Aptos is Microsoft’s new standard font, succeeding Calibri, which was made standard around 2007. Calibri in turn replaced Times New Roman and Arial as Microsoft’s default fonts. I never liked Calibri, so I welcome the change. This is dating me, but TNR is still my default serif font of choice for writing large amounts of text.
In an example of the speed at which the government moves, in January 2023 the U.S. Department of State finally made Calibri (14-point) its standard font, replacing Times New Roman due to readability factors (sans serif fonts are regarded as being more readable than serif fonts). Just in time for Calibri to no longer be Microsoft’s default font. Prior to introducing Times New Roman in 2004, the State Department used Courier New, which is a monospaced font (otherwise known as a typewriter font where each character occupies the same width on the page).
Inflation seems to be back under control and the stock market is ripping again. It’s kind of crazy. More and more people are now literally buying into the prospect of a soft landing after tearing into Jerome Powell for first acting too late, and then acting too aggressively. I still think the other shoe is yet to drop. Interest rates went up 5%+ in record time, the labor market is still tight, and everything is suddenly back to the way it was before? I don’t buy it (literally).
🎪 Knuthenborg Safaripark (Maribo, Denmark) A surprising safari park in the south of Denmark where you can drive around and see a host of animals, including elephants, giraffes, tigers, lions (evacuated from Ukraine!), ostriches, etc. Also has a huge, super fun playground for the kids. We stayed overnight in a cabin that has a window looking directly into the tiger area. ★★★★
🎪 Den Blå Planet (Kastrup, Denmark) Denmark’s national aquarium. Pretty decent, with a fun water playground outside. ★★★
🎪 Frilandsmuseet (Lyngby, Denmark) Open air museum with what must be the world’s largest collection of old barns. Decent for the kids because there are a few activities there they can do, but not so much for adults, unless you love looking at barns. ★★
🗺️ Nivå Strand (Nivå, Denmark) A nice beach for young kids with a shallow gradient out into the ocean. ★★★
📺 Jack Ryan (Season 4) A reliable show to binge watch. I liked how all its seasons were self-contained. Unfortunately this is the final season, but it looks like it will be getting a spinoff. ★★★
🎬 The Menu (2022) Nice looking movie. Characters are a bit of a mess. ★★½
On Twitter
Next time you think about trading stocks, think about your competition.
The most decorated international math olympiad participant of all time is that competition. pic.twitter.com/xoCcukVM4Q
big news for font freaks: Times New Roman is being phased out at the State Department & replaced by Calibri. Secretary Blinken sent a cable to all embassies today directing staff not to send him any more papers with Times New Roman. Subject: "The Times (New Roman) are a-Changin" pic.twitter.com/HENLbRH3UQ
In under 5 days, Threads will have hit 100 million users. Mark Zuckerberg isn’t exactly the world’s most beloved billionaire, but in a case of the enemy of my enemy, Elon Musk is even more polarizing given his behavior over the last year. Lots of people have been looking for a Twitter replacement, but the issue with any social network is achieving network effects. This issue has plagued Twitter clones like Blue Sky, Mastodon, and Post News. And the overwhelming majority of users simply don’t care whether the network is decentralized or whatever. It just can’t be a ghost town. Threads, launching off of Instagram’s 1 billion plus user base, doesn’t have that problem.
The historic adoption rate is despite Threads not being released in Europe. This is likely due to the $1.3 billion fine levied by the EU in May, which essentially said that various transfers of personal data they make from the EU to the U.S. are not legal because U.S. laws render that personal data incapable of being protected to the standards required by EU law. There is a new EU-U.S. data transfer framework that is being worked out that in theory will address the issue. However, it is the third incarnation of such a framework. After the European Court of Justice invalidated the first two, it’s uncertain whether the third attempt will actually solve the underlying problems, which all stem from U.S. government surveillance.
I was in the European Union when Threads was launched, but I was able to sign up. That is most likely because my Instagram account was pegged as an American one. The rules around which privacy laws apply to whose personal data can be complicated, but in general you can think of the laws being based on residency (where you live), not citizenship (which country’s passport you hold).
It will be interesting to see how Threads fares. It feels pretty vibrant for such a new product, but we’ll see if it has staying power. I must confess, like many others, it will be a big dose of schadenfreude to see Meta successfully swoop in and take out Twitter after months of “I know better than everyone else” behavior from its mercurial owner. And then things will eventually revert back to people criticizing Meta for how slimy a company it is.
Further Observations
In the meantime, Twitter is suing legendary law firm Wachtell over their huge fees from having successfully defended the company from Musk’s attempt to backtrack from the acquisition. Yeah, good luck there. Apparently, the $90 million fee was composed of $18 million in hourly rates, and the remainder being a massive success fee. Wachtell is perhaps one of the only corporate firms that can charge success fees.
I’ve been in the miles and points game for a decade now, but I’m still learning new tricks. Here’s how to get free cruises by gaming casino status matches. There are various on ramps to this process, but the goal is to attain status at two Vegas casino loyalty programs (Caesars Diamond or MGM Gold) and go from there. With MGM Gold status, it’s possible to get an almost-free 7+ night Carnival cruise for a couple (or a very cheap one for a family). some may be offered balcony rooms and longer cruises. The free Carnival cruise can, in turn, be matched to a free Royal Caribbean cruise. You can also match MGM Gold to Caesars Diamond, which may generate additional free cruises and other benefits like free hotel nights and credits. I have almost 2 years of Hyatt Globalist status courtesy of a Bilt promotion earlier this year, but unfortunately the challenging part is actually making the time for a cruise.
Jubilee Jim Fisk and the great Civil War score (The Boston Globe) How a 30 year old made a fortune shorting Confederate bonds at the end of the Civil War in 1865. His scheme involved trading on the news 5 days before the news reached London.
Reviews
🍽️ Kong Hans Kælder (Copenhagen, Denmark) ✽✽ A decent meal that met but did not exceed expectations. Service was attentive (they gave Susanne a Danish menu and me an English menu) but impersonal. ★★★
🎪 Experimentarium (Hellerup, Denmark) Like the Exploratorium in San Francisco, but oh so much better and with exhibits for kids of all ages. Signage is in Danish and English. ★★★★
✈️ SFO-CPH on SK936 (A330-300, Premium Economy) Solid, new hard product, but the wifi didn’t work. Service was decent for a red eye flight. We bought a couple of these inflatable seat extensions so our kids could lie down to sleep and the flight attendants seemed cool with them (not all airlines are). Flight was full. Sit on the right hand side of the plane for a nice view of Copenhagen when circling to land. ★★★★
🛋️ United Club Lounge (International Terminal, SFO) Overcrowded, cramped lounge with passable food. You’re better off hanging out in the terminal. ★★
Charts, Images & Videos
This maths lecture, “A world from a sheet of paper” by Stanford Professor Tadashi Tokieda is very engaging:
A beech tree I passed by on a walk earlier this week
Last week at work, a sizable portion of my company gathered at a physical office in San Francisco and spent the week working physically together. It wasn’t an offsite — it was actual, regular work. It’s the first time in a few years since I’ve done that with a larger group of colleagues, and it felt refreshing. Everything is just much more dynamic. The side conversations during meetings, the option to have quick “drive by” chats, the ability to “read the room”, and even the small non-work diversions that keep you feeling connected to other people throughout the day. On the flip side, the interactions eat up more time, and finding focus is more difficult when you need to do work that doesn’t involve a discussion. But, the worst part of office work is by far and away the commute, and I was reminded about that when it took me 90 minutes to drive home in traffic (double the time it took to get in).
