Jon Krohn: 00:06
This is episode number 614 with Ross Dawson, author of Thriving on Overload.
00:27
Ross, welcome to the Super Data Science Podcast. It’s awesome to have you here calling in from Australia. I guess it’s early in the morning for you since it’s late in the evening for me.
Ross Dawson: 00:37
It is. Great pleasure to be talking to you, Jon.
Jon Krohn: 00:40
Nice. So I didn’t tell you this before we started recording, but you Ross are the guinea pig for this podcast for a new era. So you are the very first guest in this deliberate new format change. So we have two episodes a week. Historically, it’s just on Tuesday episodes that we have guests. And those guests speak about topics that are specific to data science. And then on Fridays, we have these what we call Five-Minute Friday episodes where it was typically just me on my own talking about how to cope with things at work, a topic that we’re going to be talking about today, how to Excel at work, looking at emerging technologies, looking to the future.
01:25
And I started to think that why don’t we have brilliant guests like you come on and give much better, much more detailed answers for our audience than I could possibly do. So you’re the very first guinea pig in what we’ll be something that happens more and more and more on this program. So Fridays will no longer just be five minutes all the time. They will definitely be longer and more frequently. I will be accompanied by brilliant guests like yourself.
Ross Dawson: 01:57
That’s good.
Jon Krohn: 01:57
Yeah, it’s going to be fun. So Ross, you are the author of the brand new book Thriving on Overload. It looks like a great book. It was released in the United States on September 6th. And so by the time this episode is live, anybody in the US will have access to it and I imagine people pretty much all over the world.
Ross Dawson: 02:17
I’ve told more like late October for the rest of world, but it will be around soon at latest.
Jon Krohn: 02:23
Nice. Great, well, you’ll be able to access it soon if you’re not in the US. So this book answers the question when you’re managing exponential amounts of information as part of your job or as part of your daily life. I think a lot of us in the data science profession or in related fields are dealing with a situation all the time. We have tons of new papers coming out, thousands of papers every week, many of which are relevant exactly to what we’re doing. There are hundreds of books every year that we might like to read. There are patents, there are GitHub code repositories. It doesn’t end. There’s just exponentially more and more and more information for us to absorb as data scientists.
03:07
I frequently have had people reach out to me on social media and asked me the question, “How do you keep up? How do you learn in this field? How do you feel under control?” And you Ross, are the man to answer that question because your book is about managing these exponential amounts of information and transforming that from a feeling of being overwhelmed into a sense of abundance and empowerment. How do we do that at a high level?
Ross Dawson: 03:37
Well, it’s actually the same as you, I get asked that question. I’m a futurist. I talk about just about every industry you can imagine. And so I have to keep across incredible amounts of information. So people ask me, “How do you keep up?” And in a way, this is partly a self-reflection, I observe myself and what I do and also from all of the people I know that I’ve interviewed and engaged with, people I call the information masters, the ones that we can see around as who are able to effortlessly keep on top of that. And so really the book is trying to distill that into a framework, which is useful and pragmatic. So that have identified what I call the five powers, which are the foundations for being able to thrive in this world of overload.
Jon Krohn: 04:24
Nice. Yeah, let’s dig into that. I understand that the five intertwined powers to develop are purpose, framing, filtering, attention and synthesis. Maybe you can run us through those one by one. Ross, what’s the purpose power all about?
Ross Dawson: 04:40
So purpose is knowing why you are taking in information at all in the first place. And that’s, I think, something which very few people get to consciously, where instead of just, “Oh, lots of information. That’s interesting” in the sort of way you are able to play, we are as humans attracted to information where novelty gives us a dopamine reward. So always we have plenty of novelty all around us and we follow our instincts and we’re just looking for all sorts of cool and interesting stuff. So purpose starts from saying why. “Why do I want any information at all in the first place? What do I want to achieve in my life? What are the things that are important to me? What is it that serves me? What are the objectives?” And from that, understanding our relationship with information.
