You're welcome. My pleasure. So according to a Harvard Business Review article published last month, it says that the job of the data scientist is more in demand than ever with employers and recruiters. AI is increasingly popular in business and companies of all sizes and locations. And they feel they need data scientists to develop AI models. By 2019, postings for data scientist on indeed had risen by 256%. And the United States Bureau of Labor Statistics predicts data science will see more growth than almost any other field between now and 2029. So given that statistic, and given that data, I would like for you, Stefan, to give us a bit of a short abstract on the industries you've worked for, so far in your career in this field, given these statistics, in your experience.

Stefan Lorimer

Wow. Yeah. So data science is such an interesting and evolving field. Because if you think of what we're doing today, and why that demand exists, it has, you know, its foundations in math and statistics that go back to, I don't know, the 60s or so, a lot of the models that that are worked on by data science, you can actually find in industries that you might not be as aware of or consider to be data science. But you know, at the core data science is the analysis and modeling mathematical modeling of human behavior primarily. And you can if you've ever used, you know, radio, or done recordings like these, or looked at how trajectory calculations can be done with the space programs, you're gonna find early examples of how mathematicians laid the foundation for the type of work that you see today that data science is, you know, asking for future recruits score. And that's kind of interesting because, you know, in my career so far I've seen it emerge in entertainments. You certainly see it apply to a lot of the software developed for music, film and television which you know, in my younger years I spent time in I spent time in not for profit software, where we were developing features to support individuals where you had a basic rule sets based on on data you're quite an algorithms at that stage. I worked in environmental science for water data management software, where algorithms and statistical models were measuring if you know the, the flood level of rivers would increase or decrease in pollutants didn't that water met regulatory standards gaming, of course, which, you know, I've spent a lot of time and have a wide variety of people that I got to work with some really great professionals for data scientists and some of the cutting edge in the field. And now forestry were some of the data science models are going to establish what the best way to grow the most trees are and to limit To the amount of trees you have to cut down in order to meet the world demand for wood fiber and wood products. So I've seen data science in all those fields, including many others in the functions that I have been a part of.

Maureen Farmer

You've probably seen it evolve as well.

Stefan Lorimer

Absolutely. I mean, the thing that's really interesting about it is that somebody, somewhere has a model that gives you value. So let's break it down and, and kill some of these words that are probably fairly loaded, right? What's a model? What are statistics? What is data science, a model would be something where you say, well, we can observe that, you know, of all of the people who use banking software today, of the people who access their bank accounts online, right over their phones. That is an expected behavior for each and every one of us. Now, where is your mobile phone accessing its banking information from? Well, your house and your region. And so a computer can basically have a model that will say, if Maureen or Stefan or any of your listeners access their bank account via their phone within the expected area, don't prompt them with security questions. But if you go to one state or one province or one region away from where your phone is likely recorded, accessing, you know, your bank account online, that's an unexpected behavior. It could be though, that you travel every Christmas to see your family, as many of us do. Or there's another religious holiday or cultural holiday that you travel for, and you go similar, regular, so the interesting thing is going to be after you've traveled two or three times to that other region. And you've been prompted with security questions. Now the model is going to pick up that that place where your family lives in, you know, some remote area of wherever is also expected within that model. Right. So these models are, are the basis for rule sets that the system say, Is this a valid thing?

Maureen Farmer

This is machine learning, right? That's what what machine learning is referred to?

Stefan Lorimer 

Well, machine learning and artificial intelligence are very similar in the sense that one is more of a simplistic manner of looking at, you know, what are all of the locations that the model should consider that are valid or invalid. So think of it as like a classification, it's like you're working with a young child who's observing, crossing the street, and you're standing there, and you're giving them you know, instructions, like, Don't cross when the light is yellow, wait until you know the light is green. And when the light is green, wait five seconds and look, you know, left and right and make sure there's no cars that haven't stopped, right. And so those rules, and those boundaries are something that a child will learn over time, what the the standard expected things will be that you know, fall within those rules that you provided them, that's machine learning. Artificial intelligence would also interpret, you know, the many other variables that are going along with it. Like if I also have the ability to hear a sound like a chime that goes when the the walk sign goes up, what what's the different variations of the walk sign? And that all depends on the on the data that you're feeding the artificial intelligence. And so it can actually learn and interpret many different other signals, as well as the rules that you have provided to then say, well, I don't if I if I can't see, but I can hear the chime, that also means that it's safe, if I can't hear any cars with their engines, you know, accelerate, right. And so artificial intelligence can actually interpret the additional data that goes along with that situation much as a child could. And there's many different models that have been designed to do that, that learning pattern. So machine learning is basically a more basic, it's a more basic way of of interpreting the signals, that you can feed a model. And the reason why it's evolved over time has a lot to do with the computing power.

