Conversations with Mitch
Conversations with Mitch
Thinking about Web3, AI, Starlink, and subscription business models
I’m away teaching this week, but I’m delighted to share with you my recently published conversation with Mitch Joel about my new book, on his long-running podcast, Six Pixels of Separation.
You can listen to the entire hour-long conversation here.
But I want to pull out a few key sections in particular, where we discussed:
Web3, blockchain, and decentralized systems
Lessons from Starlink for emerging tech
Generative AI and Predictive AI: Hype vs. business value
Streaming, SaaS and subscription business models
“Move Fast and Break Things” vs. Rapid Experimentation
Enjoy!
P.S. We had an amazing conversation with Lucy Kueng last week on leadership lessons from the media industry. The video is now edited and captioned. It will be posted in next week’s edition of David Rogers on Digital. Stay tuned!
Web3, Blockchain, and Decentralized Systems
Mitch Joel:
You and I have spent many, many years looking at change, understanding change, thinking about it from different angles… How are consumers behaving? How do we move and adapt? I would submit that in the past little while, we're starting to see some major technological things happen that might shake the roots of what we've seen before.
The first one that I'd like to talk about and get your perspective on is this entire world. I'm not going to call it Web3. I'd rather look at it as centralized and decentralized platforms.
There is a large push towards this idea of decentralization. I have my own thoughts and philosophies about it. In particular, it's nice to be decentralized, but you live in a centralized world. So, anything that you decentralize that there's a problem, you will move to a centralized system, whether it's the legal system or accounting or whatever. You have to figure out what it is. So, it's obtuse and silly maybe to have this conversation, but it is something that a lot of people are looking about. How do we decentralize social media, how do we decentralize databases, all this stuff. What do you think about when people talk about this idea of centralized or decentralized?
David Rogers:
So, one thing I think of is -- there was an incredibly ironic bit of news where I read that someone who claims to be the originator of Bitcoin, who wrote the original white paper on blockchain…
Mitch Joel:
Not Satoshi.
David Rogers:
Yes. Someone who claims to be Satoshi.
Mitch Joel:
Okay.
David Rogers:
Yeah. I forget the gentleman's name. And he-
Mitch Joel:
“David Rogers.”
I'm kidding.
David Rogers:
Not me!
He lost a whole lot of his cryptocurrency. This happens.
This whole idea that, “Oh, this will be secure, it'll be safe.” Yeah, except if you lose your key. There are lots of scenarios. Anyone who is involved in actual security will tell you that the technology element of security is the least important. It matters, but it's the easiest one to solve. That's not the real-
Mitch Joel:
The human error is the problem.
David Rogers:
The human error. Right. Humans are the problem. The fundamental unsolvable problems. So, of course there's going to be times where you lose your asset, or it gets stolen.
And so the actual fellow who claims to be Satoshi, who started off with this vision of the permissionless decentralized architecture… he loses a bunch of money, and he's going after who he says is liable for it. So, he takes them to court in a centralized government to bring a lawsuit against them. A civil lawsuit.
That's one thing I think of.
There's Web3, the idea of: what can we develop through some different and more decentralized architectures? Will that allow us to solve problems? Will that allow us to create better systems that better meet the needs of their stakeholders?
This ideological idea that decentralized is always better, and the whole point is “let's just make everything as decentralized as possible and then the world will just”... that it will magically make everything better because in theory, “I've got this white paper explaining why the whole world will be a much better place if we do that.”
I'm sorry, that's not how change happens politically, technologically, commercially.
Let's find an application. I'm still waiting for that first Web3 application that comes out built radically differently, that works really well for the users, and draws a lot of use by a lot of people. I'm sure it's going to happen.
I'm not really interested in the long seminars about how, in theory and principle, this will bring a better world.
Mitch Joel:
Things like smart contracts make a lot of sense. I do think about things like attaching NFT-like technology. I do like this idea that in a physical world we have assets that are abundant and scarce, and digital assets should comport to the same type of world. All of these things I think are massive business model opportunities for somebody to get right.