05:33
And just as we have a relationship with money or with other people or with our health and so on, we have a relationship with information so we want to make that relationship more empowering, one which is a positive relationship rather than a negative relationship. I would say the majority of people today have a negative relationship with information. They feel as if they’re sucked into the social media vortex, and they are. Whereas feeling you have a control, and that control comes from knowing that and getting some definition. And that’s defining your topics of interest and how they serve you so that you can get some clarity. That’s clarity that enables you distinguish between what is valuable and what is not valuable.
Jon Krohn: 06:19
Nice.
Ross Dawson: 06:20
So the second power is framing. This is [inaudible 00:06:27] of my first answer when people used to ask me, “How do you do this?” And this is the idea, we have to build frameworks which really give a clarity around what, the knowledge that we are building. So we have different structures for frameworks. I, identifying the chapter three, structures, trees, networks, and systems, and where we can organize the information that comes in, the ideas, the concepts that come in and using visualizations to be able to capture how it is we see a particular domain. So if it’s any particular aspect of data science for example, we try to build a foundational knowledge. What are the structures? What’s the frameworks? How these ideas fit together, building our own unique understanding and expertise. And then that allows us then to filter. When we see a new idea, where does that fit in our structure? Or in fact, where does it not fit? In which case we need to adjust and to modify our framework.
07:31
So this is really knowledge development. There’s a lot of studies over the years of how it is experts become experts, and that is through essentially unconscious pattern recognition where we start to see patterns in things. We have a recognition in ourselves. We’ve seen a similar pattern and essentially that’s where the expert exceeds the amateur, is they’re able to digest these patterns. So the framing is really making that pattern recognition explicit, being able to bring to the surface the ways in which you’re actually building the frameworks on which your knowledge or expertise reside.
Jon Krohn: 08:15
Nice. That sounds clear. And so I imagine in your book there are multiple different ways that you identify that people could potentially be framing so any one method could be suited to some particular purpose that they have in their life.
Ross Dawson: 08:32
Exactly.
Jon Krohn: 08:33
Nice.
Ross Dawson: 08:34
So say using both structural tools or structural frameworks and also visual, what I call concept frameworks or concept visualizations.
Jon Krohn: 08:45
Nice. All right. That sounds great. So once we found our purpose, then we identify the right framing structure so that we can make explicit the kinds of pattern recognition that experts use to create meaningful structure from the information that they are absorbing. And then with that, we can go to step three, which is filtering.
Ross Dawson: 09:08
Yes. So filtering is in the essence. It’s really distinguishing between positive value information and negative value information. So most information doesn’t have zero value, either it has positive value or either it does not have as much value as the time it takes to consume it.
Jon Krohn: 09:27
Right.
Ross Dawson: 09:27
Or in fact it has even further negative value because it is misleading or incorrect or simply not constructive in our thinking. So this is partly a sensitization to what serves us in being able to build better models, build our richer understanding and what is negative value and be able to get faster at that and providing some frameworks to do that. But one of the frames is having a portfolio of information sources. So just as we have an investment portfolio and we look at how do we get, I suppose, the portfolio together to be more than the sum of its parts by through the diversity, through the richness, through the complementary of those sources, so a designing a portfolio of sources, and then being able to have a frame of mind where we can be bringing what we find is of value into our mental models effectively.
10:33
And so one of the key frames is shifting from essentially a fixed thinking in terms of having beliefs to ones of being able to modify your thinking essentially, probabilistic thinking. And so in fact, Bayesian, BATS in fact can be applied to Bayesian thinking. So when we have a revision update in terms of a Bayesian model, that’s exactly the same thing. Another name for that is a belief revision. And so we are in fact using our thinking in our mental model so that any new information can be taken to adjusting or refining our model. So rather than confirming or disconfirming a fixed belief, we are continuing to revise our thinking on the basis of what information comes in. So that sensitizes to the information that can be most valuable in being able to adjust, use our priors and to adjust them to be able to become a more accurate assessment of the world.