So you know, if we were talking, say, 2030 years ago, we'd be talking about how many hard drives and microchips you have in your home? Well, now, because of cloud computing, there's no real upper threshold, it's just a cost, a cost of doing business, you can put many millions of processors or processing power, and so you can accelerate the speed of that learning. So 20 to 30 years ago, you know, it would take many months for that mathematical model to interpret all of the different variations to then be able to say, when I hear a signal, I can also cross the street. If the walk you know, lights turn on and off. Whereas now it might take three minutes. And in fact, if you go and look at some of the really cutting edge stuff that you can find from companies like kindred AI, from Amazon, from, you know, like car driving the Tesla has done a lot of the things that are going on there is They're taking video game environments. And they're making millions of copies of the immediate environment like the one I'm describing here about crossing a road, right. And they're running simulations in parallel of millions of different variations. So you can see robots like spot the dog, you've probably seen, like the really fun dancing videos of spotted dog that was run by created by Boston Dynamics, they can now go from a robot that has no learning capacity built up to three minutes later, having the sum total of all of the spot dog robots in history so far, in just over three minutes.

So you know, it's evolved in pace and ability because of the computing power in the cloud. And that's a good thing, because you definitely don't want a robot learning, or, you know, an AI or machine learning model having to learn for each and every one of our experiences. One of the promises of the evolution of this is that, you know, the model is sort of going to, to learn and, and follow human history and only get better over time, you're going to see this in the examples of the Google AI that the defeated the best GO players in the world, it's now almost impossible to beat them. And every time that a human does learn, and to be one of these AI, the AI will then become, you know, so much better in a short period of time that you can't beat them the same way twice. And they'll interpret the waste learning.

Maureen Farmer

Oh, no, it's just completely overwhelming. And there's a lot of conversation right now about the ethics of artificial intelligence. And this is not this is not what this podcast is about, or what this conversation about is about. But I wouldn't, I'm wondering if you could make a couple of comments about the ethics around artificial intelligence and data gathering and how organizations are addressing a compliance around that practice.

Stefan Lorimer

So, on the topic of ethics, privacy and security related to artificial intelligence, what's really important to remain focused on is the current battle for minorities and of the different populations that exists today, to gain recognition and for people to not be abused by companies, right, that those are the underlying concerns of them. And those drive the questions of ethics, privacy and security today. So why is this become a big deal? Because if you go back to some of the early science fiction novelists Isaac Asimov is an easy one, because he established you know, Laws of Robotics way back when aI had laws like you can't hurt human beings that were created long before we ever had artificial intelligence to rely on. There is an innate fear that artificial intelligence will gain sentience. And you probably have seen this that, you know, a Google engineer claimed that, you know, an artificial intelligence in Google had achieved a level of sentience, a certain age of a child, and therefore, it had a lot of implications, believe it or not, that's not really where the ethics and concerns come from, it's that there will be an increased or more pronounced reinforcement of powers against against humans. So you saw this because, you know, there's a lot of issues to do with the abuse of data. Largely, the example for that would be Facebook, Facebook has probably been the biggest driver of privacy law and in Europe, as a result of misusing personal data. Now, if, as an example, you have a situation where you can learn where someone lives, and then you can change the pricing of certain things online, that's an issue, which would be very problematic for you. And I, because you could change the online price for say, fresh fruits in the future, based on where someone lives. That's, that's a human rights issue. At the end of the day, you know that an abusive business practice would be driven by an algorithm that would favor one person over another on a food price is something that, you know, is rather unconscionable, or at least creates fear for us.

The other aspect is, of course, security. Now, if the example that I'd spoken about earlier around how model can interpret, you know, where you generally access your online banking from, there's fraud and security that go along with that one of the biggest worries of today is identity theft. Well, you know, so if an AI can be used to interpret your behaviors and imitate you, that's also a rather big deal, because it can ruin people's lives. I think the biggest issue around ethics, however, comes as a result of how AI learn and interpret who it is that they are dealing with and decisions that are made as a result. And you've seen that ethics around this largely favor. Google has had a recent I wouldn't call it a scandal, but you know, there's been a big uproar about AI ethicist and and what they had found with the algorithms biasing towards certain populations and minorities. Yes. And this is very fascinating. Because you know, at the root of this, you said, Okay, well, we are fearful that this is reinforcing already. The stigma is the labeling the taboos, the unacceptable things that we're trying to rebalance in our society today with diversity, inclusion, and equity. And in fact, what it is at the heart of it is that it's a data problem primarily, it disproportionately so certain communities do not create enough data to train a model to be sensitive to their behaviors. And so if you know, you have one group of people that produce more data than the model is going to bias towards them. And you can see this in the access to technology, if you have one group of people that has less access to mobile phones or computers. So there's less data about their behaviors, then a model, which is a statistical model that favors towards averages and normals, and all kinds of stuff that we use to, you know, say, Well, what is the general trend of something, if there's more data from one group of people, then you're going to have an issue where basically, there's a bias in the model.

