It's archaeolatry. Archaeolatry. Data before it knew it was data. What do you got? I have a quipus. It kind of looks like a macrame necklace. A horizontal thread with a whole bunch of knotted threads of different colors hanging from it. Or like someone started disassembling a hanging planter, but it's actually a sophisticated method used by the Incans and their empire for recording and communicating information. Quipus relied on these combinations of knots and strings and spacing and placement to encode meaning.
Kind of like a database or spreadsheet that you wear around your neck and on your chest. An information storage system, absolutely. So, how did it work?
The locations of the knots and the colors of the threads and even the connections between them all meant something. Especially trained officials known as keepers of the knots exercised governance by maintaining their use and interpretation. So, if you had the quipus and knew how to use it, you had power.
Aha, like ancient DBAs. And I don't mean the ones that can remember using DB2 on the mainframe. No, right. But power through data hasn't changed in well, maybe ever. When was the quipus used?
Well, the use of knots and strings goes back at least 4,500 years. The quipus were specifically used by the Incans as recently as 1500 AD. That's the 16th century.
That's some pretty good staying power. Standards were established and passed down through generations. So, there really is nothing new under the sun.
Maybe not, but while society's use of data isn't a big deal and it's not new, governance and separating those duties is a big deal. So, back to data as power. And as we've said, data is anthropology. Threads and knots as a database. I hadn't thought about it quite that way before.
This is John Ladley and Mark Cooper. Welcome to the Rock Bottom Data Feed.
I feel like I should call for Mr. Wizard. Or Mr. Peabody and Sherman. Exactly. But if we're going to keep saying that data is anthropology, then maybe we should talk about some anthropology. And archaeology. And archaeology. It's all about people and modern organizations don't have the discipline and seem to struggle under the misconception that any of this is new.
And it's not. None of this is new. We've got more examples to share throughout this series. What's on for today? In this episode, we have the second half of the interview with Stan Christians, our famous game show, and another data poetry slam.
Sounds like a full plate. Then let's get started.
First of all, this monologue is coming to you from Ireland where I'm giving a talk in the beautiful Republic of Ireland. I'm not in the studio, so please excuse any shortcomings on the audio side. This podcast is about data, sure, but it is aimed at everyone. Yet I still get pushback from leaders and leadership that data is none of their business. They have people for that job. And I have said that this is a poor course of action. Mark is going to talk more about this and specifically about some examples in today's dispatches. But right now, him and I are pointed in the same direction for our episode today. So, back to this poor course of action. I'll be a bit more direct. It's a violation of leadership accountability. And I have seen in some cases, it could be considered malfeasance or even worse. Scenario one, for example, a CEO looks at financial numbers, the numbers aren't headed in the right direction. The CEO huddles with the CFO, they both roll up their sleeves and work into the wee hours to get to the root of the problem with the financial numbers. They work hard at it because the numbers are important. Now, let's look at scenario two. Marketing tells the CEO that the last campaign failed because it couldn't get us the right data. So, same store sales continue to decline. We can't understand our customer. The CMO sends a note to the CIO to fix the data and they move on. Well, the problem is they both need to have the same outcome, which is the first one. And until we have that, we're in a lot of trouble. Many organization leaders moan about data, but rarely roll up their sleeves and wade into the fix. And it's because they are not held accountable. As in throat to choke, as in you're going to lose your job if this costs us too much money. If it's the financial numbers, they can lose their job. So, data numbers? No. Big and important things will happen from bad data. And AI takes that same bad data and now will create even worse results. It doesn't take a rocket scientist to see that data should be considered a topic at high levels. Leadership must learn more. They are beginning to and I hope that trend continues because remember, it can get even worse. But Mark will talk about that. Oh, be nice to each other.
We now continue our conversation with Stan Christians. To hear part one, look for the fuss about Stan Christians at rockbottomdatafeed.com or wherever you listen to podcasts.