The toilets in the building we were working in were on the floor above us. That floor also happened to house the company shown in the photo below — a 2 year old AI company of 60 people that had literally just been acquired for a jaw-dropping amount.
MosaicML’s office
DPReview, the authoritative site for digital camera reviews, was scheduled to shut down after Amazon (its owner) conducted layoffs earlier this year. However, I was glad to learn that DPReview was acquired by Gear Patrol and will live on!
The Man Who Broke Bowling (GQ) Aussie Jason Belmonte’s two-handed technique made him an outcast. Then it made him the greatest—and changed the sport forever.
He has flown 23 million miles. Here are his travel secrets. (Washington Post) Several people sent me this article, but I learned about Tom Stuker a number of years ago. He got a great deal, but dare I say that United First would lose its novelty relatively quickly.
On Twitter
MTV Cribs profiled over 110 celebrity homes.
What was the most popular episode? It wasn’t a big mansion or exotic getaway.
It was Redman touring his modest Staten Island townhome.
Before filming, the MTV producers offered to rent the rapper a mansion but he had zero… pic.twitter.com/mMxkwrbtBC
The loss of the Titan, the submersible operated by a private company carrying five people attempting to view the Titanic, has captured the world’s attention for most of the week. After a multi-day search, vessel fragments found on the sea floor mean that the Titan had a catastrophic structural failure and imploded. Mercifully, it would have happened so rapidly that the passengers wouldn’t have realized what was happening.
“You know, at some point, safety is just pure waste.” That’s a quote from Stockton Rush, the late CEO of OceanGate who was piloting the Titan. A lot of reports now point to Rush confusing innovation with recklessness, having been repeatedly warned by experts in the past. It took several years for the warned-about risks to manifest.
“Regulations are written in blood,” is a common saying. In industries where serious injury or death is a real risk, innovation is slower for a good reason.
Hundreds are still missing from the disaster off the coast of Greece, where a boat carrying hundreds of people capsized. It is sad, but not surprising, that media coverage of the two news events has been so incredibly disproportionate.
Top 5
One reason attributed to declining birth rates in developed economies is a reluctance to bring children into a world that feels increasingly messed up. I periodically think about what kind of world my kids will be living in when they reach their 50s, by which time it will be the 2070s. Here is what I think are the most significant global generational issues that will dramatically shape how things unfold in the coming half-century:
The U.S.-China relationship (and the question of Taiwan)
Not making the list for me (yet) are AI, dwindling natural resources, interplanetary travel, and the impact of social media. Disagree with the list? Am I missing anything?
Further Observations
Welcome to summer. The summer solstice was on Wednesday — the astronomical beginning of summer in the Northern Hemisphere and the longest day of the year. In Australia, the seasons are defined by calendar months, instead (winter runs from June 1 to August 31).
It’s been an unusually cool spring and start to summer in the Bay Area, but Texas is experiencing high heat at the moment. The Dallas-Fort Worth area was forecasted to reach as high as 46°C this weekend. And, El Niño is returning…
The World’s 50 Best Restaurants list was released for 2023. In the past, some restaurants won so much that they are now on a list where they can’t win again. This year, restaurants in Lima have taken out 4 of the places, with Central taking out the top spot.
This week I learned: “Wakeup calls are a long-standing tradition of the NASA program. Each day during the mission, flight controllers in the Mission Control Center will greet the crew with an appropriate musical interlude.” Here’s an archive of wakeup calls from one of the Discovery missions. Some of the songs they played include Stairways to the Stars, Defying Gravity, and the theme from Stargate.
About the only time I listen to the radio these days is when I drive Susanne’s car to take my daughter to gym class on the weekend. The drive is only 10 minutes each way, but for the last 3 months, I will hear, without fail, The Weeknd’s Creepin’ at least twice, sometimes three times. (My daughter has heard it so much she knows the lyrics 😅.) It’s really annoying, and I recently stumbled across a video that explains why this is: Why radio stations sound the same.
Well, the whole Prigozhin thing was anticlimactic. What was the point?
📺 The Bear (Season 2) I hope they renew this for another season. My favorite episodes were Honeydew (a nice interlude in Copenhagen), Forks (nice to see a troubled member of the crew catch a break), and the incredible Fishes (a double length episode with an ensemble cast). ★★★★
'Rerender A Video' by Shuai Yang et al. uses a cluster of techniques to remove the flickering and temporal inconsistency that happens when using stable diffusion for video, while still being compatible with LoRA/ControlNet-driven content.https://t.co/WsRCSP26qgpic.twitter.com/9PeQKvBKiv
It’s the Juneteenth long weekend and for the only time in the year, our preschool is open on a holiday, so Susanne and I genuinely have the day off. Unfortunately, her Monday plans were unceremoniously derailed by not one but two clients on Friday who are pushing to sign deals early next week.
Susanne and I watched this YouTube video that plays the most popular song in each month since 1980. It’s almost an hour long, but it went by quickly. The 1980s had some amazing music, and I knew virtually every song that decade despite not being sentient (or even alive) for most of it. Studies have shown that people tend to stop discovering new music sometime around 30-35 years old. I started tapping out about midway through 2020, when I couldn’t recognize more than about 50% of songs and artists. It’s probably not a coincidence that this coincides with the start of the pandemic and the last time I had a regular commute for work.
My thinking on the Vision Pro has changed. After reading all of the glowing reviews, I think I’m going to buy one when it’s released… but only if Apple’s 14 day return policy applies to it. There are several Apple products that I’ve been indecisive about buying over the years. I ended up trying them out, only to send them back after a few days of use. They include the Apple Watch Ultra (found it to be a net detractor over the 20 year old automatic watch I currently wear that powers itself and doesn’t vibrate on my wrist every 5 minutes), the iPhone Pro 13 and iPhone 13 (I prefer the Mini’s form factor), and the AirPods Pro Gen 2 (I already have regular AirPods and the noise canceling that the Pros offer is unusable from a practical point of view when my kids are around… which is also when I probably most want to use them). I suspect the Vision Pro at home is an itch that will get scratched by trying it out for a few days.