So, this is an oversimplification, of course, and it is to point out that, you know, like, there's, there's bias in all of the machines that we have, but until we can make it so that the the AI that we build, considers the amount of data considers that not all people are same, considers that on access to technology are the same, we are going to have these great ethical issues about just, you know, who is an AI learning from, at the end of the day, right. So if you're, if you're always, if you're always teaching a child from one perspective about one group of people, then they're going to have inherent bias about the perspective on the world. And you see this in history books, and he does see this in the way that people are, are raised in certain education systems, and it would be the same for the artificial intelligence.

So, there's an evolution there. And the most important thing to note is that you have to be considering what it means when you're creating artificial intelligence, that not everybody that you are learning from, is the same, and not all the data is the same. And so you have to be able to interpret that difference in that variation that makes humanity so special in the way that you're designing your artificial intelligence, so that it does not favor certain individuals and groups as a result.

Maureen Farmer

So that has been very helpful for me, Stefan, because you have just simplified, for me, a very complicated issue that is in is in the media, it's everywhere, and the discussion around it, and you know, your perspective on what we've already had already have already had inherent bias in our tech in our textbooks and the way we learn and the the lack of a multi dimensional perspective on society, whatever it might be. So this is incredibly helpful in my own understanding of data science, artificial intelligence. And I know that you have recently not recently, I guess, looks like it's been a while now took on a new role where there were some really complicated integrations of products. As a result of a bunch of mergers. I would love to ask you as a business leader, and as a data scientist in a product management expert. What do you see as the main challenges to post merger product integrations?

Stefan Lorimer

This is a fun question!

So, let's reach into a little bit of technology about how this works. And let's reach into a few metaphors and examples to make sure that you know, this is understandable. And we don't get lost into the technical details. It's very easy to get into what some of the more experienced professionals have called the how hole, which is fascinating. It's like how do you solve this actual problem with technology? At the simple heart of it, you know, the things that you're going to see in any corporate leadership and corporate coaching exists, you know, like in a post merger situation where you're looking at products, right, the comment that you hear is, Well, our teams exist in silos. Well, we haven't integrated yet. Well, we don't have shared processes. Well, we don't have shared culture. And those are the things that you're going to address on the people side related to, you know, when you acquire a company and how you bring them into part of a larger strategy. And you know, what's fascinating about that is that you can have really successful companies and really successful products individually, but the same thing exists on the technical side of things when we're talking about what do you do post merger or acquisition, the first thing you have to do is you have to look at what is shared. And start from there. Right? What is the common ground. Now with products, the fascinating connective fabric and tissue about, you know, two different products is that they share something in common right out of the gate, that's users. It's people, right? People who use software.

Now, everybody can relate to this, we all have user accounts with passwords. So let's say you have a company, two companies, and there's one bigger one and one, medium one, and they're both successful. And, you know, it wasn't a crisis that caused the acquisition, they decided that it was the right thing for both companies to merge. And then you say, well, we need a joint roadmap for these two products to go successfully together and become something together, the first thing that you're going to have to look at is the underlying data structures, about the common things, and then the underlying common technology that you're going to need for those common things. What are the common things, our user accounts, our password systems, our shared security, how data is imported and exported and shared. Those are systems that are common to both in the same way we have, you know, groups of people who exist in silos until you have a process where you have a joint quarterly meeting for everybody, and you start talking about culture, and you start talking about shared HR processes, you don't have anything to start from. So before you get into well, the two software's overlap, and they have, you know, the same abilities for the users. And so we should have one instead of two development teams working on a common feature set, you know, within their age products, you have to take the data, and you have to figure out how you're going to map it into, you know, a shared system. So you'll do things like you'll say, Okay, well, if we had two products, and we had two development teams in each of the companies, and they were both working on identity, user profiles and security, how do we make one team and that one team is now going to be responsible for users, not authenticating and security? And if they do that sort of stuff? How do we have to change the two products that we originally had not one to, to have a common system? How do you change those two existing products to now connect to that shared user system? And what is behind all of that, and makes it sing is the shared and common data structures.

