SITALWeek #277
Welcome to Stuff I Thought About Last Week, a collection of topics on tech, innovation, science, the digital economic transition, the finance industry, ancient food carts, 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: The importance of seeking 2020's lost randomness; Intel’s paths forward; cash everywhere and the specter of inflation; AI-driven nuclear fusion; multiplatform gaming; lockdowns drive better sleep; negative-sum business model of food delivery; the limitless potential of reinforcement learning; and lots more below...
Stuff about Innovation and Technology
Biohybrid Smellicopter
University of Washington scientists have developed the smellicopter: “an autonomous drone that uses a live antenna from a moth to navigate toward smells”. The video shows the drone navigating with an antenna of a moth connecting two wires, enabling conversion of chemical signals (odors) to electrical ones. Following smells autonomously could allow it to find gas leaks or unexploded bombs.
20 Seconds of Fusion
The Korean Superconducting Tokamak Advanced Research (KSTAR) nuclear fusion device, aka an artificial sun, was able to sustain plasma temperatures of 100 million degrees for 20 seconds. That’s more than twice the eight-second record achieved in 2019. The research group has a goal of five minutes of operation by 2025. DeepMind’s Demis Hassabis also let slip that Google is working on AI to control plasma in nuclear fusion.
Amazon’s ‘Couture’ Clothing
Amazon’s new “Made for You” clothing uses images taken by your phone to create a body double. You can then customize clothing (for now it appears to only offer a t-shirt for a reasonable price of $25) and see how it will appear on you.
It’s Among Us
Popular bluffing game Among Us continues its meteoric rise from 295M monthly active users in October to nearly 500M in November. That’s enough to beat out top mobile game rivals Candy Crush and Pokémon Go. Among Us appears to be the most popular game ever as measured by the number of monthly players! And, game-maker InnerSloth only has four employees. As more games go multiplatform (allowing simultaneous play across mobile, PC, and console) we may see more fickle behavior and boom/bust cycles for games than in the past.
COVID Streaming Surge
The number of streaming subscriptions grew 50% in 2020 to just over 250M total in the US, with the household average going from 2.7 to 3.1 services according to data in the WSJ.
Better Living through Lockdown?
Smart-ring maker Oura discussed its 2020 data suggesting that social distancing might be good for our health. I might modify that to say physical health (and perhaps not so much mental health). The data from 45 million nights of sleep showed a significant drop in nighttime resting heart rates (RHR), which is typically a sign of better health and recovery, from 2019 to 2020. Oura also discovered that RHRs follow seasonal patterns – peaking in December and troughing in the summer, although the exact reasons for that remain under investigation. The lack of going out on weekends was a big contributor to lower RHRs in 2020, as alcohol, lack of sleep, and eating late can all increase RHRs. People are also going to bed earlier and waking up later (perhaps the effect of not having to commute in the morning as much). I can’t help but think about how much of this benefit is accruing to white collar workers who have fared much better during the pandemic – another example of widening inequality.
Instacatch-22
Americans spent $53B less on food and drink through November 2020 (compared to the same 11 month period of 2019) as grocery and liquor stores gained $80B in sales while restaurants/bars lost $133B according to US retail data tallied by Jason Goldberg of Publicis. Despite near-ideal tailwinds in 2020, the business of delivering some of those food/beverages continued to search for a model to support the value placed on it by investors. I have been tempted, like a good Bayesian, to update my paper from 2019, The Evolution of the Meal, but it’s not clear the tumult of 2020 changed my credences. It continues to appear that vertical integration and a mix of store pickup options are currently the only path to profitable digital food sales of any significant size. Beyond that, a profitable delivery model will also likely require subscriptions and routing to grow beyond a large niche market of affluent households. I continue to think most food delivery service providers find themselves in a Catch-22 (see #252): if they want to be profitable, they will need to begin to build and operate their own stores/restaurants; but, if they want to scale quickly, they will also need to maintain their existing store/restaurant customers, which are increasingly not best served by third-party delivery services. The WSJ reported on the predicament of Instacart, which had its first profitable month in 2020 since being founded in 2012. However, their grocery store customers complain they don’t make any money off the orders after the 10% commission. Most delivery businesses today are zero-sum or perhaps even negative-sum propositions (in which the end consumer benefits at the expense of the store/restaurant). If you further take into account the treatment of the contracted drivers, then it seems clearly negative sum. And, the delivery businesses are still being subsidized by investors making an increasingly narrow prediction that scale will improve economics over time without significant changes to the business model.