The S&P 500 index is almost back to where it was in March 2022, when the Fed started the fastest series of rate hikes in history. Kind of crazy.
Anduril has prohibited TikTok/ByteDance from all devices used to access Anduril systems, including personal devices.
Two reasons: 1) The No TikTok on Government Devices Act requires this of defense contractors. It should have gone further. 2) TikTok is nation-state warfare.
Apple launched its $3,499 Vision Pro AR/VR headset on Monday. With sales tax, the price will be closer to $3,850 where I live. If you wear glasses, like myself, you have to buy additional lenses that magnetically attach to the device. I imagine those will cost a few hundred dollars.
Nonetheless, as long as Apple decides to continue developing the product, we can expect the Vision Pro to get cheaper and smaller over the years. They’ll probably release a non-pro Apple Vision at some stage.
My take? The Vision Pro probably represents the best shot this decade at seeing whether AR/VR is ready for widespread daily adoption (if it ever becomes ready — I’m doubtful).
As I wrote last week, the promise of AR/VR is mainly hamstrung by the hardware. For example, if I could magically have some of the functionality of the Vision Pro in the glasses I already wear (kind of like a heads up display), without materially increasing their weight or looking dorky, I’d be more willing to stand in line for it.
What we have today with the Vision Pro is the best mass-production headset that you can design with today’s technology. It’s entirely controlled by voice and hand gestures, so you don’t have to lug hand controls (the VR world’s equivalent of the stylus) around like most other headsets. Apparently the eye tracking is pinpoint accurate. The battery has been offloaded to an external battery pack you stick in your pocket, ostensibly to reduce the weight of the headset and thus user fatigue. The screen resolution is the best in the industry. They try to mitigate the issue of being blocked off from the physical world by including a creepy simulated-eyes display in the front, and giving the user the ability to control how much of the outside world they want to see. Because wearing glasses degrades the experience, they require glasses wearers to buy special Zeiss lenses that attach to the device and use those instead. They even have a solid approach to privacy (advertisers can’t do analytics on where you’re looking).
The headset is far from perfect. Reports are that the headset will still starts to feel uncomfortable after some period of use. The corded external battery is still something extra to lug around. But let’s acknowledge for a moment that there is an incredible amount of technology packed into this thing. It’s been pointed out that it represents the culmination of all the different things that Apple has been working on in its other products — from the Apple Watch’s digital crown to the iPad’s Lidar, to the iPhone’s TrueDepth camera for Face ID, and the M2 chip — and then some.
Who knows if it will find meaningful traction? We’ll find out next year.
Some additional random thoughts:
John Gruber has a great review of his 30 minutes with the Vision Pro. “Most impressively, and uncannily, the field of view seemingly exactly matches what you see naturally. It’s not even slightly wider angle, or even slightly more telephoto. There is no fisheye effect and no aberrations or distortion in your peripheral vision. What you see in front of your face exactly matches what your own eyes see when you lift the Vision Pro up over your eyes.” Sounds like Apple absolutely nailed the technology.
Here’s what Mark Zuckerberg thinks. I mean, it’s what you’d expect him to say, and he kind of has to publicly say that, even if he’s privately sweating. The cost is the main competitive advantage he has today. He points out that Apple’s portrayal of the Vision Pro world feels less social that Meta’s vision of their metaverse. While that may be true today (it’s a little tough to buy everyone in the family their own headset!), that will change if the price becomes more accessible. After all, the social aspect is just software and Apple has a bristling developer ecosystem. The hardware is the hard part.
The iPhone was niche when it was launched. But I was totally hyped about it and stood in line at the Apple Store in Sydney when the first one was released in the Australian market. (I had to ask my boss at the time for permission to leave work early for it. He’s one of the best technology lawyers I’ve ever worked with but even he didn’t appear all that curious about it and seemed amused at my earnestness!)
The iPhone is everywhere now (at least with the world’s richest 1 billion consumers) and the average price point today is, what, $1,000? When the iPhone was released it was around $600, whereas a BlackBerry could be had for around maybe half that (and Nokias were even cheaper). People will pony up lots of cash if something is good enough and essential enough. Phones are essential in today’s world, but were pretty much essential before the iPhone arrived on the scene too. AR/VR doesn’t have that positioning.
The simulated eyes thing are an interesting approach, but I find them creepy. I certainly wouldn’t want to interact with my kids that way. How we look at eyes for feedback is a primal thing… uncanny valley and all that.
The scene in the ad for the Vision Pro that I liked the most is where a woman is sitting on a plane and she tunes everything out by watching a movie (at the 0:49 mark). I wonder if the AirPods Pro will pair with the Vision Pro? It needs noise cancelling to be completely immersive. That’s escapism. Maybe a little bit dystopic too. From a practical point of view, it’s probably going to be annoying for people sitting next to you who need to pass by to use the toilet (both to get your attention and to maneuver past).
Apple was very careful not to mention the terms AR, VR, metaverse, or headset in its launch. It prefers the term “spatial computer”.
If weight fatigue is an issue, I can’t see it being used for more than an hour or two a day by most.
Bask Bank offers a savings account that pays interest not in cash, but in American Airlines miles. As interest rates have risen, so too has their mileage rate, and each dollar deposited with them now earns 2.5 AA miles per year. They are currently running a promotion where if you deposit a certain amount of money with them for at least 180 days, they will give bonus miles. $50,000 gets you a 20,000 mile bonus on top of the regular 62,500 miles. So, is it worth it, compared to sticking your money in a 26-week treasury bill?
Here are the factors:
Treasury Bill: A 26-week treasury currently pays out 5.483% p.a. when held to maturity. The interest is not subject to state tax. I regard the post-tax interest on the t-bill as the opportunity cost here.
Bask Mileage Account: $50,000 deposited for 6 months nets 82,500 miles. Those miles are taxable, and Bask issues a 1099 that values miles at 0.42 cents each, for a taxable amount of $346.50.
Here’s how it breaks down:
View From the Wing values AA miles at 1.3 cents per mile (cpm). OMAAT values them at 1.5 cpm. That feels about right. Just note that miles are subject to devaluation over time, so they tend to become worth significantly less as time goes on.
In theory, a one-way first class partner award booking from the U.S. to Asia can cost as little as 80,000 AA miles… if you can find availability.
Further Observations
I have the details of two of my credit cards committed to memory. This can be a surprisingly useful thing to do.
Donald Trump was indicted on federal charges on Thursday relating to improper handling of classified documents. “Wouldn’t it be better if we just told them we don’t have anything here?” Trump is alleged to have said. Uncharted waters…
Both Binance and Coinbase were sued by the SEC this week. The SEC’s complaint against Binance features this amazing quote:
The other hilarious thing is that the name of the “compliance officer” referenced is Alvin Bro. There are tons more choice quotes in depositions and other evidence released by the SEC.