So, getting back to this question of, you know, data science, I mean, this is not really a data science question. But one of the major obstacles that consumes data scientists time is when they're having to create these really complex models that do really amazing things, but they're spending 60-70% of their time cleaning data. You know, they're looking at six different systems and saying, well, Maureen, or Stefan, or any one of our listeners are sitting somewhere, and they're showing up in systems six different times what's common their email addresses? Well, now we get into one of these ethics questions right about privacy, are you allowed to look at their email addresses? Are you allowed to use their first names to say that Maureen is Maureen, you know, when we, when we sign these, these agreements, these end user license agreements with companies, there are legal restrictions. So this is where you wade into the legal issue.

So, at the heart of post merger and acquisition, steps that you need to take is, first and foremost, you have to start talking about data governance and data structures that are common because data scientists can't even start their work, unless they've cleaned, labeled, identified, standardized the datasets that they're pulling from all these systems, let alone just the regular engineers that are going to build the next version of that amazing software that gives you great abilities.

So you know, how do you create common...the user in themselves? How do you make sense of the data? How do you get a common technology stack that you're building on? Not three different types of technologies, but one that you share? How do you make that efficient? How do you communicate the same language? A user is a user? Or are they? Are they the same? Legally, are we protecting those users? Because that's the promise of the software is not only that, you get these amazing magical abilities. But you, but you also have it done in a secure fashion, it's not going to allow somebody to take that user information and say, call up your bank account and change your password because they know your email password and the most common security questions that you get asked.

So, it's about the data at the end of the day. And then it's about the systems that those are wrapped in. And then it's about the silos of the people that go along with it, because they are going to be making those changes. So if you start with what's common, establish a common framework. It's like building a house

Maureen Farmer

You make it sound so simple! So, I have a question for you, because I'm very, very curious about this. Can you predict when you go into an organization, how long a milestone is going to take? So say, for example, the examination of the different data sets? Do you know that going in? Like, can you predict that? Or is it sort of like, well, it takes as long as it takes, like, how does that work?

Stefan Lorimer

Yeah, it's a fabulous question. So let's rewind to a statement earlier that you made about how data scientists are so in demand. Right? Right. I mean, the difference is, do you have people who know what they are doing? Right, because if they do, then there are very fast ways of doing that. And they will be in the experience that, you know, a data scientist or a data engineer, or any number of the flavors of roles of data management taxonomists, you know, like, there's all kinds of really fun titles and, and roles that get into managing an identifying data.

Maureen Farmer

Right, I know. I know something about you that you have a unique ability to shorten timelines for very complex business integrations, I know that and, and I don't want to you to give away your IP or your secret sauce. But I, I really would love to go back to that question for a moment. Because the ability to take a new product integration, whatever that is, to market, and a period of time is a predictor of profitability, and I don't want to make it about profit is about a whole bunch of other things. But as a simple KPI is a simple measurement for that process. You know, I'm very, very fascinated by the that challenge of, you know, walking into an organization, after there has been a merger of two products or two companies, or both, presumably, there will be both. And, you know, in that governance processes, or do you know that when you walk in their middle, and I'll let you answer the question.

Stefan Lorimer

I mean, the short answer is generally yes, because of the developed practices in the teams that are doing the work. And so not to give away the whole process, because there's a lot of nitty gritty details that ultimately come out of this. But you can kind of understand how much time it will generally take when you look at the teams that are working on any part of the technology that you're using, and whether or not they have best practices and whether or not they understand the challenge, the underlying challenge that comes from having data managed correctly.

Maureen Farmer

When you go in, do you do like a gap analysis with the competencies of people across the teams? Is that one of the things that you do?

Stefan Lorimer

Yes, that would be one of the things you can do. You can approach this from sort of two different angles. One is you can do a gap analysis of like the types of roles that you were seeing, because there's more generalist roles and there's more specialist roles when you're talking about data scientists, you know, you're starting to see their specialization in data science, of course, but when you have groups of engineers that are generally br generous, I mean, they're building entire web applications, right? They don't have the time. And they have not focused on the specialization within the data domains themselves, you have engineers who specialize there. And so when you look at the people, manual things, and you just look at the roles, and you understand what their general responsibilities are, you can know whether or not there is enough experience in that team to be able to address the questions. So it would be a good example like this, you know, like when we're talking about building a house, and you're building that strong foundation at the bottom, right?