Jon Krohn: 11:44
Nice. How many podcasts do you come on where you can just start talking about Bayesian stats and be confident that the audience will know what’s going on?
Ross Dawson: 11:51
Not many.
Jon Krohn: 11:54
I am so happy that you did that. So, yeah, for listeners who are unfamiliar with Bayesian statistics, we have had a number of great guests on the show discussing it in detail. One of the most in-depth introductions we’ve had and actually the most popular episode of Super Data Science in 2021 was episode number 507 with Rob Trangucci. So if you’d an introduction to Bayesian stats, check that out. But it is like Ross is describing, so you have a prior belief and then you use information to adjust that belief into what we call the posterior. And even that posterior, it’s a probability distribution. So it isn’t a point belief single scale or value. I really love that way of thinking about the way that we absorb information. That’s brilliant. Thank you, Ross. So once we’ve applied our filter, are we into the attention point now? Or is that still coming up?
Ross Dawson: 12:53
Yeah, that’s sort of a reasonable distillation of the filtering aspect. So attention, I think of what’s critical to recognize is that it’s not as if we either have attention or not attention. The default mode network of our brains essentially describes that our mind wandering is the default state of our minds. Unless we are specifically focused on something, our mind is wandering. So there are in fact a number of different activities and different types of attention that we need in order to be able to process information effectively. So for example, one mode is scanning, where basically that is one particular attention mode where it’s relatively light as it were. We are looking through a predefined set of sources and we are looking for things where we want to potentially dive in further. So this is one particular mode. It’s not necessarily requiring incredible focus, but is one where we are looking across sources, be able to identify, “Okay, not interested. Maybe interesting” and you go through those.
14:02
So a quite distinct mode is assimilating where you have identified, “Ah, this is potentially of interest. This is something which is relevant to my interests. I’ll check that against my purpose.” And then you spend time in a different frame of mind, different tension mode where you’re actually, “Okay, I’m looking for the ideas in this. I’m correlating that to my existing mental models or frameworks,” bringing that in, being able to consider that, potentially make some notes and so on. And so there’s six attention modes in all. So there’s scanning, assimilating, exploring, seeking, deep diving, and that’s absolutely critical. Deep diving is a state where you take everything out. No distractions, no notifications, not going to be interrupted by anybody unless it’s the end of the world. You have the conditions where you can be focused and you practice getting into this deep diving mode.
15:03
One simple recommendation for anybody, and I think your audience would probably already be a lot better than most getting into that deep diving state because you have to really build models and create structures in software, but is that you have at least one period of time every week where you get into this, everything gone, getting focused, and you’re practicing that ability to focus. So through the chapters, also a number of different exercises or practice for how we can refine our attention. This is a muscle. This is something which we can exercise. This is something which we can develop. And as we get better at attention, we can apply that when we choose to do so.
15:49
And the sixth attention mode, which is just as important as any of the others, is regenerating because our attention of our brain is finite. It is limited. It is we can’t sustain continuous attention. And in fact, the best way to be able to have our highest levels of attention is to stop, turn off all digital things to take ourselves out of information. And in fact, the research shows that in fact places of nature are the ones that are best at regenerating our ability for attention.
Jon Krohn: 16:26
Wow. Is there any literature on the bare minimum amount of regeneration that we need in a day or in a week to be optimal?
Ross Dawson: 16:37
There’s a variety of research, but I mean I think it is very contextual in terms of individuals and situations and so on.
Jon Krohn: 16:46
Right.
Ross Dawson: 16:46
But essentially in terms of, 30 minutes a day, I would say, is a bare minimum. There’s also some great work by the Kaplans of a couple who are a foundation of a lot of this research including what they call Attention Restoration Theory, which is probably what we all need. In that research, they distinguish between what they call hard fascination and soft fascination. And so this is where the fascination is when you are seeing thing and you are drawn into that. So the soft fascination is, for example, you’re looking at some trees in the wind or the ocean. Or even in, I think, people’s faces, this is something where there’s nothing specific to focus on, you are engaged in that.