Yeah, I'm going to throw something out here that I have not heard and this is what I teach and you know, okay, so it's why I teach it. The fifth thing you mentioned, which is the consumer and who you're talking to, that speaks of what I call a data supply chain. Okay? And I take it farther than that and I think the education should. And that is that not only do you have to worry who your customer is in your organization, but organizations are now interconnected by data at a level that we never even imagined. So, we're already always crossing firewalls and distributing data. So, that's one thing I really hammer on. The other thing is and I know this sounds super elementary, but I have found it to be effective and I'd like your feelings on this, Stan, or if you've seen this or not. What is data? What are we talking about? Because I have had CEOs in the room go, 'Well, yeah, data, all my transactions.' No. And then when a CEO says, 'Well, just give my marketing guy access to all the transactional systems and he'll have all the data and you guys can just go do whatever you were doing again out of my hair.' Which we know that doesn't work. And explaining that stuff to them sometimes, have you come across that at all or am I just out here and I'm out in the weeds going, 'Yoo-hoo, here's another crackpot.'
Could be true, right, John? But yes, I've seen it. Data is, I mean, we're getting to a philosophical point, you know, what is data? In its most raw form, it's some sort of observation, a sort of measurement maybe of the world. But if you've never worked with data, which maybe most of those CEOs never have done directly, they can think of it as a vague, very abstract concept, which has been sold to them many a time as a silver bullet. You know, we'll predict the future if only we had this widget around all of our data. But once you've worked with data, be it through wrangling a SQL file or even in a spreadsheet, then you know, okay, data can get kind of messy very quickly. And to get from the mess, to get out of the mud into that insight and action takes a number of steps and takes effort. But that connection is usually only seen when you have to do it. Then you feel, okay, this is kind of messy. Because if you look at data as only your transactions, that's a very myopic view. Data is also the user interactions that we have with our application or what is in the log files behind it because that hides some sort of pattern behavior or process repetition. But data is not just within our walls but also outside of our companies' walls, our partners' data or external data sets we may be able to buy and then merge with our own to get a better insight of our customers or products or what have you. And then for the ones who are really switched on when it comes to that topic, they will actually rebuild their product, their service, their industry with the purpose of collecting more data that didn't exist before. The classic example there being the self-driving car. It needs to be a car that can measure the world around it. Otherwise, it can never be self-driving. But that data didn't exist, so you actually had to build or rebuild the machine and in this case the industry for that data to come alive and to exist. And then all the use cases like self-driving built on top of it, but probably many more like new kinds of insurance.
So speaking of cases, you've seen a lot of companies and you've seen it from the sell to them and their customers and so you get a longer view perhaps even more than a consultant does. Is there any examples, and you mentioned that you have several that have moved towards business as usual type language and stuff? Do you have any, because one thing we hear from our peers is, you know, and you go to MIT or Harvard Business Review and you don't get a published clear start-to-finish data management, data governance, data strategy case study. They are few and far between. Do you have any examples and you can obviously redact names and things like that, that really stand out as someone who has tackled this world of data, better data discipline, data management, data governance to use a tool or two, yours, somebody else's, whatever and hit the ball out of the park, really did something good for themselves. Is there anything that stands out to you in your experience?
One company? I think there's a number. Well, there's a number of them. I can even start scrolling through our website with the public cases listed there. And they're in all industries. They're not specific to just banking or insurance. They're in all industries even going into cruises and pharmaceuticals and gambling and energy and governments. But I think John, with respect to their success end-to-end and having reached business as usual, I don't think any of them are there either. I think what you're seeing is that they have a good initiative which may be a good start, which may even be a good 3-year initiative and then some of them are even beyond it. So that you could argue they've gotten closer to that business as usual. But then there's organizational changes or what have you, how the world works. So there's always external disruptors that can come into play. So the way we look at it is that this business as usual is measured mostly by adoption and we ourselves measure that through very simple, monthly active users or some other similar type of metric whereby you see, okay, this organization let's say it has 100,000 people and now there's 10,000 or whatever monthly active users. Well, then you're 10% of the way there because really what you want to do is you want to catch them all. I mean, to a various degree, not everybody has to be a data curator or something, but they have to have some touchpoint with the data process that you want to capture. And that for me is a measure of how far organizations are at business as usual. How many of the people touch the data processes?
Or that's the logical aspect, yeah. When we take it to the money asset, right? We look at, you know, how business as usual is the money asset. Well, then we would measure, for example, are people putting in their expenses? Are people doing their budgeting? And then you measure how much of the population is properly doing their expenses. And even there, although the money asset is plenty older than the data asset, even there you will find different degrees of business as usual because people just don't always do a good job.