One of the predictors of like, you know, building a house going off the rails is whether or not you have dependencies identified. Now, if there's a crack in the foundation of your house, and the people that are building the house come along and say, Well, the first thing that we have to do is we have to build the walls and build the structure on top of the foundation, they don't address that crack, then at a later stage, when you're trying to build the things on top of that foundation, you're going to have to address the fact that there's a crack. So if you look at the estimates, and you say something along the lines of You know, there's no cement in the estimate of time, or there's no one who knows anything about cement, you know that there's going to be this knock on effect and an extension of the milestones in your roadmap. And it's similar when you look at how software would be developed or product to be integrated, or you know, a company would be merged.

The reality is the foundation that everything gets built on is data. And if you don't have somebody specialized to handle the standardization, the harmonization, the association of the data structures to integrate via this is how it's generally done, but like, you know, API's, SDK's, common data and data governance models, normalization or some other terminology of how the the data entities themselves, you know, whether it's a house or a floor or a tree or a game or water. If you don't know what all of the data is, you need to build that structure, inevitably, there's going to be a knock on effect, and it's going to extend the timeframe. So you can come in and say, okay, well, we acquired this company, now we're going to merge their product line into our product line, go and build it.

And you know, you're gonna end up with a new app. The app will not integrate into the shared login experience. Why? Well, because it doesn't access the same user login, and it doesn't access the same profile of your user. Well, why is that? Oh, because it handles first name, last name, email and passwords differently. That's a that's a data issue. It's not an engineering issue in the sense that, you know, you have to build a software around it. It quite literally is, is the email held in plain texts? Right. So as you typed it, this is a frequent problem of security that has come up over the last however many years, right, so there's this thing about is data even handled correctly? You know, is it is it encrypted, so that security, things can come along? Those are all of the things that are extending timelines, you're going to have a great app, and you'd be like, Oh, this does exactly what the users need. It's fantastic. And the first thing that happens when you go through a security audit is they say, No, the underlying foundation of your security and your data structures will allow it to be hacked. And now what are you going to do? Well, lengthly fixes and quality assurance runs, that's going to extend the timeline, because the underlying foundation of the data is not safe, secure, standardized, normalized, accessible, right.

So any way that you look at it, the process, and the things that you're going to tackle are about ensuring that you have a common data set that is documented that everybody builds from for a common foundation. You know, today, the ability for us to stand up apps have been commodified. I mean, you've got kids who are as young as like, you know, eight or nine, as soon as they gain computer literacy, being able to stand up mobile apps and ship them on the app stores. Right. The construction of the software is not, is not a challenge and obstacle, it's the data literacy, you know, in when you when you when you see the the development of apps, it's all very nice, but you don't see people talking about, well, what is the type of the data types within a database? You know, that's not something you're teaching like a 10 year old when they were learning to code that is the foundation have to build on? How do I store? How do I process? How do I manage the data?

So, those are the kind of knock on effects. It's like, if you know what to look for around the data integration itself, then you will address how to get to market faster, because building the actual apps and the products is actually relatively quick, if you've got a good team, but addressing the underlying cracked foundation between say, you know, like the left side and the right side of the business or two companies that have emerged, you have to, you know, fix that gap and communicate between those two.

So, it has a lot to do with addressing the complexities of the data structures and the documentation, and the merging of the datasets into being one unified data set that ultimately makes the difference between a predictable timeline on the roadmap, or one that will see many iterations and extensions. There are of course, many other variables. That is an oversimplification.

Maureen Farmer

So, you have this data science expertise, there's no question but you also have what I just mentioned, is this ability to predict timelines. So where did you learn that from? Is that just an innate skill that you have? Or did you learn it from experience? Was it something that you knew how to do as a kid? You were really good at estimating? I know, I know, growing up, one of the things that our teachers always wanted us to do is like, Look, you need to be able to estimate things, you're never going to always have the time to have it down, you know, to an absolute decimal point. And I know in project management working in corporate as well, Project overruns, Project overruns, that were always a big deal in corporate. And some of the people that I have the privilege of working with today, some of them have actually been able to under promise and over deliver, they're actually going they're actually able to achieve these milestones ahead of schedule and ahead of budget, because I'm very curious about a about a business problem that a lot of organizations have in this complex post merger world where we have products that we're bringing together, how do we stay or how do you stay on budget and on time?

Stefan Lorimer

The short answer is I have had a lot of different kinds of jobs in my career and learned...the best way of putting it is trial and error, the humanity behind those numbers that you need to hit.

Maureen Farmer

The humanity behind the numbers. I love that, let's talk about that. 

