Welcome to Stuff I Thought About Last Week, a collection of topics on tech, innovation, science, the digital economic transition, the finance industry, free energy, and whatever else made me think last week. Please grab me on Twitter with any thoughts or feedback.

Click HERE to SIGN UP for SITALWeek’s Sunday EMAIL (please note some ad blocking software may disrupt the sign up form; if you have any issues or questions please email sitalweek@nzscapital.com)

In today’s post: building 3D maps of the world; Arm’s accelerating implementation for machine learning; Google Pay looking to become a super app; wearables predict sickness; digital DJs to curate algorithmic content; extreme land demand as suburbs sprawl again; inverting our view of how the brain works leads to insights about decision making; and lots more below...

Stuff about Innovation and Technology
Companies Mortgaging Their Future
It’s been over two years since we wrote in this MarketWatch post about the Walgreens Boots Alliance CEO (Stefano Pessina, at the time) saying he was “not particularly worried” about Amazon buying online pharmacy PillPack. Last week, Amazon finally announced prescriptions available for 2-day shipping to Prime members in 45 states. The point we made in the MarketWatch article was that legacy companies were overly focused on hoovering up their own stock instead of investing capital in building digital platforms to meet the evolving needs of modern, Information-Age customers. This practice of favoring financial engineering over innovation is a way to mortgage the future of a company. Since then, Walgreens has bought back ~$5B of its stock, which, incidentally, has declined ~40%, while the S&P500 is up ~23% and Amazon is up ~58%. The Walgreen’s CFO commented last week that the drop in the stock was “disappointing”. Today, it’s not much easier to get prescriptions from Walgreens, and, meanwhile, the digital platforms continue to invest in offering their customers more product and convenience for less money. Amazon doesn’t yet have local pharmacies for same-day prescription fulfillment, and there is a lot of anti-competitive collusion between insurance companies, pharmacies, and pharmacy benefit managers that will require time and creativity to break, but the writing is on the wall. There are a huge number of legacy, 1900’s Industrial-Age business models that have failed to integrate with or compete against information-based platforms, and their days are numbered.

Android and iOS Building a 3D Map of the World
This CNET interview with the AR team at Apple underscores the point of how well positioned Android and iOS are to build the 3D maps of structures and the outside world for developers to leverage in new applications. AR is a feature today in many phone-based apps and will steadily be more and more front and center and functional. Last week, I mentioned this article on Niantic building an AR map of the world as people play Pokémon.

Arm Enables Mac-Based ML
Google’s newly-released TensorFlow – optimized for Apple’s new M1 Arm processor – boasts up to a 7x performance increase over the Intel-powered Macs. This new fork of TensorFlow could significantly lower the cost to train machine learning models. According to this detailed Arstechnica interview with the team at Apple, the designers used a unified memory architecture (UMA) to drive the performance: “all the components—a central processor (CPU), a graphics processor (GPU), a neural processor (NPU), an image signal processor (ISP), and so on—share one pool of very fast memory, positioned very close to all of them. This is counter to a common desktop paradigm, of say, dedicating one pool of memory to the CPU and another to the GPU on the other side of the board.” Despite some good reviews, the new M1 has a lot of problems if you venture off Safari and try to run anything non-natively. Once the software adapts, there will be broad demand for a chip like this and other Arm processors in the data center in coming years, but TSMC is selling out for leading-edge manufacturing, so it may be awhile before we see a lot of machine learning done on Arm-based chips. If the Nvidia purchase of Arm is approved in around 18 months, the timing could be right for Arm to start taking data center workloads en masse.

Google Pay Upgrade
Google has a new version of Google Pay that’s looking to take a much broader role in peer-to-peer and business payments, as well as banking, rewards, food ordering, etc. I've been using the app for a few days, and I am surprised at how much gamification they have built into the interface to unlock cash back and offers. Google Pay currently has 150M users in 30 countries, which, frankly, I find surprisingly low given that it’s the primary channel for contactless payments on Android phones. With the new version, you can now centralize your credit cards and other financial accounts for a more comprehensive view of your spending and saving habits. Google also plans to launch Plex in 2021, which will integrate with 11 mobile-first bank account providers, which I think is an interesting example of the big tech platforms enabling new digital competition in the economy; however, we should be cautious about how Google uses or abuses this position of power and the data it enables them to collect. The new Google Pay is an ambitious effort, which could be met with regulatory scrutiny. But, it could also enable competition against the archaic legacy banks and financial institutions.

