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: structural batteries; health-monitoring apps and tracking vaccines; NZS discusses semis on The Knowledge Project podcast; virtual theme parks; cloud-based PCs may eliminate need for local processing; pros and cons of layering AI onto real-time conversations; history of indoor air quality and how to fix it; what humans and AI can learn from the free energy principle; rising fear in China; and lots more below...

Stuff about Innovation and Technology
“Semiconductors: The Ultimate Bargaining Chip”
Jon and Brinton joined Shane Parrish for a wide-ranging discussion of the semiconductor industry’s history, status, and future prospects on The Knowledge Project podcast, and it’s a must listen! The podcast and transcript are available to TKP members here, and there will be a free version available soon on YouTube (as of today’s newsletter, it had not yet been posted). Many thanks to Shane for the engaging conversation.

Help Me Obi-Wan
Researchers at Samsung have used a special backlight with a beam deflector to tilt the angle of laser beams to create a realistic 3D holographic display with improved viewing angle. They haven’t quite reached Star-Wars-grade tech yet, but the 10-inch device only requires a depth of about 1cm to fit the optics. The team detailed their new system in Nature Communications.

Animatronic Humanoid with “Illusion of Life”
Disney researchers have taken their animatronic tech to a whole new level with vastly improved human characteristics. The new robot heads have much more fluid movements and better gazing functionality to hold eye contact or mimic actions such as reading. (Gazing, in particular, plays an important role in non-verbal communication.) Other mimicry aspects – including blinking, respiration, and head tilt – also contribute to the life-like result. There is a short video showing off some of the new features. With AR coming over the next few years, it will be easy to overlay these types of realistic humanoid expressions onto a plain robotic face using software and sensors, but these mechanistic hardware achievements from Disney are nonetheless cool even if they are just for The Hall of Presidents.

Refocusing on Indoor Air Quality
Before modern air conditioning was developed in the mid-20th century, preserving/dissipating heat (depending on local climate) and exchanging stale air (once we understood the dangers of airborne diseases) were important elements of building design, according to this Wired article. With the onset of air conditioning, however, architects started focusing their attention elsewhere, leading to, in many cases, a near-total reliance on HVAC to heat, cool, filter, and ventilate indoor air. Unfortunately, the filtration and ventilation functionalities of HVAC systems have long-been ignored and neglected, and now residential and commercial buildings need a big refresh to deal with their stale, germ-laden air. However, revamping HVAC is likely to require systems that use more energy as they run more often and/or push more air through filters. Better ventilation could have several health benefits – from reduced asthma to improved cognition – besides just limiting viral spread. As part of this refocus on indoor air quality, I expect we will also see an investment in data collection/analysis, with air quality sensors that feed into algorithms and connect across buildings and homes to improve air quality and health. The HVAC upgrade cycle is an example of what Jon describes as “the next big thing is there’s no next big thing” for the semiconductor industry; instead, there are a lot of new use cases popping up all over the economy enabled by chips, sensors, data, and the cloud.

Structural Batteries/Battery Packs and Energy Scavengers
Housing batteries inside structural elements in mechanical devices is a practice that’s been around a while, and recently Musk discussed Tesla’s ambition to transition the EV’s “skateboard” battery pack into structural elements. Some researchers believe you could go a step further in commercial production and make the structural elements themselves store energy rather than just house batteries, e.g., a car’s frame would be made from thin layers of composite materials capable of storing electrical charge. This Wired article explores a number of different efforts underway to incorporate energy storage and discharge into the physical elements of cars, planes, and smaller devices.

And, if the device is small enough, it may not need batteries in the future. Arm’s Project Triffid aims to create chips and sensors called CRFIDs that can scavenge power from wireless signals and forgo the need for batteries. This breakthrough would enable a wide range of new use cases, e.g., light/temperature monitoring for stored goods, remote sensors for hard-to-reach locations, and fitness tracking shoes that never need to be plugged in to recharge.

Vaccine Tracking and Logistics
It’s possible that some public events and places will require proof of vaccination to enter, particularly for large indoor gatherings. There was a rumor in Billboard magazine that Ticketmaster would be requiring live-event attendees to verify that they have either been vaccinated for COVID-19 or show evidence of a recent negative test result (Ticketmaster subsequently said such enforcement would be up to the event organizer, and that Ticketmaster was working through how they could help accommodate verification). Monitoring proof of vaccination and test results would likely happen via third-party apps, like CLEAR, which could automatically control ticket validation for concert goers. Verifying health status will obviously be a complex orchestration, but one that may ultimately be required more generally for all types of access – from eating out to air travel.

I still believe wearables will play a significant role for ongoing health monitoring and access to public/shared space beyond COVID to mitigate spread of whichever virus manifests next. Wearables can provide real-time temperature, heart rate variability, respiratory rate, etc. to health-pass applications on an ongoing basis, providing a general assessment of health as opposed to a narrow analysis of one specific infection. And, I remain hopeful that the data from widespread health wearables will help shift the healthcare industry’s focus from treatment to prevention.