Stefan Lorimer

The humanity behind the numbers when you're talking about timelines and estimates are a lot to do with how you pull off coming in, you know, under budget and within timeframes. The first thing to note is that humans established trust. And trust is also something that is dangerous because it leads to groupthink. You know, and you might hear about groupthink. If you haven't studied groupthink in the topics that you cover in your podcast, it's a good one to crack open. So, when you have a good, strong, trusting team, but you don't have opposing opinions, sometimes the strongest voice or the most trusted voice will make a statement and people just go along with it. Like, yes, that will take one month, we're good.

Maureen Farmer

So, how do you overcome that then?

Stefan Lorimer

So, the first thing to note is that I encourage—it's not dissent—but I try to build teams, where the first thing that we're going to do is ensure that we've got opposing opinions that show up at the table for everything that we do before you commit to anything. So there's this really big drive for, you know, DEI initiatives, diversity, equity inclusion, right? Well, there's a much better way of looking at DEI issues, which is like, do you have enough different knowledge and perspectives to be able to question that statement that is being made by any one of your team members, to have all of the different analysis done quickly, about how long it would take, and what might be the dependencies and what other variables might come into play, when you're going to be putting forward a timeline or a cost? The more diversity that you have, the more analysis that you've got, we've been talking about data scientists and artificial intelligence, one of the most powerful processing engines we still have access to today is the human brain. So if you have, you know, we can't communicate like a computer can but if you got a group of like six to eight people who are good with communicating each other, when you go in and you say, How long could this take a project manager who doesn't know the domain that will ask that question, you've got a team that's made about six to eight people, that can give you different opinions, and they are comfortable with each other, and they can communicate effectively with each other. And they can quickly get to that answer. Now you have the framework on which you will get into all of the potential risks that might influence any one major tasks or group of tasks to be able to execute on a timeline.

So, that's the first thing, that humanity behind the numbers, that's the thing that really, you know, makes teams really effective at coming in under budget, and being able to under promise and over deliver is that they're not going to know it, it's not one person has the experience is that the collective intelligence of a group of human beings in the right kind of environment with the right kind of leadership to feel supported to speak up, is going to get you the fastest responses with the most variation of that answer, then you get to the fun problem about the numbers themselves. Right, so people will speak up. And they'll and as long as you have, you know, dissent, you're gonna get a wide range of above and under what your range is, right.

But now, there's an analysis component, which is: Do you understand the roles themselves? Do you understand the tasks and the technology themselves enough to be able to take with rigor, the question of, if you build a new feature, if you build a new model, then what are every single task that you might have to do? Now this falls into another task of humanity that you know, we used to communicate with each other is this oversimplification of things, I'm doing it today. We want to be able to communicate quickly, and we want to be able to communicate simply. Well, the truth is that simplification is the enemy of good timelines, and good budgets, because the details...assumptions...exactly.

So the humanity behind the numbers is kind of the first question. And the second one is also kind of the humanity behind the questions on the opposite. It wasn't just that because of that simplification, we're also going to say, oh, yeah, you know, take me two hours to do that task. No, the truth is, is going to take you a day. And so we oversimplify, and we use optimism, and we use all kinds of things to highlight the inherent bias towards making someone feel good about the answer that we are giving them when in truth, it's not going to take you two hours, it's going to take you one hour to prepare. It's going to take you one hour to start the work. It's going to take you one hour to correct for the work because you know, you learned something in the process of this new task that you're having to do that was not disclosed by your co worker or the documentation. Then you're gonna get into having to communicate with it. everybody what this changes where if you're a good team member and the knock on effect of it, then you're going to complete the work in the afternoon before you finally turn around and sort of communicate out to everybody the impact and just the timeline.

The truth is, we all do this inherent bias towards not conservatism, but optimism in our numbers. And so the rigor in which you take first of all that variation, and a group's abilities to state what the potential ranges, and then you really dig into the details of knowing exactly what the knock on effect and your ability to create parallelism, not an assembly line of one task after another handing along the line. But the most intense version of how many tasks you can do in parallel that are non dependent. There is entire fields of study in business to do with supply and logistics planning, Tim Cook, the CEO of Apple, in fact, this is his area of specialty that they're not really accompany today of, you know, the most innovative things, they're a company of like managing their parallelism of tasks, you know, exactly when somebody is going to deliver. So you have to then get into the science of supply chain and logistics of tasks to really be able to isolate like where there are dependencies and how those dependency have to have knock ons, and how accurate you can get with the most critical tasks, so that you take the humanity out of the numbers. And now you're just looking at the trends in how humans behave, and the variation that might cause a knock on effect.