17:34
Hard fascination is watching a movie. So that’s something where you are fascinated, you are absorbed in it, but that is not regenerating your attention. And so if we are spending all of our time on Netflix or streaming, that is not in fact truly regenerating our attention. It is relaxing perhaps, but that is not the same degree of regenerating that we need for truly being able to pay high focus attention when we want to.
Jon Krohn: 18:10
That is crystal clear. I really appreciate that quite practical advice. It also is in line… So it’s nice that there’s research by these Kaplans that backs something that I think I intuitively agree with. You don’t come out of scanning a social media feed or watching a film often feeling rejuvenated and excited to dig into reading a book-
Ross Dawson: 18:36
Indeed.
Jon Krohn: 18:36
… in the same way that you might if you take the dog out for a walk in the park.
Ross Dawson: 18:40
Exactly.
Jon Krohn: 18:41
Awesome. All right. So we’ve got through purpose, framing, filtering attention. We’ve just got number five left, synthesis.
Ross Dawson: 18:49
So synthesis is the master skill. I think it’s actually quite pertinent for a data science audience. One of the key aspects that I describe of synthesis, that it is the capability that will keep humans ahead of machines, where this is the ability to pull together disparate data into understanding. Literally being able to see the whole. So analysis is the slicing and the dicing and the breaking things into smaller and smaller parts. The synthesis is the opposite, it’s pulling that all together to see how parts connect to be able to perceive a whole. This is something where a lot of this does reside essentially at our subconscious level. I talked before about how essentially patent recognition is something which happens at subconscious level and that analysis is a very conscious activity, consciously breaking things down to pieces.
19:51
A lot of the insight, so if you look at any great scientists or artists or engineers or any people that are coming up with new ideas, they usually experience this moment of insight. There has been some great research done on this moment of insight and the states of our brain in which we are able to receive these. So part of what we can start to uncover is what it is that nurtures those states of mind that bring us the insight, the ones that enables us to make these creative connections, the ones that enable us to be open to new data or new information that we can incorporate into our models.
20:36
Machine learning is, by definition, domain bounded because you’ve got a data set, that data set has certain parameters and you can train that, but that’s only applicable of course within those particular parameters of that data set. And so what humans can do, which machines cannot fit the foreseeable future, is to be able to pull together different things of different classes or categories or different levels of logic and to pull those together to make sense. So this is really the master capability. And as, indeed, we have more and more models and algorithms and data models that can often exceed human performance in any specific decision-making domain, where humans will continue to be superior and to be able to keep ahead is via nurturing that capability of synthesis.
Jon Krohn: 21:34
Nice. I love it. All right. So Ross, you’ve given us the five intertwined powers to develop in order to thrive on overload. This information overload that so many of us data scientists face. Yeah, so purpose, framing, filtering, attention, synthesis. We can get more on each of those out of your book. Lots of detail around, say the specific frameworks that we could use to frame more information on the six types of attention. It all sounds like really valuable reading. But one thing I don’t think we’d be able to get out of your book that I’d love for you to tell us on air is your thoughts about how our listeners, how this data science audience could develop tools or machine learning models or products to automate or augment or assist any of these five hours.
Ross Dawson: 22:29
So I think there’s immense possibilities. One of the most obvious is simply in content filtering and recommendations. And this is interesting. I’ve actually recently set up a startup to be able to play in the space of information productivity, how we can create value from-
Jon Krohn: 22:50
Instill all your intellectual property on air press, please.
Ross Dawson: 22:53
Sure. This is about sharing. I think that this is a common space of play. What I would say one fairly obvious thing is that content recommendations is something which in 2022 we still haven’t cracked.
Jon Krohn: 23:10
Right.
Ross Dawson: 23:10
It’s quite kind of amazing. If you look over the last 20 years, you think, “Oh, well surely, AI could tell us what information we want and what we don’t want.” Well, I’d like to see the products which do that. And that is still something where we will crack this better than we have, but we haven’t yet.