International sports are so dirty. The merger of the PGA Tour with LIV Golf after so much moral grandstanding by the PGA Tour was shocking. Jay Monahan, the PGA Tour’s commissioner, is probably the golfing world’s most unpopular person at the moment. At least Messi turned down the Saudis.
📖 How To Invest: Masters of the Craft (David Rubenstein) Interviews given by Carlyle Group co-founder David Rubenstein with investing legends such as Seth Klarman, Ray Dalio, Stan Druckenmiller, Jim Simons, Orlando Bravo, Marc Andreessen and Michael Moritz. Full of intriguing insights — into their backgrounds as well as how they think. ★★★★
Nvidia currently has only $26 billion in sales versus $208-525 billion for the Big 4 (Microsoft, Apple, Google and Amazon). Just how high are these expectations? Nvidia now trades at over 38x sales and over 200x earnings. We’ve never before seen a price to sales ratio that high… pic.twitter.com/PyaVHd2g6j
Apple’s long-awaited VR/AR headset is rumored to be launched at WWDC on Monday. Around the internet, it seems that people are skeptical whether the device will become a successful product line for Apple, but no one is willing to rule them out given their track record of taking over markets they enter (see phones, tablets, wireless earphones, watches, etc.). However, the anticipated $3,000 price tag is a barrier that will probably see the first version remain a niche object, kind of like its top of the line Pro Display XDR monitor. Let’s see what they can do with it. (Meta also announced its Quest 3 headset. I bought a Quest 2 during the pandemic but it hasn’t seen much use.)
When I was in high school in the late 90s, I remember being totally hyped for the release of Ultima Online, one of the first massively multiplayer online role-playing games (MMORPGs) that, I’m surprised to see, is still operating today. The promise it held in my mind was a persistent world that I could explore with a bunch of friends. But it was not just about exploring dungeons and killing monsters—you could do stuff like form a guild, trade goods, and even buy virtual property. This meant it could be a place you could just hang out in and chat and “build a second life” in, so to speak, in between all the adventuring. (Second Life came a few years later.) It seemed to be a great place to hang out with real life friends and friends that I had made online.
The reality was decidedly less rosy. The first issue was that things were hamstrung by technology. Broadband in Australia was in its infancy and cost prohibitive, and accessing UO over a dial-up connection led to crazy bad lag. It was unplayable. The second issue, is that even when the connection was tolerable, Britannia (the name of the world in which UO is set) wasn’t a particularly fun place to actually hang out. While clearly there were enough people who the game engaged that it has survived to this day, I don’t remember playing it for more than a few weeks even though I played a lot of computer games back then.
MMORPGs are the original metaverses and they have been around for decades. While VR is tightly bound with the metaverse (as in, a way to fully immerse yourself in one), VR has broader applications. Nonetheless, just like UO, I think the early days of consumer-accessible VR are still pretty niche. Until a few things change, I see things remaining that way (although, with Apple’s reach, even a niche offering involves a lot of people).
First, the form factor is cumbersome. Fabricated mockups of Apple’s headset show a sleek pair of ski goggles which are easier to whip on than the Quest, but there is an external battery pack which I assume needs to be clipped somewhere to your body. This adds quite a bit of friction each time you want to use it, which contrasts to the “instant on” nature of its phones, computers, etc. (I remember the old days where computers didn’t have sleep mode and it could take 3+ minutes to boot up a desktop. That made using a computer a very intentional act.)
Second, current forms of VR disconnect you from the physical world. I imagine the Apple headset will have look-through functionality, but it still involves putting something in front of your eyes—your window to the world—blocking them completely. So it may be great for games and experiencing things that aren’t possible in the real world, but will that provide widespread daily utility?
Third, are there any real world interactions that are genuinely made better by VR goggles? I certainly don’t want to have meetings at work wearing them. Maybe there are some limited applications like doing virtual walkthroughs of buildings in real estate that are more convenient (but still inferior to actually being there).
Fourth, all of Apple’s major product lines are used by all age groups. It will be interesting to see what the uptake is of older age groups. Maybe it’s me just getting older, but the Quest 2 makes me dizzy sometimes.
All that said, MMORPGs have come a long way and some of them have developed very strong communities and are actually places in which a lot of people now hang out and socialize (and sometimes meet and get married IRL). And some people even make a bit of a living through it by selling virtual goods. I have no doubt that VR will continue to evolve and grow, but similar to MMORPGs, I’m not sure it will become something all pervasive, like the smartphone, or this new generation of AI technology that’s currently set the tech world ablaze. There’s nothing wrong with that, of course, but with Apple I suppose there’s a certain weight of expectation that whatever they do will literally change the world and become ubiquitous.
Augmented reality is probably the more interesting application, but I think it demands a different form factor to find more widespread adoption. But what would that be? Google Glass was an attempt, but it looked dorky. Maybe one day we’ll be able to integrate a chip and power source into glasses that are inconspicuous. Or maybe it will be something like this, where the technology just gets out of your way altogether.
Nonetheless, Apple’s headset is still probably going to be able to do some pretty cool stuff!
Further Observations
Every few issues of this newsletter, I will sign up a random friend or two to it. If you’re wondering why you suddenly started receiving these, that’s why. (Funnily enough, no one has ever replied to me asking, “Hey, why did I start getting these?”)
A shout out to the Dim Sum WhatsApp group! It was great seeing you all yesterday… and actually eating dim sum together for the first time in ages!
📺 Succession (Season 4) First class TV, and they successfully landed the plane with the series finale. ★★★★★
📺 The Diplomat (Season 1) Susanne put me on to this one. It’s really good. ★★★★
📺 The Night Agent (Season 1) Watchable, but pretty standard spy show fare. ★★½
🎪 Gilroy Gardens Theme Park (Gilroy, CA) Took the kids here over the Memorial Day weekend and it wasn’t as busy as I feared, which is probably telling. It’s a bit of a weird, aging, and tired theme park, and two of the rides broke down just as we reached the front of the line (which did not make for happy children). Membership is apparently tax-deductible. ★★
So I followed @GaryMarcus's suggestion and had my undergrad class use ChatGPT for a critical assignment. I had them all generate an essay using a prompt I gave them, and then their job was to "grade" it–look for hallucinated info and critique its analysis. *All 63* essays had
Inflation is coming down, but I remain skeptical that we’ll get a soft landing.
A soft landing supposes that, by keeping interest rates roughly where they are, inflation will start to moderate back down to its 2% target and low unemployment and positive productivity growth will be maintained. Paul Krugman suggests things will be more resilient than the doom and gloom suggest.
But interest rates have skyrocketed in the last 12 months, after more than a decade of ZIRP and a generation of workers who haven’t experienced an environment in which money was not cheap.