Another tech challenge related to the Pfizer vaccine is transport and storage – the extreme temperatures (−70℃/−94℉!) required to preserve the genetic material in the new mRNA vaccine create a host of logistical issues. It will be important to orchestrate a transport/distribution network that can monitor and maintain the temperature from manufacture to injection to ensure the vaccine remains viable. The tracking and monitoring of vaccines is another great use case for distributed low-power chips and sensors that feed into data analysis in the cloud. Recall in SITALWeek #266 I mentioned the collaboration between Amazon’s AWS and Carrier for tracking and monitoring the movement of food, medicine, and vaccines. Efficient vaccine transport seems likely to be remedied, but the ultra-cold, −80℃ freezers needed for storage beyond a few days are so expensive (not to mention finicky and difficult to maintain, according to SITALWeek’s Editor: innumerable research lab managers have no doubt been woken in the wee hours of the morning to respond to high-temperature alarms and frantically redistribute freezer contents from seizing to functioning units) that many rural hospitals won’t have the budget to store the vaccines for local distribution.

PokéMap
Niantic, the Google spinout and maker of popular AR game Pokémon Go, has been working on dramatically improving 3D mapping of the real world. Game players are contributing to the effort by submitting videos of public spaces. Niantic wants to eventually create a continuously updating 3D map of the world with the capability of identifying all objects and surfaces. Pokémon Go has raked in $1B+ in revenues in 2020, the game’s best year so far.

"Virtual Theme Parks" Future of Entertainment?
Owen Mahoney, CEO of desktop and mobile gaming company Nexon, discussed the addition of Disney veteran Kevin Mayer to the company’s Board of Directors on last week’s earnings call, characterizing the appointment as a recognition of the burgeoning transition of entertainment from passive to interactive. It’s much more likely that we will be active, embedded participants in the next generation of entertainment – “virtual theme parks” as Mahoney says – rather than consume it passively:
"But we also are of the belief that the media companies, the big entertainment companies in the next 50 years are going to be much different than the media companies and big entertainment companies in the last 50 years. And you've heard us talk about this before. We think that there's a massive secular shift going on in the entertainment business right now. It's not well understood. And to make a long story short, it's about the shift from off-line to online, from linear to interactive, from the physical to the virtual. And we think this is a very, very important secular shift that's going on...So as we work through these strategies, there's going to be a new set of linkages and a new flywheel that develops. And we think that virtual theme parks are going to be the center of the next 50 years, the way that physical theme parks were at the center of Disney's strategy last 50 -- the last 10 and 50 years. So we're very, very happy to leverage his advice and counsel on the Board of Directors."

Private 5G May Replace Wi-Fi
The WSJ reported on the rise of private 5G networks bringing high-speed Internet to rural areas using open-source software and unlicensed/lightly-regulated spectrum. 5G has benefits over Wi-Fi in latency and throughput, and it’s not impossible that we could see the new technology ultimately replace Wi-Fi at some point. Back in SITALWeek #229, we explored this article from IEEE that covers the possibilities afforded by private 5G.

Encouraging Inter-Monopoly Competition
We may have a lot of monopolies, but we also have a lot of monopolies competing with each other. While the DoJ is concerned about Visa’s acquisition of fintech startup Plaid, Facebook has proposed an entirely new payments network, FastPay, that is reportedly seven times faster than Visa’s. Yet, thanks to government fear, Facebook’s prior attempt to enter the payments market with Libra – and offer Visa some healthy competition – struggled to get off the ground with Facebook’s involvement. Going forward, it will be increasingly important to let monopolies try to enter each other’s competitive turf, but the government doesn’t seem to see it that way. Facebook engineers even used a cloud service from another “monopoly”, Amazon, to test the new transaction network using 48 Intel processors and 384Gb of memory to handle 160,000 transactions per second on computers that cost less than $4,000/month to run. The implementation of FastPay has been open sourced by Facebook.

Layering AI onto Zoom
In SITALWeek #267, I wrote about Otter.ai, which will record and real-time transcribe your video meetings for live or later analysis. Last week, I read about Macro.io, which adds a real-time visual overlay to Zoom meetings to keep tabs on each participant’s relative contribution to the conversation. The tool also lets you customize the Zoom interface, take notes, and work with various collaboration suites during a call. I’m somewhat apprehensive that layering on AI will change the conversation in bad and counterproductive ways, yet I also see the value in removing all types of internal and external bias while improving productivity. Take for example a team of investors trying to decide whether or not to buy a stock. If you had long-term data on who said what during meetings, including who talked the most/least, tone, disagreements, etc., and correlated that to the investment outcome in the portfolio, maybe it would help eliminate common cognitive biases and improve stock picking. I’ve not used any of these new tools, and, yes, I am afraid to know how much I talk during NZS team meetings!