Jon Krohn: 23:28
What I’ve heard… Well, I do have a TikTok account, but I’ve never really used it. My social media team for the podcast creates 30-second clips of what they think is a highlight of a Super Data Science episode. And then they convert that into a TikTok video and they post it on my TikTok account. So I have this developing TikTok account. I’ve never been on myself, but what I’ve heard from people is not in the kind of content recommending system that you are talking about where it’s useful information that’s going to help us with our jobs. But apparently, TikTok has cracked an algorithm in a way in terms of finding you amusing information that’s tailored to you. People describe it, compared to the preceding kinds of big social media networks like Facebook or Instagram, apparently, TikTok is really on to something with keeping people amused. Maybe they don’t come out of it with a sense of regeneration, but they do come out of it without that sense of gloom that I think has been associated with a lot of the social media applications of the past. Anyway, I am digressing.
Ross Dawson: 24:39
Well, on that point, I mean, I describe TikTok as having pushed out the frontiers of attention hijacking. And so essentially, the various social platforms over time got better at it. TikTok has absolutely pushed up the bar, but this is in terms of hijacking your attention and pulling you into that vortex of, “Okay, well, there’s this, and there’s this, and there’s this, and there’s this,” but that is not really functional from the issue of-
Jon Krohn: 25:05
Yeah, it doesn’t.
Ross Dawson: 25:06
… “Okay. This is the most relevant information. This is not one which is necessarily appropriate.” Some things, maybe 30 seconds is what you need. Other things, maybe you need a bit longer, and so moving on the next thing is not going to be the most functional for us.
Jon Krohn: 25:19
Exactly. Yeah, it’s not effective for a scanning. It’s not effective for a simulation. It’s certainly not effective for a deep dive and it’s not effective for a regeneration, that TikTok experience. So I guess what you’re saying is that if somebody could come up with a content recommendation tool that allowed us to scan and assimilate, and maybe even then, I’m imagining a hypothetical application here, but you could have one that a machine learning algorithm that facilitated scanning over the types of information that are useful to you, say professionally, and then you could be doing some assimilation and then decide, “Okay, I’ve found a paper that I’d like to dig into for the next hour” and at the click of a button, have all other applications be unable to notify you and then you could deep dive on that article for an hour, maybe just be able to take notes and that’s it.
Ross Dawson: 26:16
Well, that’s an overlay, I would think, to the data science aspect of that and that these are some of the functional tools of controlling your attention. But I think in terms of the data science, it is being able to identify which of the N papers that have emerged recently are the ones which are going to be particularly pertinent to you. And this is where, for example, some of the data inputs are eye-gaze. So if you’re looking at eye-gaze tracking, you can start to get better insights into what it is of interest and relevant, be able to find other signals which can identify what’s going to give you data around what else would be interesting to you.
27:01
But the other thing I would add, and this is something I’ve written publicly about for over a dozen years and I haven’t really seen taken up, is this idea of a serendipity dial. This is not just for content.
Jon Krohn: 27:14
Totally.
Ross Dawson: 27:14
This could be for music. It could be for movies.
Jon Krohn: 27:17
100%.
Ross Dawson: 27:18
It could be for anything else, saying, “Okay, at some points I don’t want any serendipity at all. I just want to get only the music which I know that I love or only the content that I know that I love.” Or sometimes I want to crank up the serendipity and I want to say, “Okay, actually I want to hear something which I haven’t heard before” which it might not be something I love, but it might be something I love which I’ve never heard before or seen or whatever. I think this is something where we get locked into particular parameters in how we present content broadly to people. And so this is something I’ve been trying to… It’s my hobby horse in a way, this idea, but I think that that’s one of the important parameters. And so maybe it’s a dial where you give the user control of that. Maybe it’s something you just consider. “All right, what degree of serendipity I’m going to offer people?”
28:12
And it’s interesting. In Spotify, apparently they have their Discover Weekly where basically every week it gives you some recommendations. They found that people had greater engagement with that Discover Weekly if they included songs that that person had already listened to before.