Rate rises have never occurred so quickly in history, and I don’t think they’ve had a chance to bite. The longer you keep rates at this level, the longer the situation gets worse for people and businesses who are sensitive to rate rises. People are not used to interest rates being over 1%, much less over 5%. The effects of high rates should eventually flow through in the form of increased bankruptcies, defaults, and restructurings. The regional banks aside, we haven’t seen much of that yet, but it doesn’t gel with me that whoever levered up to the gills in the last few years isn’t going to face some sort of reckoning when their interest rates have suddenly multiplied. It will take some time to play out.
We’re already seeing some signs of stress.
Credit card debt did not decline in Q1 like it normally does and delinquencies are rising. The commercial real estate market is a ticking time bomb. The regional banking crisis is still lurking in the wings.
Economic cycles have tended to consist of periods of gradual growth followed by short and sharp declines. I’m not sure why it would be different this time when the powder keg is so large. Eventually, something will spark a prolonged panic and precipitate the next recession.
And maybe we’ll see cycles that are more volatile — a sudden slowdown could push the Fed to cut rates back down to zero again in a knee jerk reaction, and you end up with an economy in pilot-induced oscillation (“when the airplane begins a departure from the desired flight path, and the pilot applies inappropriate, excessive or mis-timed corrections that result in ever-increasing excursions that threaten to force the airplane out of control”).
Perhaps something at play is that things happen so fast these days, and attention spans are short. The lack of bad things happening quickly might be interpreted as a sign that the bears are forever crying that the sky is falling, but we’re actually on a path to recovery.
I just don’t see it happening.
Further Observations
Ron DeSantis launched his presidential campaign on Twitter Spaces in a conversation with Elon Musk and David Sacks. As is common for Musk’s events, it started late and then just got more awkward as the servers overloaded.
Uber’s Chief DEI Officer was placed on leave after a couple “Don’t Call Me Karen” talks she hosted went sideways after various employees felt antagonized by what she said. Things feel a bit out of control, and it’s telling that I don’t really feel comfortable publicly expressing a view what was said, knowing that it could come back to bite me at some stage. All I’ll say is that this chilling effect on speech isn’t a good outcome.
This was NVIDIA after earnings release on Wednesday. That stratospheric price reflects $180B+ of market cap added in the space of an hour, all because of AI. (Only Apple and Amazon have recorded larger one-day increases.) Its PE ratio is now over 200.
Generative Fill, a new superpower integrated throughout Photoshop, launching in beta today.
Powered by Firefly, our generative AI family of models, Photoshop now let’s you summon new objects and augment creations layer by layer. Saves time, increases possibility, and pretty 🤯 pic.twitter.com/ARxhclFshO
The Republicans are playing chicken with the debt ceiling again. It’s a travesty how the system works here – namely, that the budget is approved by Congress first, and then as a separate step, Congress has to approve raising the debt ceiling so that the U.S. Treasury can issue new debt to pay for the spending that Congress already approved. This process is almost unique in the world. (Denmark is the other country that works this way, but they don’t practice brinkmanship over there.)
It’s almost impossible to imagine Congress actually taking things over the precipice, but no one is 100% absolutely sure that won’t happen, even though the consequences would be catastrophic if the U.S. were to default on its debt.
After months of President Biden effectively saying that he wouldn’t negotiate with terrorists who were holding the debt ceiling hostage, it’s telling that he is now negotiating. Clearly, there’s pressure from the Democrats who think the Republicans might let things go too far.
“X-date” is the deadline where the government effectively runs out of money. It fluctuates because it depends on day-to-day tax revenue collected and spending, but “as early as June 1” is the guidance that the Treasury has given.
The last time things got this down to the wire was in 2011 when the debt ceiling was raised with potentially only a day or two before the government ran out of money. In response, S&P downgraded the U.S.’ credit rating from AAA (the highest) to AA+. In the month leading up to the resolution of the crisis, the stock market fell 17%.
U.S. Treasury Bonds are regarded as “risk free” assets (or the closest thing you can have to risk free) given the reserve status of the U.S. dollar, the fact that the U.S. economy is the largest in the world, and the ability for the U.S. government to print more money at will. If the U.S. were to default on its debt, it would be entirely self-inflicted.
Interestingly, the market is not reacting like it did in 2011, staying relatively buoyant despite June 1 being a little over a week away. This means the market is expecting things will be resolved before x-date. It has been observed that this means that insurance against a debt default is selling for cheap. This insurance comes in the form of put options on the S&P index (via an S&P index ETF), or alternatively call options on the price of gold or against the VIX (volatility index).
On May 18, one unit of SPY (an S&P index ETF) was about $415 (about the same as where it was a month ago). A put option on SPY expiring on June 2 at a strike price of $385 (~7% under the ETF’s current value), was selling for $0.40. Should SPY decrease by 17% (the same as in 2011) over the next 2 weeks to $345, the option would return about $40 for a ~100x return. This has not gone unnoticed.
One potential solution to the crisis, where the U.S. Mint mints a $1 trillion coin and deposits it in the Federal Reserve’s Treasury account, has been taken off the table by Treasury Secretary Janet Yellen. As explained by Wikipedia: “The concept of striking a trillion-dollar coin that would generate one trillion dollars in seigniorage, which would be off-budget, or numismatic profit, which would be on-budget, and be transferred to the Treasury, is based on the authority granted by Section 31 U.S.C.§ 5112 of the United States Code for the Treasury Department to “mint and issue platinum bullion coins” in any denominations the Secretary of the Treasury may choose. Thus, if the Treasury were to mint one-trillion dollar coins, it could deposit such coins at the Federal Reserve’s Treasury account instead of issuing new debt.”
Further Observations
Michael Lewis’ book on Sam Bankman-Fried is called Going Infinite and it will be released in October. I’m looking forward to it.
Sometimes I’m so exhausted that I fall asleep putting my kids to bed and end up waking up a few hours later at around 11pm. After shaking off the wooziness, I then spend the next few hours relatively clear-headed and productive, catching up on emails, work, and household admin, before heading back to bed at around 2am or 3am. I was worried that this messed up schedule might become a bad habit, but it turns out that it’s a natural sleep pattern called biphasic sleep that was common in medieval times: “A couple of hours later, people would begin rousing from this initial slumber. The night-time wakefulness usually lasted from around 23:00 to about 01:00, depending on what time they went to bed. It was not generally caused by noise or other disturbances in the night – and neither was it initiated by any kind of alarm … The period of wakefulness that followed was known as ‘the watch’ – and it was a surprisingly useful window in which to get things done. ‘[The records] describe how people did just about anything and everything after they awakened from their first sleep,’ says Ekirch. … One servant Ekirch came across even brewed a batch of beer for her Westmorland employer one night, between midnight and 02:00.” That makes me feel a bit better.