Investors Back “Challenger” Banks
Speaking of mobile-first banks targeting the next generation of financial services customers, online banks are on track in 2020 to raise even more than last year’s $5.3B in funding rounds, according to Conor Witt at CB Insights. One example is Chime, which recently raised $485M at a $14.5B valuation.

Research Supports Wearables
In a preprint (not yet peer reviewed) article, researchers from the TemPredict study showed that Oura Ring’s temperature monitoring could predict the onset of COVID-19. I’ve written about the promise of health monitoring devices in the past, and this latest research indicates that 76% of the first 50 study participants to be diagnosed with COVID had detectable fevers before reporting any symptoms of being sick. The authors conclude that corroborating temp data with other markers, such as increased heart rate and decreased heart rate variability, is an important method for using wearables to predict illness. A healthy heart has highly variable beating, but when the body enters any sort of fighting mode, the heart starts to beat in a more regular rhythm.

Unleashing Game Developer Creativity
Embark, a game studio owned by Nexon, is working on leveraging the Unreal Engine with new layers of AI and tools to transform the way games are developed. The old way still involves a lot of manual processes (like coloring leaves by hand) and does not scale to meet the design needs of today’s always-updating virtual worlds. A couple of interesting blog posts (here and here) highlight the opportunity to shift game developers from repetitive tasks to more creative ones.

High on Social Media
In the ongoing political theater hearings last week, Dorsey and Zuckerberg were asked about whether or not social media is addictive. While Zuck said the research was not conclusive, Dorsey more thoughtfully responded: "I do think, like anything else, these tools can be addictive, and we should be aware of that, acknowledge it, and make sure that we are making our customers aware of better patterns of usage. The more information the better here." When the two CEOs were asked if they had seen the Netflix docudrama, The Social Dilemma, which explored the intentionally addictive nature of social networks, Zuck said: “I’m familiar with it” while Dorsey said he had not seen it.

Reinventing the DJ for the Information Age
It seems a poor assumption that the big tech platforms have devised the best algorithms to serve their customers. Many of their searches/feeds are optimized for ads, clickbait, or addictive hooks (just don’t tell Zuckerberg that), or to reward a set of user behaviors that a platform’s machine learning routines have deemed useful. In the Liberty’s Highlights newsletter a couple of weeks ago, the author posited that Netflix should allow third parties to program channels for viewers. Expanded broadly to all types of information/media consumption, perhaps folks would prefer to have human curators – content jockeys – whose selections they prefer over those of a purely computational algorithm (with questionable and/or oblique intentions). I am reminded of the music genome project at Pandora (I still remember watching musicians sit and categorize stacks of CDs at the Pandora headquarters), which, in my opinion, created a richer, more enjoyable music streaming service. Jack Dorsey expressed excitement over the possibility of third-party Twitter feed algorithms (first suggested by Stephen Wolfram), and seemed to indicate that the best algorithms might come from outside the company. All this dialog makes me thoroughly intrigued by the idea of having a smorgasbord of algorithms guided by a diversity of human views for myriad use cases and consumers. As Liberty suggested, an entire economy of influencers and corporations could emerge. I can also foresee all sorts of echo chamber risks and potential abuses (although we seem to have that situation now, albeit with less consumer upside). In an ever-changing landscape of disruptive innovation, no company should have the hubris to believe they have a monopoly on the best algorithm for anything.

Miscellaneous Stuff
Hollywood’s Hollywood Obsession
One of Hollywood’s biggest obsessions continues to be Hollywood itself. Variety covers David Fincher’s new Netflix movie, Mank, a behind-the-scenes look at the making of Citizen Kane that’s centered on the writer of the classic film. Other projects are in the works for movies about making the movies Chinatown and The Godfather. Perhaps the curious habit exists because the creative forces of movie making make for good drama, as Fincher explains: “I believe filmmaking owes a lot more to demolition derby than it does to neurosurgery. It’s a miracle when it goes off the way you had it in your head. For the most part it doesn’t.” Fincher, for his part, was happy to partner with Netflix on a big-budget, somewhat obscure, black-and-white film, and he thinks the picture quality of in-home TVs should continue to make streaming a good venue for big films. Maybe Hollywood could go even more meta and make movies about the making of movies about Hollywood – perhaps a behind-the-scenes movie about the making of the meta-Hollywood movie The Player.