Once you have that model, it's just a process of a third thing, which is elimination of risks, which can only really come with experience, right, you can only really look at it and say, Well, you know, if I run a restaurant, or if I run a medical group, or if I run a Construction Group, or find a private equity company, I have people who know what it means when a plan goes off track and what to look for. That's why you bring in specialists and management consultants for that kind of expertise.

So, after you've looked at the humanity behind the numbers, and after you've taken the humanity out of the numbers, now you have a situation where you have to have domain expertise, really looking at it and saying, Did you account for the fact that you need to handle security and third party security audits? And that usually takes three weeks and a staff of two, let's say, to mitigate for those risks. Did you integrate the data? And do you have a data scientist on your team? Do you have you know, somebody who's going to adjust the API's and republish them. And that usually takes two weeks, right? So it's the knowledge of the domain expertise, finally, that makes the difference. So really good, junior project managers don't have to have domain expertise, but they have to be good with people. So you can get the humanity and the range and the communication. And then you have to be able to isolate for the humanity and remove that and be able to analytically break it down before ultimately, the domain expertise shows up.

Maureen Farmer

Something tells me, Stefan, that you're really good at speaking to power because I do know, no matter how much experience you have, you have to have the courage or maybe the confidence is a better word, to be able to have a dissenting opinion on a process, especially walking into a new organization.

Stefan Lorimer

Yeah, we can get into all kinds of fabulous stories about you know, identity and upbringing here and reminisce and get nostalgic, I think we all could, right? Because this is, this is sort of like a thing that we talk about in leadership circles about how to create safe spaces. And not everybody comes from, you know, like a safe space right? So how do you create an environment where ultimate people trust that they can step into a place where they can say, well, I understand what you were saying. And here's my perspective. And it is also valid, because of my lived experience. And so, you know, the thing about truth to power is that, you know, it's quite often seen as sort of a losing battle, or it's taking a step back...

Maureen Farmer

It's seen as a conflict, or it's seen as adversarial. And I don't like the term I hardly ever use it, but it just occurred to me when you were explaining to me about dissenting opinions, there is a private equity company that I'm aware of that when they are about to do an investment in a company, Stefan, they actually bring in a consultant, they hire a consultant to blow holes through the planning. Often they'll hire that consultant on to help them with the integration or help them with the you know, the the transformation of the company that they're investing in.

So, it's a really, really fascinating topic to me and I could talk about this forever. Given that our time is almost coming to a close. I would love to ask you a couple more questions and they're fun questions.

So, I would love to ask you...what has surprised you most in your career so far?

Stefan Lorimer

Oh, gosh. Okay. So what surprised me in my career so far, is a very surprising pattern that I've seen with people. And I think we've probably all heard, at some point in our careers, something about this, but how in every work culture and every region, I've worked with how people limit themselves or create obstacles to growth, and to moving forward and to accomplishing shared goals. And there's many reasons for this. And, you know, a discussion earlier and topic that you raised about speaking truth to power and safe spaces, and, you know, having diverse teams, it all plays a factor into this. And I'm sure we all do this. In fact, I'm, I have to check myself regularly. We know when I, when I'm showing up, and trying to perform as part of my roles, right? A great speaker who's who's getting a lot of traffic these days, online, I think it was named Simon Sinek. He speaks about team members, he speaks about people, you know, he's always talking about, you have to set people first. So he's probably touches on this topic as well, because I've heard him speak about it.

The biggest surprise is how hard people want to remain stuck in a loop, where they're limiting themselves due to fear, right? And it's also equally surprising to see what people are capable, you're able to unlock them and move past that fear. And so you'll see this, surprisingly, you know, I thought this was one culture. When I started out my career in software, I was like, this is one culture, this is one person, this is one country, this is one team is one company. No, it's a human behavior, you're running into a fight or flight response, you're running into a series of things about you know, like how you see yourself and your position at work, the best leaders are able to unlock potential to move people out of the fear, and the barriers that those people are creating for themselves. And it's not that it's being done all the time, or you know that any one person is doing it, it's more when it is there, can you unlock it?

So, I've been lucky enough to witness how trust in the right environment and guidance, you know, from a leadership position that I've been set someone up to succeed. So it is always a surprise to see how these very basic changes in perception and trust and a lack of fear, a lot of growth and opportunity.