A Berkeley student commuted from Los Angeles for an entire academic year. He did this ostensibly to avoid paying rent in the Bay Area (he had free accommodation in LA). From a pure money perspective, the numbers checked out. From a practical perspective, well… His daily commute averaged 4-5 hours each way, and involved a car, plane, and BART ride. He managed to schedule his classes so he only had 3 days a week for most of the year, but during some periods he did have a 5 day week. On some days, he needed to wake up before 4am to catch his flight, and some days his flight back departed at 9pm. But, he managed to make it to every class, every day. The sheer planning and discipline involved, and fortune to not be impacted by delayed or canceled flights, is impressive. He breaks everything down in this great Flyertalk post. There was a 6-month period of my life when I was commuting at least 5 hours a day, every weekday, and I swore I’d never do that again. (Standing in a hot un-air conditioned train for an hour every day is not fun. I once fell asleep while standing up.)
Here’s how one of his days looked:
“8/11/2022 Thursday LAX-SFO-LAX AS3342 LAX-SFO E175 $5.60+5000 AS miles AS3435 SFO-LAX E175 $5.60+5000 AS miles Leave home 410am Car Parked 436am Reach Airport 456am Flight Departs 600am Flight Arrives 745am BART Departs 935am BART Arrives 1047am Arrive Class 1100am [4 hours 20 minutes at class] Leave Class 320pm BART Departs 333pm BART Arrives 448pm Flight Departs 623pm Flight Arrives 757pm Airport Landside 804pm Reach Parking Lot 822pm Arrive Home 858pm
That’s when I started questioning myself: 6am flight everyday, is it really worth it? Especially given that I only slept 4.5hrs the night before. But anyways, once I boarded, I slept like a baby until we reached SFO.”
— Aakash Gupta 🚀 Product Growth Guy (@aakashg0) May 18, 2023
✅GOOD MANAGERS ask open-ended questions in 1-1s like “What’s on your mind?” and “What could be going better?” They look for what's not being said and try to help.
❌BAD MANAGERS worry about unimportant details or company politics and forget the big picture in 1-1s.
I’m somewhat embarrassed to have only discovered the amazing library system in our part of the Bay Area. There is a small offshoot branch of the main city library about 2 minutes’ drive from our house that has a great selection of new releases and children’s books. And, through that library, I can request books online from any of the dozen or so other libraries in the same system. They ship the book to my local branch and notify me when it’s ready for pickup. They seem to get all the major new releases. Self-checkout is simple — they have a neat scanning machine where you pile all the books you’re checking out on top of the checkout surface. The machine detects all of them and deactivates the security tags in one fell swoop. The books are initially lent for 3 weeks, but as long as no one has a hold placed on them, the library will automatically renew them for you. They let you check out 50 books at a time! I visit with my daughter every second Saturday.
Anyway, the library is not what this post is actually about. While at the library, I picked up a field guide to identifying mushrooms (on a total whim) and learned a few things, including what a “spore print” was. Basically, it’s a method of identifying mushrooms by observing the attributes of the spores they drop, such as their color and drop pattern.
I tried to make one. It’s easy: I bought a portabella mushroom from the supermarket, cut the stem, placed it gills down on a piece of paper and covered it overnight with a glass container. The next day, you get your spore print on the paper. (And if you spray it with a fixative, like hair spray, you have something you can hang on the wall.) Below was the result:
I also learned that portabella mushrooms are the same species of mushrooms as white mushrooms, button mushrooms, and cremini (brown) mushrooms — all supermarket staples.
Mushrooms are just one part of a fungus. Similar to an apple on an apple tree, mushrooms are the fruiting body for the underlying mycelium (a threadlike network that exists in soil or a substrate like wood) that makes up the rest of the fungus. Mushrooms are the way fungi reproduce, since they are basically receptacles for dispersing spores.
Mushrooms are incredibly diverse, but I was particularly intrigued by giant puffballs. They can grow larger than a soccer ball and, for a limited span of their lifecycle, are edible!
Further Observations
In a civil case, Trump was found liable for sexual abuse and defamation. He is appealing. I doubt this will significantly change whatever trajectory he’s currently on. Once DeSantis enters the race, things will get interesting.
In another case of chickens appearing to come home to roost, habitual liar and member of Congress George Santos has been indicted on 13 federal counts including wire fraud, money laundering, theft of public funds, and making false statements to Congress. He has pled not guilty.
Beef (Season 1) Sucked me in deeper and deeperwith each episode. The main characters are mostly Asian and about my age, which in a strange way (because things get really crazy) made this show more relatable. Worth a watch. ★★★★
Air I’m a sucker for Damon/Affleck flicks. The 80s montages were also spot on. ★★★
One of our kids got the whole family sick this week, so just a short update.
Following the demise of First Republic Bank, all regional bank stocks took a bath on Thursday. They rallied slightly on Friday, but the signal sent by plummeting stock prices could affect confidence and lead to a run on a bank by its depositors. If this continues, it’s hard to see where it stops with regional banks unless the government steps in and changes policy — most likely by doing something with the FDIC’s 250K insurance limit. Pac West and Western Alliance appear to be the most vulnerable at the moment.
King Charles III had his coronation ceremony yesterday. I have been told by our resident royal watcher that historically the UK’s coronation ceremonies were not attended by foreign monarchs because the ceremonies were considered to be the business of the people of the UK alone. Now that the British Royal Family is considered to be one of the UK’s primary “exports,” times have changed… If you are curious who attended, here’s a list.
On Friday evening, we arrived home after dinner to find that one of the large Modesto Ash trees in our front yard had split in two. The split half was being precariously held up by the branch of another tree, and that branch was the only thing preventing everything from collapsing. Nonetheless, the whole tree was now tilting like the Leaning Tower of Pisa, and hanging over the sidewalk like a gigantic Sword of Damocles for people taking their evening walks.
I quickly printed out a couple warning signs and stuck them to the tree. But now what?
We had actually realized the tree was in the process of dying a couple months ago and called an arborist. After tapping the trunk and several branches with a pen, he recommended we take the tree down, but there wasn’t any rush as the winter storm season had passed. We signed a contract with them ($1,500 for the removal, plus more for stump grinding) a couple weeks ago, but then they strangely ghosted us, despite multiple follow ups. A call to their emergency line went unanswered.
In any event, after Googling, it turns out that even when a tree of this size is clearly dead, you need a tree permit from the city to remove it. Tree permits take about 10 days to process, unless it’s an emergency. Two problems: the permitting office was closed since the weekend had basically started, and the tree permit required an arborist’s report for submission.
We didn’t feel comfortable waiting until Monday, and it was a toss up who would collapse first over the weekend: First Republic Bank, the Warriors, or our tree.
Susanne got some quotes from other arborists through Yelp. We were surprised to get several immediate responses. Some were willing to come out on Saturday to give us a quote. One asked for $2,000 off the bat (ouch).