The Lamentable Rise of “Ambient TV” (aka Netflix)
Fincher’s project stands in contrast to what the New Yorker recently called “ambient TV” - borrowing from Brian Eno’s description of ambient music: “as ignorable as it is interesting”. The author Kyle Chayka went into more detail in his newsletter: "Streaming wasn’t supposed to be a passive viewer experience: we pick what we want to watch, when we want to watch it. But the profusion of ambient shows turn streaming into a passive experience like cable, where we just leave it on and pay attention to it or not. Netflix produces ambient content intentionally, because that’s how some people use its service." This idea is something I was getting at in SITALWeek #269 when I talked about the increasingly formulaic Hollywood content that Netflix is amplifying. Meanwhile, a world of creative expression is exploding on Twitch, YouTube, social networks, video gaming, etc. and is increasingly gaining traction with viewers.

Stuff about Geopolitics, Economics, and the Finance Industry
Extreme Land Demand
Land for new houses in the US is becoming increasingly hard to find with the pandemic activating latent demand from the 30-something sneaker wave (the bolus of ~1M extra people entering their 30s looking to homestead) and accelerating the burgeoning urban-to-suburban migration. John Burns Real Estate Consulting reports: “96% of top land brokers rated their markets as Hot or On Fire during 3Q20”. Given the time it takes to buy, provision, and permit land, the problem of land scarcity isn’t likely to be solved anytime soon. Meanwhile, Redfin reports a record gain for home sales of 24% and a 100% increase in demand for second homes (y/y for October). At the opposite end of the Millennial and Gen Z demand surge are the Boomers, who are largely remaining in their homes longer than prior generations. Once Boomers do decide to downsize (or hit the road in RVs) over the coming decades, it should create a steady stream of available housing in established suburban neighborhoods. It will be interesting to see if retailers can shift to accommodate the flood of younger, more dispersed homeowners, who are likely accustomed to the amenities and conveniences they had in cities, like same-day delivery.

Outsmarting Your Brain: Substituting Pattern Recognition with Adaptability
Last week I discussed Karl Friston’s free energy minimization theory of how the brain operates:
This can be viewed as minimizing free energy, which is simply the difference between what you expect to happen and what your bodily senses are telling you is actually happening. For example, if I expect that I will warm up by stepping from shade into sunlight, and then proceed to do so, odds are the temperature receptors in my skin will confirm that prediction – no surprise and minimized free energy. Underlying the free energy principle is the idea that the brain is a Bayesian probability machine. We talk a lot about Bayesian logic – the constant, objective adjustment of prior credences based on new information – when it comes to picking stocks. If the brain acts as a Bayesian machine, it will constantly adjust predictions based on new sensory inputs.
The better the predictions, the better the metabolic efficiency.

As I reflected more on this topic – and listened to Lisa Feldman Barrett discuss her insightful new collection of essays, 7 and a Half Lessons about the Brain, on Sean Carroll’s podcast last week – I decided it would be helpful to connect a few more concepts. There are multiple lines of reasoning from various fields of study that all seem to point to this idea that our brain makes a prediction about the world and then tests that model against sensory input, adjusting the model as necessary. In other words, our brain has preconceptions about what we are experiencing before our current senses have a chance to exert an influence. Feldman Barrett characterizes the hypothesis-then-test neural algorithm as allostasis, or “automatically predicting and preparing to meet the body’s needs before they arise” (p. 8). Whatever we call it, there is a clear process by which our brain is constantly making predictions about the state of the world before checking against internal and external inputs. This is an inversion from the way we like to think the brain works, i.e., it seems to take in inputs, and then make a decision, which yields a false sense of agency over the entire process of thinking.