Intel’s Paths Forward
Intel was in the news as the company’s ongoing manufacturing struggles caused an activist investor to send a letter to the board. The letter seemed to simultaneously suggest that Intel consider exiting or selling its semiconductor fabs and also enter the business of designing different, custom chips for cloud giants, which would require access to leading-edge fabs and a host of design capabilities Intel doesn’t currently possess. Putting aside the poorly-informed suggestions in the letter, there are some good options for Intel to get back on track. First, replace the financial-minded CEO with a product person, preferably someone with really strong engineering chops. Second, Intel should keep and milk its cash cow x86 manufacturing capacity (focusing on leading-edge rather than bleeding-edge manufacturing). X86 chips might be losing share of compute workloads as Arm moves into PCs, but it will be a predictable fade that will allow Intel to generate FCF for several more years. Meanwhile, x86 servers will stay in high demand even as workloads grow on Arm, GPUs, and custom silicon – it’s a situation of “and” not “or” for x86. Third, Intel should outsource the leading-edge portion of its chips to TSMC or Samsung (this can be done in the world of chiplets, where components are made separately and then assembled into one 2.5D/3D multi-die package). Fourth, Intel should get together with Apple, Amazon, Google, Microsoft, and others to form a JV with either TSMC or Samsung. This should be a massive $100B+ effort to build a leading-edge chip supply chain on US soil for geopolitical reasons (and common sense). And fifth, Intel should put in a $50B+ rival offer to acquire Arm with the same JV fab consortium of Internet platforms and foundries. That last item is a long shot (and there are a lot of reasons why it’s an insane suggestion, but it sure would be fun to watch play out). Among other things, it would be a $50B acqui-hire of Simon Segars to be the CEO of Intel. Those pipe dreams of investing in their future aside, what will most likely happen is Intel will just buy some more stock back and end up as one of the more spectacular examples of roadkill in the brutal technological disruption cycles. Failure to adapt creates an existential crisis, which can happen for a number of complex reasons even in a tech industry as old as semis. If you’re interested in learning more about semi tech and where the sector is headed, check out Brinton and Jon on The Knowledge Project podcast and our whitepaper and podcast from earlier this year.
YouTube Phenom MrBeast
MrBeast (Jimmy Donaldson) spent the last several years figuring out what types of videos would produce the most views on YouTube and leveraging the site’s mysterious algorithm to go viral. “Make a clip too long, no one watches or wants to watch another. Make one too short, people won’t linger. Use a bad thumbnail photo or title and no one will click....His videos often blend three popular YouTube genres. There’s the outrageous challenge, such as staying inside a block of ice for a day or being the last one to leave a vat of ramen noodles. There’s the celebrity guest appearance...And there’s the reaction video...” According to Bloomberg, his videos cost on average $300,000 to produce, and MrBeast is said to generate tens of millions of dollars in advertising across his various streaming and social media accounts. The MrBeast YouTube channel has nearly 50M subscribers. Donaldson was in the news recently for launching a nationwide burger chain overnight (well, likely with months of planning) leveraging existing kitchens in 300 restaurants. The MrBeast Burger app was the 2nd most popular free iOS download on the launch day.
Miscellaneous Stuff
Ancient ‘Hot Dog’ Cart
Scientists unearthed a very well-preserved fast-food stand in Pompeii. The brightly painted stand – known as a thermopolium – sold various dishes, including duck, pig, goat, fish, and snails, out of earthenware bowls built into the counter along with wine and hot beverages. In addition to the striking paintings on the food stand, it is quite remarkable how little has changed in 2000 years, except perhaps that now you can get fast food delivered thanks to generous investors.
Is Minimizing Free Energy the Ultimate Reward?
David Silver is the principal reinforcement-learning research scientist at Google’s DeepMind for AlphaGo, AlphaZero, and MuZero, an AI engine that learns on its own as described in a recent Nature publication. This Wired interview contains several interesting comments from Silver, who believes that a machine will eventually be able to do anything a human can: “The brain is a computational process, I don't think there's any magic going on there”. Silver views reinforcement learning, i.e., reward-based learning, as the entirety of AI (and, by extension, how the human brain may work). According to Silver: “the reward-is-enough hypothesis...says that the essential process of intelligence could be as simple as a system seeking to maximize its reward, and that process of trying to achieve a goal and trying to maximize reward is enough to give rise to all the attributes of intelligence that we see in natural intelligence”. Recall in SITALWeek #271 we discussed Karl Friston’s principle of free energy (i.e., the brain is optimized to minimize prediction error). Friston believes that giving AI the directive to minimize surprise would make for a superior training model. Importantly, because minimizing free energy is an internal (rather than an external) reward system, it’s a good candidate to be the ultimate reward in the hypothesis Silver referenced. I haven’t seen mention of DeepMind focusing on free energy, but a 2018 Wired article notes that some of Friston’s students have gone on to work there. David Silver was also on Lex Fridman’s podcast back in April.
