. You'll hear from us soon!
Oops! Something went wrong while submitting the form.