Do you think about maybe the lack of enthusiasm we're seeing around blockchain? Or are you more on the camp of there are so many people using this type of technology that it may not be visible as a commercial product, that we're not seeing how strong it's being used? Is there some tie in with that and decentralized or newer technologies too?
There's a lot of talk. It was like all hype and then it gets used, and then it gets pulled back in type of thing.
Lessons from Starlink for Emerging Tech
David Rogers:
Yeah. I would draw an analogy between blockchain and satellite internet with Starlink.
Folks were saying, "Oh, Elon Musk is going to disrupt the entire telco industry. He's going to build this satellite network." And then, “who needs all these fixed lines and so forth?” And just in principle it sounds like, oh wow, this is going to be a classic, low-end disruption, and take over.
I talk in my last book about different parameters of disruptive business models. Would this be one of those cases where it doesn't disrupt just a section of the market, but it takes over the whole market? Which we certainly see in some cases like with network effects.
So, that was the thinking. There was all this talk, oh, the whole industry is going to go this direction.
And then I heard a very frank discussion by someone who'd been in the industry for a long time who said... It was at one of the big established players, and they were up front… They said, "Yeah, we're looking at this. It's not the best technology for a number of circumstances."
They talked about urban density and geography and populations and so forth, and explained, just from the basics of the technology, why in terms of scalability and reliability and speed—depending on, again, density and topography, and so forth—cell towers were just a better solution for some populations than [satellite].
So, I recognized early on, “Okay, this is going to maybe work in some scenarios, but not in others.” Maybe this is really a lot of excitement over really nothing. Then of course, I didn't think about it that far ahead. I wasn't investing in the sector or advising a telco on that particular decision.
A few years later now, we suddenly see the scenario—because of Ukraine—where satellite internet is not incrementally better, but radically better. The classic marker of business model disruption. It is in situations of great instability. Warfare, obviously being one of them, but the aftermath of a hurricane and other cases as well.
And all of a sudden it becomes like, “Oh, why didn't the US military buy a 40% stake in Starlink as soon as this was going?”
So, there's obviously going to be some areas, some applications, some use cases where that technology is not just a nice option, or somewhat better, but radically better.
And I think the same thing is true in blockchain. It's a technology which is really interesting, has certain amazing capabilities, has certain really strong limitations. Now, some of them, like the move from proof [of work] to proof-of-stake reduces one of them, which was this terrible cost on the environment and carbon pollution. But there's still others. Versus a traditional centralized database, it's incredibly resource inefficient.
So, it's not something that is going to replace every centralized application or process, if you understand the technology strengths and weaknesses.
But it's going to be very interesting and useful in many applications, and there'll likely be some applications where it allows you to do something you simply could not do otherwise.
So, I would take the same approach in thinking about blockchain.
Generative AI and Predictive AI
Mitch Joel:
I know when I was teeing up my question about centralized versus decentralized, you probably thought I was going to ask you about artificial intelligence. And I didn't, but I will now. So, let's talk a little bit about this.
If we walk just a little bit further down the road on AI, it is inevitable that AI will in and of itself be developing technology that we will use, which will be part of The Digital Transformation Roadmap. I think about this very esoterically. This is interesting.
You have a technology that's going to develop a software, which means it won't be coded by human and typing and actual keys, and then how are we going to then-
David Rogers:
But so do compiler languages, right? Compilers. We don't write in binary.
It's a tool.
Mitch Joel:
But it was based on that. This might be thinking in other pathways, or creations that we may not be able to communicate with, in theory, through the simple interface of a keyboard. I'm just walking down this idea of: how would you debug it in a world where it wasn't even created on a keyboard? Again, maybe more of a science fictiony thought, but I think it's actually very pragmatic and we're arriving there soon.
I think at a greater scale, not generative AI tools, but more artificial intelligence we are probably in some type of precipice of redefining what knowledge even means, which will be a big challenge to us.
How are you thinking about it at that level? So, I'm not looking for, wow, isn't generative AI tool cool? Yeah, it does a lot of stuff. I'm watching every company integrate it into some way, shape or form of what they're doing. Everybody's “AI first.” Yeah.