Jon Krohn: 28:33
Right.
Ross Dawson: 28:34
So this is something where Spotify uses algorithms to be able to identify what broadly was their most interest. But I, and I know that there’s plenty of other people, think, “What the f***? This is supposed to be Discover. I already know this.”
Jon Krohn: 28:47
Right. Yeah. Yeah. Yeah. Yeah, they’re solving for something different from you. Similar to a TikTok, they’re trying to capture your attention for as much as possible. They’re not necessarily trying to provide you with the most authentically Discovery experience. Yeah, so I’ve thought about this a lot. It’s something that I felt like when now I am… Yeah, I use apps for my music like probably everyone else. I happen to use Apple Music these days. Something that I miss from listening to the radio regularly, I certainly don’t miss the ads, I certainly like being able to pick whatever song I want to listen to when I want to listen to it, but I do miss the serendipity element that radio broadcast hosts seem to provide.
29:41
So I, for example, I grew up loving Rock Radio, but sometimes other music would come along where a DJ was like, “You know, this isn’t rock, but you’ve got to check this out” and they play. So things like Eminem or something back when I was a teenager that there were like… There were some Eminem songs that they played on Rock Radio, or Outcast, who are really clever musicians, kind of genre-defying musicians. It was really cool to have that come in a way that probably wouldn’t happen if I was just listening to the rock playlist on Apple Music or Spotify. Yeah, so that’s great guidance Ross. This has been such a fun conversation. I’ve absolutely loved talking to you about these topics. I feel energized by the possibility of minimizing my information overwhelm and maximizing my sense of abundance and empowerment. That just sounds so wonderful. Have you attained something like that as a quick side? Do you feel…
Ross Dawson: 30:51
Well, no, I’m not…
Jon Krohn: 30:54
You’re not an information master yourself. You just study them?
Ross Dawson: 30:57
No, I am… This is one of my superpowers, I’m not perfect and I’m working on it. So I’m good. There’s a lot of things which I do very well, but occasionally I pick up with my phone and scroll through social media where it’s probably not the optimal thing to do. So it’s a journey. And that’s it. We need to continue to improve. But as I continue to evolve, I am pretty good. In a way, I can observe what I’m doing in order to be able to teach others, but I also can observe myself and say, “Hey, actually, there’s probably some better things I can do.” And this is a journey. Nobody’s ever going to be perfect in any of these things. And I think that that’s empowering as well to be able to say, “This is an opportunity for improvement and I can get better.”
31:53
And I think everyone as well. I think you can look at your own habits and say, “Oh, that’s not so great,” but you can say, “Well, I can look at my habits and say, ‘Well, okay. I actually can see plenty of opportunities for me to have information habits’,” which are not just empowering in terms of your ability to be an expert and your ability to do great at your work, but also to be happier, to have a greater wellbeing in your life. I mean, I would say our information environment is deeply destructive. We’ve created it because that’s what humans want. We want lots sort of innovation. We’ve created this world, which actually if we let ourselves get sucked into it, it’s really destructive. And so we do need to empower ourselves, and so we say, “Well, okay, I’ve got some choices.” And just being conscious. Just start to think, “What [inaudible 00:32:46] one or two or three things, which I can do to make myself that little bit better?” and you’ll experience the difference.
Jon Krohn: 32:55
Nice. Yeah. Aimlessly going through social media less, getting deep, carving out some of that time to go deep into topics without distractions for a bit of time every week and maybe cranking that up over time if it feels like the right thing as you start to ease into it, training up that attention muscles as you describe it. That all sounds like great advice, Ross. Thank you so much for coming on the Super Data Science Podcast for being our guinea pig for this new Friday guest format. And yeah, listener, check out Thriving on Overload, which is available now in the US and the rest of the world within the next month at least. Have a brilliant rest.
Ross Dawson: 33:38
Pleasure to talk to you, Jon.
Jon Krohn: 33:41
Yeah. Catch you in a bit.