Cloud-Based PC
Windows cloud desktops appear to finally be around the corner. Microsoft will be offering Azure-based remote Windows machines accessible on any local OS including Mac, iOS, or Android. I’ve used Amazon’s Windows desktop service before and found it to be sort of ok and quite expensive, but still better than Citrix. Hopefully, this new cloud-native offering from Microsoft is better, as there is little reason for most office workers to be running OSs/apps locally. Apple's Arm processor announcement last week was curious because we've been writing about and hoping for Arm to finally work on a local PC or Mac for so long I can't even remember when we first started talking about it. And yet, the argument for having that kind of local processing power – given cloud computing’s low latency and high bandwidth capacity – is increasingly diminutive. Outside of high-end gaming rigs, it's sort of becoming an anachronism to have beefy local processing.

Improving Performance with ASICs
Amazon has transitioned Alexa voice AI responses from Nvidia chips to the cloud giant’s own custom silicon, called Inferentia. The shift has resulted in 25% lower latency with a 30% cost reduction. More broadly, I expect there will be a significant trend going forward of ASICs (application-specific integrated circuits) replacing general-purpose chips for the highest volume workloads at the big clouds. Google’s TPU is another successful example. The high cost and lengthy timeline for ASIC development (starting around $50M and typically 18-24 months) will continue to limit the number of practical applications. Amazon’s chip was first announced two years ago and is used exclusively for the inference part of AI. (Inference is basically answering questions after a machine learning model has already been trained, with training largely still taking place on Nvidia chips.) According to AWS CEO Andy Jassy, other companies are using Inferentia on AWS, including social app Snap, which implies the new chip has broader applicability for inferencing against machine learning models.

Miscellaneous Stuff
Rare Earths in Space
A popular topic from SITALWeek #255 was a protein that binds selectively to the crucial rare earth elements (REE) necessary for just about everything that enables modern electronics. The protein-capture method would make it easier to extract REEs in a cheaper, more environmentally friendly way while also countering China’s stranglehold on the industry. In recent REE news, experiments on the International Space Station in zero and simulated-Mars gravity showed that it might be possible to extract REEs on asteroids or other planets using bacteria to speed up the process.

AI and The Free Energy Principle
Renowned neuroscientist Karl Friston believes that what defines life is a tendency to minimize the “gulf between expectations and sensory inputs”. 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 investing. If the brain acts as a Bayesian machine, it will constantly adjust predictions based on new sensory inputs (interesting aside: when it comes to processing news and stories, however, our Bayesian machinery has apparently been largely hijacked by social networking to the profound detriment of our objective reasoning skills – an idea I’ll return to momentarily). Friston expanded the idea of the brain as an inference machine only capable of evaluating information to also encompass action. According to his free energy principle, if the brain makes a prediction that appears incorrect, it can respond in one of two ways: accepting the surprise and modifying its version of the world (Bayesian inference) or by acting to make the prediction true (what Friston calls active inference). Returning to the example above: if I step into the sunlight and do NOT feel warmer, I might then realize it is because I am standing behind a nearly invisible piece of insulating glass. If I take the action of circumventing the barrier to feel the warmth of direct sunlight on my skin, I will have resolved the perception gulf.

But what happens if we are overwhelmed by prediction errors and can’t immediately correct them through thought or action? Indeed, “The free energy principle, it turns out, isn’t just a unified theory of action, perception, and planning; it’s also a theory of mental illness. When the brain assigns too little or too much weight to evidence pouring in from the senses, trouble occurs. Someone with schizophrenia, for example, may fail to update their model of the world to account for sensory input from the eyes…‘If you think about psychiatric conditions, and indeed most neurological conditions, they are just broken beliefs or false inference—hallucinations and delusions,’ Friston says.” So, concerning my point above, perhaps the social-network-amplified deluge of (mis)information we are continuously exposed to has overloaded our brain’s modus operandi, leaving many of us in a veritable fugue state, unable to correctly assign value to information who’s veracity we can’t confirm (or no longer have the capacity to question). It’s an interesting possibility to contemplate. And, not only can the free energy principle potentially explain the apparent collective human delusion of the post-truth era, it may also offer a better path toward creating smarter AI systems. The reward for machine learning using free energy minimization is reduced surprise – a readily adaptable internal goal – in contrast to the external input required in both pattern recognition and reinforcement learning, which have both proven fallible as training models. This article in Wired attempts to explain the free energy principle, and Friston was also on Lex Fridman’s podcast earlier this year.

Stuff about Geopolitics, Economics, and the Finance Industry
Innovation in the Shadow of Fear
The halting of the Ant IPO and the new antitrust guidelines aimed at China’s largest Internet platforms are sending a clear message that the government’s objectives take precedence over disruptive technology’s progress. Can China’s innovation trend survive the latest crackdowns? To put it more bluntly, can innovation survive in a climate of rising fear and declining freedom? Now that the rewards of money and status are diminished in China compared to the last two decades, will there still be sufficient individual motivation to move their tech sector forward? Importantly, what are the implications for the China's domestic chip development efforts, which have so far floundered? Force and authority have historically been the enemy of reason and creativity, and only time will tell if China can continue to innovate. See also: How I Learned to Stop Worrying About China in SITALWeek #226.

✌

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.