I'm looking at people like Yuval Noah Harari or Douglas Rushkoff trying to debate whether or not it's going to kill us or it's a stupid web interface. I like those two. I'm somewhere in the middle of this, and thinking that there's something different here that might change who we are as people and employees. And I'm wondering how you see this in the world of transformation.
David Rogers:
Yeah. I think that's true.
I think first of all, if we talk about artificial intelligence, we're really talking about machine learning and a series of technologies that have emerged at a commercial scale in the last 15 years or so, which are based on artificial neural networks, deep learning, etc.
And it's worth bearing in mind: we're on a couple tracks at this point. So, we had one whole set of breakthroughs applying this technology in a predictive fashion. So, what you could call predictive AI.
And these are building models for things like machine vision: recognizing what's that thing on the road in front of our increasingly autonomous vehicle. (They're not fully autonomous.)
It's a drone flying over a bridge on behalf of a mechanical engineering firm, and it's scanning pictures and identifying what is or isn't likely to be a crack that does or does not need reinforcement or support in terms of maintaining that asset.
It's algorithms listening to your callers on a phone line and recognizing which ones are irritated and need a different level of customer service.
We're already very far along the path of applying predictive AI to problems inside organizations in many different industries. So, it's already reaping a lot of value. We're at this disjuncture right now where I would say predictive AI is generating 95 to 99% of the value of AI right now. Generative AI is generating 95 to 99% of the hype.
So, it's really these two streams that are going.
And the thing about generative AI… It's much earlier on. It's what predictive AI with these models looked like a little over a decade ago when we were doing things like: IBM having it play Jeopardy! and beat the best human players.
It's another thing to say, how do we bring AI into a radiology lab? Do we have it read the MRI scans for us? Or make the report? Or is it a tool used by the radiologist? And which things does it do better or worse, and which things are better done by a person in conjunction with this technology?
We've done that journey now for several years with predictive AI. We're just at the beginning on generative AI. Large language models are going to have a lot of power. Unfortunately, a lot of the early applications [have been in] some of the things that it will be least useful for—because of its inherent factual unreliability, and some other elements of unreliability based on training data and other causes. Those are all workable.
There's unreliability. And there's the “black box problem” of not knowing how it's working. That's been true of predictive AI as well, so that doesn't mean we aren't going to find really powerful applications for generative AI.
So, I think it is the kind of technology that does not stand by itself. It tees up and it unleashes, or it catalyzes, a lot more waves of change. The same way that the smartphone did. You couldn't really see in 2008 with the first iPhone what this was going to mean in terms of new business models that would be running on it in 10 years, once everyone had an internet-connected computer on them at all times.
Similarly, we're not going to be able to really see ... We've got the first glimmers, but it's going to be years before we really start to see, wow, how is this really going to impact society, business, government, our own personal experiences, our mental health, perhaps all kinds of things? Opportunities and challenges.
I think we're just starting down the path. We're at the beginning of a big learning journey, and I think it's great that people are raising concerns. I may quibble with some of the concerns I think are the wrong ones to be focused on, or they're fictional or fantastical. But obviously there are concerns.
So, it's great that people are saying, let's not just say “put this everywhere and do anything with it.” I think it's going to be one of the key external drivers pushing companies to keep transforming over the next decade, without a doubt.
Streaming, SaaS and Subscription Business Models
Mitch Joel:
Do you see things more trend-based, and the speed of which they're moving? Or do you see them being foundational?
I'm thinking in particular about subscription models. Everything was moving to subscription models and opportunities. Some of them that you wouldn't even pay for, they were just engines of marketing.
And suddenly we find ourselves in a place where consumers are saying, “I've had enough about paying for access to all of these things.” And then, it goes where? It goes back to one-time fees. Are there things in that world that are interesting to us?
David Rogers:
What category are you thinking of?
Because if I look at B2B subscriptions and consumer products & services, I don't think there's a universal trend going on.
Which are you thinking of Mitch?
Mitch Joel:
Yeah. I might argue that they are universal, that we had an habituation through streaming services to pay a little bit for having access to full libraries, and that B2B piled on and you had the SaaS and all of this happening at the same time.