Maureen Farmer

There's no question, there's absolutely no question because when you're in an environment where you don't feel safe, you cannot be productive, you cannot be innovative, because you have cortisol running through your body, you have adrenaline running through your body. There's a lot of PS PTSD in in the workplace right now. And I myself have experienced that in the past. So I know exactly what it's like, you know, the transformation, I was talking to a gentleman in, he's an investor, he's a portfolio manager. And he's, he's absolutely brilliant. And he moved from one big, big company, it's a global brand. And he almost left the world of finance, because he thought, This is what it's like, everywhere. But he ended up getting recruited by another company in the same industry, same size. He said, Maureen, I cannot believe how different these two companies are, we're doing exactly the same function. And he said, high integrity, he said, I almost left the world of finance, I almost left, I almost gave up during my CFA exams, because I was so miserable in that first company. And the transformation in that culture, and in the relationships is absolutely the opposite. So you know, this whole idea of corporate culture of organizational culture and, and leadership and building trust, it is so important, I believe in the success of an organization. And it's been proven time and time again. So I really appreciate your perspective. And it's a very analytical answer, but it's also a very powerful answer to that question. And so I thank you for it. And my last question is around restaurants. So Maddie and I, our Podcast Producer, and I are putting together a list of restaurants for a list towards the end of the year, I would love to get a couple of names of places that you'd like to frequent.

Stefan Lorimer 

Oh, my goodness. Okay, so this is gonna be a fun question. Because of course, as you know, I'm into products and restaurants and experiences and services can all be bundled up in a product but for those who during the COVID period, couldn't go to restaurants and started ordering online. We now have many more options than just restaurants. So I'm going to take a bit of a different tack on your question. Awesome. Some of some of the best product design supplement I've seen in like the last couple years are how very small food companies have moved online and provided niche products that you just never would have expected minor variations and just common things that we have every day. And you're like, Well, how did those things work together? Yes. And they're delicious. So the first thing I'm going to call out, that's one of my favorites is a company called Coho Collective, which actually popped up in Vancouver, I would say, I don't know, like five to six years ago, it picked up on what the what's called ghost kitchens. So a lot of really small companies in the food, business or restaurant business or catering, they don't have commercial kitchens. And that's a huge investment, you can get started and you can start making food, but you need a commercial kitchen or make it viable ghost kitchens are basically like leasable spaces at low rates.

And so, Coho Collective kind of found this niche just before COVID when all of a sudden you could go online with a website, you could be delivering food all over the place. So I love cocoa collective, because it's got this really great variety of really small, you know, food businesses that are starting up, you can order stuff, you check it out. I recommend that your your listeners go and check that out, because you can probably order it anywhere globally, knowing how shipping works these days. They're doing great work. Some of the products you see coming out of small businesses are the next issue you'll see in an upscale restaurant tomorrow, and it's just fabulous.

So, you have to try is Ernest ice cream, or they got started a few houses down from where I lived in Easton. You know, years ago when I was part of, you know, startups in Vancouver. They had amazing flavors of ice cream. There's chocolates, you've never try their seasonal varieties. They use spruce boughs versus a tree so shout out to my work in a forest for these a spruce bows are kind of this lemony scented thing that we're used in First Nations cultures and cuisine and that sort of stuff. You can eat it. They'll put that in like a vanilla and it just gives like a lemony zest to a vanilla and they'll serve that up when you can have the spruce boughs coming off the trees. It's incredible what they do with flavors.

Maureen Farmer

Fantastic. We'll make sure that we have these in the show notes for sure.

Stefan Lorimer

The last place I'd recommend if you're really a foodie is anything off Salt Spring Island, it's really interesting how that company or how that place has become a business incubator for food. You know, they don't scale businesses. In fact, the general community there is one of general hostility for the larger businesses. As a result, what's really interesting is the tourism has made it so you've got you know, Salt Spring cheese companies, they're fabulous Salt Spring kitchen company, Salt Spring coffees, these popped up serve a tourist market. You can order their products online, in a lot of cases, and then they move off to kind of come to other parts and start shipping and exporting their foods.

So, Salt Spring has been a great place food products and any of those Coho Collective, Earnest ice cream, anything Salt Spring...those are my favorite food products ordered.

Maureen Farmer

That's awesome. I'm heading to Vancouver. We're doing the California coast next April—April 2023. That's the plan and all the way up to Vancouver. So I'll be going to Salt Spring Island. We have friends who live there, and I will definitely check it all out. So Stefan, it's been an absolute pleasure hosting you today. And I cannot wait to have this this episode launched. And look forward to maybe hosting you again another time.

Stefan Lorimer

I'd love that. Thank you for having me. 

Maureen Farmer

Thank you, Stefan.