My first exposure to this idea was in Damasio’s book, Descartes’ Error, many years ago. Damasio teaches us that the concept of the mind as a separate intellect – a blank slate that objectively parses the body's sensory input with logic and an appropriate dose of emotion to decide upon a course of action – is completely upside down compared to how we really operate (falsifying “I think therefore I am” as a philosophy). Rather, the mind and body are deeply interconnected, with the body constantly directing the brain – often without us consciously knowing. Which brings us to interoception, a concept I discussed briefly in #268: “Your heart doesn’t race because you are scared, you are scared because your heart is racing”. When we learn to hone our interoception and read the signals from the body – looking at the ‘raw data’ instead of relying solely on our brain’s ‘digest’ version – we can minimize the often counterproductive mental states engaged by our brain’s response to external stimuli, which tend to rely heavily on historical and evolutionarily-ingrained pattern recognition.

It’s not just our instinctual, flight-or-fight responses that may be out of sync with modern life, but also what we might consider our more intellectual world-building models and decisions, which tend to go awry simply because we are (sub)consciously using the same historical and ingrained data sets to make predictions. We know from complex adaptive systems that making predictions is largely a fool’s game – it’s much better to have a ‘beginner’s mind’ blank-slate approach to what might come next rather than to artificially narrow the boundaries of possibility through formalized prediction. Additionally, trying to frame future hypotheticals based on historical pattern recognition is highly susceptible to false positives/negatives (from a basic probability standpoint) just given the sheer volume of information we now deal with on a daily basis and the difficulty of distinguishing signal from noise.

So, if the human brain has been honed to navigate the world by ensuring that our experiences match up to our predictions (i.e., reducing free energy or maintaining allostasis/homeostasis as Friston, Feldman Barrett, and Damasio would respectively say) by relying on a narrow historical lens, then navigating through our current era of mushrooming uncertainty (thanks to ever-accelerating technological progress, disruptive creation, and informational deluge) makes for an unnerving, uphill battle. Indeed, it’s profoundly uncomfortable to face an unknown future armed only with narrow predictions (like an investor relying on DCFs and earnings estimates or a CEO running a company based on a five-year strategic plan).

Luckily, it’s entirely possible to retune your way of thinking to accord less weight to prior models/experiences and focus instead on adaptability and learning to take advantage of whatever happens, good or bad. Just understanding the brain’s limitations is a first step towards overcoming the inherent biases built into the way it operates. From a Bayesian perspective, the typical decision maker might accord significant credence to prior convictions; however, the best way to approach our world’s complex systems (such as the economy) is to view those priors with a hint of suspicion. I can have high credence in certain things I understand that repeat with high fidelity; for example, past experience has led me to believe my desk is solid, and if I place my coffee cup on its surface, it won’t pass through the wood. Unfortunately, it turns out that there are very few predictions that can safely rely on such solid prior evidence. By becoming mindful of the idea that decisions you are making are overly reliant on past models, you provide yourself an opportunity to consciously step in and downweigh those assumptions. Also, remind yourself that there are alternate methods for dealing with an unpredictable future such as our prioritization of adaptability as the new margin of safety when we analyze companies.

✌

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is simply an informal gathering of topics I’ve recently read and thought about. It generally covers topics related to the digitization of the global economy, technology and innovation, macro and geopolitics, as well as scientific progress, especially in the fields of cosmology and the brain. I will frequently state things in the newsletter that contradict my own views in order to be provocative. Often I try to make jokes, and they aren’t very funny – sorry. 

I may include links to third-party websites as a convenience, and the inclusion of such links does not imply any endorsement, approval, investigation, verification or monitoring by NZS Capital, LLC. If you choose to visit the linked sites, you do so at your own risk, and you will be subject to such sites' terms of use and privacy policies, over which NZS Capital, LLC has no control. In no event will NZS Capital, LLC be responsible for any information or content within the linked sites or your use of the linked sites.

Nothing in this newsletter should be construed as investment advice. The information contained herein is only as current as of the date indicated and may be superseded by subsequent market events or for other reasons. There is no guarantee that the information supplied is accurate, complete, or timely. Past performance is not a guarantee of future results. 

Investing involves risk, including the possible loss of principal and fluctuation of value. Nothing contained in this newsletter is an offer to sell or solicit any investment services or securities. Initial Public Offerings (IPOs) are highly speculative investments and may be subject to lower liquidity and greater volatility. Special risks associated with IPOs include limited operating history, unseasoned trading, high turnover and non-repeatable performance.