I called the city works department, which was closed, but a recorded message pointed me to an emergency line for issues like burst water mains and leaking sewer lines. The person who answered was as helpful as they could be — they didn’t know who would handle our issue, but took down my contact details and promised to get back to me.
Not knowing how long that would take, I then turned to Nextdoor and a few helpful neighbors suggested I call the non-emergency line for the police. (I later read in the city ordinances that the police can authorize a tree being removed if it’s a safety threat.) After calling, we were told someone else had apparently reported the issue already (it wasn’t our earlier call to the works department), and they would be sending someone around to inspect.
About an hour later, we got a knock on the door from a city worker. We took him to the tree. He shone his torch on the trunk, “Hmm, yeah I can see the split.” Then we directed him to look into the canopy where the other tree was supporting its half-fallen compatriot like a drunken sailor. “Oh. Oh yeah. That’s not safe. That’s not safe. I didn’t even see that. This is a tree on your property, but we’re going to call someone out to take some of these branches down because it’s not safe. We can cut it up but we can’t haul the debris away for you.”
Not a problem with us! We had already planned to pay for that, and now the city was telling us they were going to do it for free. We asked how long that would take, expecting it to be some time the next day, but were amazed to learn a crew would be out within the hour. (He was actually apologetic, “Well the guys have got to go and pick up the equipment they need first.”)
Sure enough, a guy with a cherry picker and a chainsaw showed up and spent the next couple of hours cutting down the tree, one chunk at a time. He was done just before midnight and we were super appreciative.
Because bureaucracy is bureaucracy, he told us we still needed to file for an emergency tree permit for the removal. “However, it shouldn’t be a problem because this was an emergency situation and my boss is the person who approves those permits.”
Further Observations
First Republic Bank is going the way of Silicon Valley Bank. A lot of the VCs that were extremely vocal on Twitter about saving Silicon Valley Bank depositors when the bank was melting down have been strangely silent. “The government has to save all the depositors or the regional banking sector is going to die!” But this weekend, not a peep from the Twitterati. Not hard to figure out why.
H-1B visas are the most common form of visa for international students looking to stay in the U.S. to work. If you have a master’s degree, there are only 85,000 available each year. This April there were over 780,000 applications. Yikes. When I applied for my H-1B back in 2010, the silver lining of being in the depths of the great financial crisis was that there were more than enough H-1Bs to go around.
If you think you’re having a hard week, just be glad you’re not doing this:
On Twitter
In case you’re wondering who‘s arguing with you on the internet… here’s a video of people walking into a building and being “scanned” with a lint roller by a man who is not a security guard, no questions asked. pic.twitter.com/lLL8vDhMRt
Joe Biden asked the president of South Korea to sing the song American Pie when he found out that Yoon Suk Yeol is fond of American music. And he unexpectedly agreed. pic.twitter.com/6PVTteSAVV
This week was the series finale of Star Trek: Picard. I’ve been a Trekkie for a long time and I have blog entries dating back from as long as 25 years ago (!!) writing about Trek. Given the special nature of the season, it felt appropriate to write a few more words about it. Spoilers ahead.
The last time we saw the whole TNG cast together was in Star Trek: Nemesis. Unlike the original series cast’s swan song in Star Trek VI, the TNG cast never had a proper send off, as poor results at the box office for the decidedly sub-standard Nemesis saw their time on the screen come to an abrupt end.
Now, a little over 20 years later, the whole senior TNG cast — naturally playing characters that are also 20+ years older — has been reunited in a fitting end and satisfying closure to the TNG crew’s journeys together.
The previous two seasons of Picard were forgettable and the writers of Season 3 did their best to do so. Season 3 was much better.
I found Season 3 deeply nostalgic, and the writers made sure to throw in tons of callbacks and easter eggs to stoke that nostalgia. So much fan service, and I loved it. It would have been nice to see more cameos from DS9 and Voyager folks (apparently they ran out of budget), but they did include an oblique reference to Odo, who was played by the late Rene Auberjonois.
When you see someone continually over the course of years, you don’t notice the aging process so much. But the last time we saw the crew, they were much younger. So suddenly seeing them all together again on the screen, 20 years later, was a stark and constant reminder of the passage of time. For me, it brought back memories of the times I spent watching and discussing episodes of TNG and DS9 with a couple of my best friends during high school and university (screening at 11pm on Channel 9 during a weeknight). Occasionally we would scoop up whatever VHS tapes we could get our hands on at the local Video Ezy and watch episodes late into the night. And it was also a sobering reminder that life has moved on — with the onset of middle age, we now each have our own “next generation” to bring into the world.
The closing scenes were great — Picard delivering lines from Julius Caesar and a poker game to mirror the end of All Good Things… Apparently they filmed the cast casually playing poker for about 45 minutes, took an excerpt from it, and rolled it over the closing credits. (Oh, and Q is back.)
I was asked for my thoughts on TikTok. My thoughts are pretty simple.
The potential for social media to do harm in various forms has been recognized for many years now. Ad-driven social media business models rely on engagement and attention, and their algorithms optimize for that. That often means displaying content that is provocative or even incendiary. It leads users down rabbit holes and into echo chambers. It influences emotions and perspectives and messes with adults psychologically. For kids, it’s even worse. I believe it can really mess up kids during their most formative years — especially if they are left with it for hours unsupervised, each day. (Gen Z believes it too. See “Do the Kids Think They’re Alright?” in the Articles section below.)
TikTok has incredible engagement and reach. People spend hours on it each day. It has over 100 million users in the U.S. As a business, I wish I could invest in it. As a product, I’m not a user. But some of the content on there and the creativity on display is pretty incredible.
Overlaid on top of all of this is the fact that TikTok is owned by ByteDance, a $300B Chinese company. Chinese companies (and particularly tech companies, as of late) are vulnerable to interference by the Chinese government far beyond a level that exists in the U.S. The Chinese government even has a board seat and some ownership of a key ByteDance company. (ByteDance also has several other apps that rank very well in U.S. app stores.)
Facebook has been used by domestic and foreign actors to spread misinformation, influence hearts and minds, and generally damage the fabric of society.
It then doesn’t take a lot of imagination to realize that if the company responsible for the algorithm itself were trying to set an agenda (in this case, at the behest of the CCP), it could do so by directly altering the algorithm in a way that could have much more significant impact than by external actors trying to game the system. It’d be a good way to subtly spread propaganda or stoke social discord in a mostly inscrutible way.
Data surveillance is also an issue, but that doesn’t seem as significant. The U.S. government certainly has the ability to covertly compel production of certain data from private companies. You’d also think someone would notice if the apps were trojan horses for spying on whatever else is happening on a user’s device.
TikTok is banned within China itself, which is pretty telling.
The result is a national security threat that seems very real. It’s hard to see another outcome from this situation other than for TikTok to be spun, or banned. I won’t be sad if either of those things happen.