Seeking Randomness
Several years ago, engineer Max Hawkins wrote an algorithm to randomize his life. He had realized that his life was so predictable that it was entirely questionable as to whether he, as a conscious organism, was even necessary for his routine to be dutifully completed. So, Max let a randomizer take over. Instead of deciding where to eat, shop, socialize, exercise, or even live, the program would make the calls (like hailing an Uber for a destination unknown...or ordering a move to Mumbai) with no input or oversight from Max. Thanks to his years-long experiment, he discovered an astounding, enjoyable array of experiences he would have missed if the algorithm hadn’t forced him out of his comfort zone. The Debugger article from Medium goes on to suggest that the increasing dominance of algorithms from social media, streaming services, etc. are exerting control over our lives and digging deeper and deeper ruts – narrowing and reinforcing our routine rather than expanding our horizons (which, for globally-connected network platforms, seems rather disappointing). This article struck a chord with me because the invariant repetition dictated by 2020’s pandemic lockdown has served to dig those grooves so deep that we are now desperately lacking randomness – the ad hoc, unexpected, and crucial connections that make life interesting and create opportunities for creativity and growth. On the surface, the fact that we rebel against monotony might seem to contradict Friston’s theory of free energy minimization being the driving force for life – isn’t lack of surprise what the brain wants? Recall, however, that curiosity and a drive to explore are critical components of the free energy principle – if you don’t create a knowledge map of your environment, you leave yourself open to surprise. Apparently, when everything is the same, there’s no new data for our Bayesian-based neural prediction machines to chew on, creating an internal crisis. Our rate of learning even impacts our perception of the passage of time – without new experiences, time seems to move more quickly (see this Vox article on COVID’s impact on our sense of time). For many people right now, the grooves are so deep the record player isn’t even playing music anymore – a potentially dangerous situation. The Debugger article reminds us that once we get used to following orders (e.g., from questionable algorithms, ads, and/or the echo chamber of social media) the mind control can lead to unwanted behaviors with potentially devastating consequences. It’s good to remember that it was curiosity, randomness, and luck that got humans to where we are today instead of exclusively remaining part of the fossil record. If you find yourself still stuck in a 2020-era rut, it might be worth thinking about ways to introduce more positive randomness into your life.
Stuff about Geopolitics, Economics, and the Finance Industry
$3.4 Trillion
The seven largest tech companies added $3.4T in market cap in 2020. And, coincidentally, S&P 500 companies hold $3.4T in cash on their balance sheets (up from $1.3T in 2019) as US companies overall borrowed $2.5T in 2020. The FT also reports that, according to Leuthold, the number of zombie companies – those with debt payments higher than profits for three years in a row – has risen to an all-time high (e.g., an estimated 15% of US small-cap companies). Household cash levels also rose in 2020, in part thanks to the financial health provided by the previously mentioned $53B decline in spending on food/beverages as well as the $66B decline in spending on apparel (it seems that dressing up to go out to eat puts a big burden on our wallets!). Even colleges are getting in on the cash-raising action – selling a record amount of debt in 2020, according to the WSJ, as investors look to earn a little scratch on all that cash sitting around (a situation reminiscent of charitable foundations selling debt earlier this year). The specter of short-term inflation is slowly going from translucent ectoplasm to potentially visible ghoul as an economic post-vaccine rebound seems increasingly likely. The Economist reports that, in 800 years of data, typically inflation goes up after pandemics. It would seem crazy not to expect pockets of rising inflation short term. Long term, however, it still feels like a tug of war between cash in the system and deflationary/disinflationary trends that have been underway for 40 years – and potentially just beginning to accelerate in the AI Age. For more on that topic, see the end of SITALWeek #258: Can We Harness Technology’s Deflationary Pressure?
Amplified Vaccine Skepticism
The first 14 million doses of vaccines will start expiring in late January, and with the current paltry pace of injections, more than 4M doses may very well end up wasted. The NY Times reports on the myriad reasons the US has failed so far to effectively administer vaccines efficiently. And, the LA Times indicates vaccine skepticism is one of the reasons that 20-40% of frontline healthcare workers in LA County decided to delay or pass on being vaccinated. Facebook, YouTube, and (to a lesser extent) other Internet platforms not taking responsibility for the misinformation on their sites is now putting human lives at risk on a grander scale than ever before.
✌
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.