I like the idea of moving away from traditional metadata and all the baggage that comes with it and thinking about them as context and control engines. And looking at it from that perspective because it's more than just semantics. I think it really is a change in mindset and making those things business as usual.
Yes, and you touched the magic word for me, Mark, right? Because when we did start the spin-off of Collibra, we were doing that research on semantics literally in the lab. The lab at that time wasn't like a chemistry lab, but there were just computers. But yes, semantics is making a resurgence with this AI phenomenon even. It's becoming a better word again. It used to be a dirty word, the big O word included in it. But now it's, I would even argue it's getting a sort of hotness again.
For the, what's the O word? Because we have people of questionable character listening to this show. What do you mean by O word?
Or speaking, but O, I mean ontology.
Yes, I know you meant ontology, but I was just hoping that someone didn't write us a letter that said, you know, that we were talking filth.
Okay, so only data filth, but we have a data quality solution to clean it.
I was going to say if we have dirty data, we can fix it.
All righty. Yeah, well, that's not what I mean by dirty data, but that's strictly another conversation for we hit rock bottom, John. Yes, well, that is our mission here. Welcome to it. Welcome to rock bottom. Mark, before we wrap up here with our standard questions, you have anything else?
Now, you mentioned earlier that it's clear that we don't need to explain to people or at least most people the importance of data governance when it comes to AI, but the actual doing it still seems to be a challenge. Are you seeing the use of AI facilitating data governance actually getting done? And what about the use of the metadata? You know, I mean, it's one thing to have everything all nice in your repository, but then it's another thing to actually use it to drive business change and also the corporate behavioral change in order to capture and again use the metadata in the first place.
Yeah, I think this new AI wave, it applies to all jobs, all disciplines, all industries, so including also, if you will, the data governance profession. And there, for example, I'll give you two examples there, Mark, how it's helping and I'm sure there's many other ways, but we always worked with classification to say, you know, you have your physical level and then you have your conceptual level and you can classify your table or column or what have you to your logical model. Because large organizations, you have architects and you have enterprise models. And then we discovered, well, they don't always have them or they're not always accessible, what have you. So we said, okay, why don't we give you a human-in-the-loop semantic model type agent that will actually help you make one on the fly iteratively. Okay, it may not be perfect, but it'll be there within minutes to a couple of hours of work while you're doing the classification at the same time. Right? So that's an example of how that helps. Now, in a broader sense, I would say that what you're seeing with AI and governance is it's pulling governance to become more important and more necessary. But at the same time, Mark, I'm seeing almost echoes of the past. Again, if you go back 10, 15 years and you touch on the data governance topic, it was known, but it was not always properly executed upon. That changed. The data office became a business function or a back office function. The chief data officer became a job that actually existed and started to grow in numbers. But in the early days before all of that happened, when the data governance topic came up, it was about, well, what do we need to do? How do we need to do this? How do we measure that we're making progress in it? Who needs to do it? When? For what reason? By when? And so on and so forth. So all of those questions, I see them now resurface with the AI governance topic. Where people are asking, well, what do we need to do? How do we need to do this? By when? Oh, there's a regulation. No, we're doing it for other reasons. So all of that same organizational uncertainty is sort of echoing with the data governance uncertainty from the past. So I think we know as organizations what we need to do. I think we know what governance means for AI and data and where the overlap is. But I think the challenge lies in the organization figuring out where and how to do that. And I think the challenge also lies in because the underlying technology has a rapid evolution. You know, last year it was all prompts. Now it's all agents. But essentially we are going to see over the next couple of years new governance requirements because of that rapid technology evolution. But the technology's evolving, the hot topics are evolving, things keep moving forward, the data challenges remain.
Yep. You still have the issue with the dirty data, the poorly understood data. And you know, I think the AI is helping in terms of generating the metadata or generating the models, you're just generating those artifacts that are really, you know, as John describes them as the blocking and tackling of the data business there. But it's, but they're still, it seems too often, continuing not to be done. And how is that happening?
That's a skill we have to learn, Mark, right? Because that data asset organization on the business side is such a young discipline compared to, for example, the IT asset and the money asset again. So if you look at it, it's young. It's maybe in its teenager years, about to go into university. So I think it's the rebellious time. Yes, it's sort of getting there and it's getting into that maturity curve. You can almost smell it at some point.