And I think people are looking at all of these things and thinking, “I'm paying a lot for everything now.”
David Rogers:
Yeah. I think subscription versus purchase, it may be a bit like the old bundling and un-bundling.
There's a certain school of thought that all business model innovation is either bundling or un-bundling. So, pendulums will swing back and forth I think within industries, but it's driven by different factors.
The only thing I would say in common between something like consumer media streaming, video streaming subscriptions, and something like SaaS [B2B] business models like Slack or many others… would be that venture capital was attracted to investing on both sides because they're familiar. They have similar KPIs, similar financial metrics, and financial advantages if you get them right.
In terms of the specific case of consumer media streaming: We're on this curve where you had Netflix first breaking the ice, and then you had Hulu and Amazon Prime in different markets, and Prime Video coming on next.
And then eventually the legacy players. Hulu actually was a legacy player, so that's an exception for the markets they were in, with HBO being that tipping point. And now we have this flood of all these subscriptions, and we say, “How many subscriptions am I paying for?”
The prices are going up. It's not just Disney, it's Netflix, it's legacy and digital-first companies. They're all raising the prices because guess what? They want at least break even.
Ideally, if you're a primarily, or exclusively, a media company, you want to make a profit [off streaming]. And so that's a bit of, again… shifting from venture-backed, where we can wait for some future date to make a profit and that is the maturation of the category where the capital's not there.
Plus of course the cost of capital has gone up. That's been a big driver here. You've had industries wide, streaming and others, that have been subsidized by cheap capital, because interest rates globally were so low. Now they're higher. All of a sudden, companies that are not year three, year four… “Hey, you've been in the market seven, 10 years. Guess what? We expect you to make a profit now.”
That's been the shift That has been just one of the elements, obviously not the big one, but that's been one inflationary element.
But guess what? For business for B2B, in most cases, it is far advantageous for the business customer to buy services on a SaaS basis.
Mitch Joel:
Sure.
David Rogers:
Instead of a license to install some enterprise application that you're stuck with and is not very flexible and customizable and then you're just like, “oh, well it's here.” We've sunk all this money in it, and so you have all this technical debt to deal with.
It's a whole different ball game.
So, I think the subscription business model in each sector plays out differently based on access to capital, customer price elasticity, and how many entrants? There's probably far too many players right now in the consumer streaming video. Not necessarily the case in some categories of SaaS subscription models for businesses.
Are there too many [streaming] players? Is it going to shake out? And so, that's why you're going to see a natural process. People are going to say, “I'm going to cancel a few of these.” Great. That's what happens. And somebody's going to fall by the wayside. I think that's just creative destruction and evolution in the market.
Mitch Joel:
You were heading into one of my favorite terms in the B2B SaaS side, which was VAR, value added resellers.
You buy the massive software you'd bring in the consultants that would optimize it and fix it. By the time it was done, there were six versions down the road. You're like ... Yeah.
I don't think people realized how much cloud computing really changed this notion of how that business operates. It's insane to think that we live in a world now where it seems so obvious. Of course, we're all on the same version of Shopify!
David Rogers:
Yeah. How could we ever not be?
Because it wasn't in the cloud. You had all these premises, all these instances in different places around the world and they were all out of sync.
It's been a real turning of the page. When we talk about the shift to cloud, there are certain things that are inherent benefits of the cloud itself. Things like scalability and resilience and so forth, and, for a large majority of players, better security, simply by making that change.
But the other thing is that enables... It goes hand in hand with a shift towards a more modular architecture of IT which is much more flexible, more adaptable. Everything I said about experimentation... This is what, because of the whole movement of agile software development, this is what all IT has gone to for the last 20 years.
They realize: you can't build software the way they used to anymore. Not at the scale that we need it. It's just completely impossible.
And we're still trying to run the rest of the business—marketing and supply chain and inventory and human resources in many companies—we're still trying to run it in the way that software was built 20 years ago. To build those giant impossible systems that you can't change anything once it's in place.
We need to make the same revolution.
“Move Fast and Break Things” vs. Rapid Experimentation
Mitch Joel:
You mentioned this idea of rapid testing and learning. And I think the derivative of that was the way in which Facebook changed that to be "move fast and break things."
Which has become somewhat synonymous with the problematic nature of big tech and data, and what's happening [with] the hoovering up of our data and ownership and things like that.
Can you help people reconcile this?
Because it's easy to stand as the prophet before us and tell us about the imperatives of digital transformation. And I won't be an apologist for the work that I've done over 20 years in building the agency that I subsequently sold.
But it is a bit of a different moment in time, where people are not so excited about rapid testing and learning and doing all this stuff.
David Rogers:
I would make a distinction. It is not the same thing to say "We need to move quickly and test and learn" versus "We need to move fast and break things."
So, the idea of that mantra... which by the way, Meta itself discarded years ago before everything blew up reputationally for them. Simply [in] getting to scale, they had stopped saying that.
[Meta:] "Yeah, we're a little too large, actually.” We've got to put the word out that we need to be a little more cautious about thinking about gaming out with any changes we're making.” “What's the impact going to be on our established business, on our markets, on our business model and so forth?"
Which was a bit of growing up for them.
But the phrase stuck with them. and they didn't change enough the culture. And they didn't think broadly enough about what risks they needed to consider. It wasn't just about “how do we manage our advertising revenue.”
But that's really a different...
It's related, but it's not the same thing as rapidly innovating through experimentation. Which is simply about a couple of things. One is thinking like a scientist, being humble, not going into any change, any new initiative, any innovation, any new product launch, any shift in your operating model with the traditional mindset, which is: let's do a lot of planning. First, let's gather a lot of third-party data about other people's experiences, none of it directly measured by us. Second, let's do a lot of analysis, a long process, gaming up different business cases where we imagine we can predict the future if we go down one path or another. And then we have a top executive make the decision. Once we've done that, we plow as many resources as we can into it.
We see this time and again, that's where a lot of companies go down launching a new digital product or service or business model, and it blows up in their face and is extremely costly. They say, "oh, we can't do any more of this."
But they've taken the completely wrong approach.
If you think about validating all the assumptions, you start with the humility to say, "Hey, we're CNN and we think we want to start streaming our content through an app." Or "We're Amazon, we're a retailer and we think we could go into cloud computing." Or you're a Walmart and you say, "Hey, what if we built an advertising business? Or how are we going to deliver groceries to the customer?"
And you don't go in and plan the whole thing out and get the CEO approval and then put a thousand people on it, right? (That's what we see in all these digital transformations that fail.)
Instead, you figure out what's the problem you're trying to solve or the opportunity you think there is for growth, and then you come up with possible solutions. But as the saying goes, you “fall in love with the problem and not the solution.”
And then for any solution you're considering, you figure out what is the cheapest, quickest, fastest way, least-risk, least-impact way that we can start to just validate some of the basic questions here?
Like "does the customer even want this?" or "who would the customer actually be with this product or service?" or "what would it take to actually deliver this?" or "what is the regulatory risk if we were actually try to bring this to the market?" All sorts of questions.
You can actually validate much more quickly if you simply build and bake a different process into how you manage teams in the organization. So, that's what this is about.
Experimentation is much lower risk than planning. Everyone thinks, "oh, you're moving fast, that's got to be high risk." But you're moving fast in a totally different way that is actually understanding the risks that you're taking on.
And so, you have actually much less risk than the traditional "let's make a big plan and then finally decide if we're going to pull the trigger and throw everything at it." That is incredibly risky, and that's why we see a lot of failure.
Stay Tuned…
Stay tuned next week for a video recording, and transcript, of my recent live conversation with Lucy Kueng last week on digital leadership lessons from the media industry.
NEW BOOK:
“THE DIGITAL TRANSFORMATION ROADMAP: Rebuild Your Organization for Continuous Change”
ORDER NOW:
Hardcover: https://amzn.to/41U85dl
Kindle: https://amzn.to/3OWD437
Audiobook: https://bit.ly/DXR-Audiobook
Bulk orders up to 60% off: https://bit.ly/DXR-bulk-orders