Some may argue that banning TikTok would be hypocritical of the U.S. The U.S. regularly decries China for muscling out U.S. tech companies from the Chinese market — only to threaten to turn around and do the same thing. This is a whataboutism-type argument which doesn’t respond to, nor undermine, the genuine concerns above. Nor does it undermine the notions of sovereignty and what’s in the national interest that allow countries to engage in this behavior. It’s not ideal, but there’s no reciprocity here.
Further Observations
I found out what happens when you forget to remove crayons from the pockets of your pants before you throw them in the wash. It looks like my toddler drew all over our clothes. Bits of melted wax everywhere. The internet’s solution of washing the clothes in very hot water, dish detergent, vinegar and washing detergent was not completely effective. ChatGPT was unhelpful. Wife not amused.
And while I was at it, my wallet also took a bath in the washing machine.
We’re almost at the end of Season 3 of Picard and not only is it far superior to the other two seasons, they dialed the nostalgia up to 11, and they did it pretty tastefully. This week’s episode is the penultimate one for the series, and it was amazing. TNG was an indelible part of my adolescence, and it’s kind of wild seeing the entire cast re-assembled, 30 years later.
Do the Kids Think They’re Alright? (After Babel) “It’s hard to find members of Gen Z who think their phone-based childhoods benefitted their generation.”
The kids picked up something nasty at pre-school and it took the whole family down this weekend. So, only a short update for this week, but a long and interesting articles list below.
Google Flights is my tool of choice for finding paid airplane tickets. It’s quick, flexible, and links off to other sites to complete the booking (I greatly prefer booking directly with airlines rather than third parties like Booking.com, Dreamz, etc., which can be a huge pain to deal with if something goes wrong.) Google flights now offers a limited guarantee where they will reimburse the difference if the fare drops after you book. There are limitations, of course:
You need to be signed in your Google account and use U.S. details when you make your booking (currency, phone, address).
The booking needs to be no more than 60 days in the future
It’s only offered for some itineraries (denoted by a badge).
$500 maximum reimbursement, max 3 times a year.
Apparently an unofficial slide that an associate at Paul Hastings prepared has been making the rounds:
While Paul Hastings was quick to disclaim this as the firm’s official position, about 70% of this is accurate, I think. Even for someone who routinely spends these dollars on outside counsel, #3 is not a reasonable expectation (except, perhaps, during crunch time of a major transaction) and firms should be able to manage staffing to give their staff time off and appropriate coverage. #4 I don’t expect everything to be done yesterday. It’s not always possible, I try to give reasonable timeframes and don’t like to put outside counsel through needless fire drills. #7 “No poor connections” — we’re all at the mercy of Comcast. #9 is totally an acceptable answer for an associate, as long as you follow it up with, “but I’ll find out”. #5 is an important point. If you pay $1,000+ per hour to anyone for anything in life, you’re going to expect gold-plated service.
One of the most famous memos in all of Microsoft history was a first memo about cost-cutting which came about in 1993 (see early 90s recession). Mike Murray, leading HR, wrote the memo "Shrimp and Weenies"—before MS, Mike was at Apple, having led OG Mac marketing. 1/ (third try)
PEOPLE OF THE EARTH. Figured out how to ask natural language questions of current financial data and get back a natural language response…with correct/relevant numbers! 🤯💰✨
So. Freaking. Cool.
1. Send database schema along with the question and relevant scope/parameters… pic.twitter.com/AE49RBTIrH
I have been mulling writing a lengthy post about AI, but there is so much happening, and it’s happening very quickly, so I’m still processing it all. So instead, here is an assortment of initial thoughts. I’m sure there will be more in the future.
The New New Thing. In my lifetime, foundational developments in information technology have happened every decade or so: the personal computer, the internet, and the rise of mobile. All of these developments changed the world. (You’ll notice that absent from my list is crypto and the metaverse.) Generative AI now joins that list and I think not only is it as significant as the others, it’s on track to be in a league of its own.
Speed. The pace at which the technology is developing, and the practical applications and industry springing up to take advantage of it is crazy fast.
The most prominent player in the GAI space is OpenAI. GPT-3 was originally released in mid-2020, but it wasn’t until OpenAI released ChatGPT, a chat bot based on the GPT-3 large language model (GPT-3.5 to be precise), that the technology sprang into the public consciousness. ChatGPT received a mind-blowing 100 million sign ups in 2 days in November 2022. OpenAI released GPT-4 last month, and upgraded ChatGPT to use it (available as a paid subscription to ChatGPT Plus). GPT-4 is significantly better than GPT-3, and can handle both text and image inputs. GPT-5 is rumored to be released by the end of the year.
The GPT models are static, relying on the information they were trained on at the time. As such, they are more or less frozen in time. ChatGPT plugins allow ChatGPT to access external information sources, and perform actions against external systems. This multiplies the power of ChatGPT. The fascinating thing is that you can use ChatGPT to help integrate itself with other systems.
There are tons of companies, products, and use cases that have been created just this year that leverage GPT as well as other AI models that generate images, audio, video, and code.
How It Works. Read Stephen Wolfram’s article for a detailed description of the concepts behind text-generative AI. Even if you don’t understand all of it (I certainly didn’t), you’ll still get a good idea of how it conceptually works even if you have to give up halfway through.
Use Cases. There are far too many use cases and examples of amazing things generative AI can do to list. Here are a smattering of interesting ones I have been thinking about lately:
Mimic someone’s voice just by providing an AI model with 30-60 seconds of audio. The output comes complete with accents, foreign languages, and vocal tics.
Out of all the major voice assistants (Google, Alexa, Siri), Google is the most capable. But it’s no ChatGPT. Once you plug AI into them, you’re going to see supercharged voice assistants capable of conducting conversations. It’s not only the quality of their answers that will improve, or what actions they will be able to perform, but AI will improve the quality of voice recognition too. And if you want your voice assistant to sound like anyone on the planet, it’ll probably be possible to do that one day (see “mimic someone’s voice” above).
Within literally 10-15 minutes, I was able to create a working script for monitoring the prices of securities (both by scraping stock prices from a webpage and crypto prices from the Coinbase API), recording prices to a database, and emailing price alerts, just by asking ChatGPT about 5 questions (some debugging was still required).
In 2017, I spent a week writing code that helped me arbitrage BTC and ETH between a half dozen exchanges. I estimate I could have completed that in maybe a third to half of that time if I had help from AI.
Technical analysis and algorithmic trading.
Legal document analysis and review.
Existential Risk. AI has issues. A ton of issues. The most splashy one is existential risk. The basic idea is that AI at some point will be able to develop itself, leading to a runaway super-intelligence that outstrips our ability to control it (a technological singularity). The human race becomes casualties, whether through some concept of malevolence or just collateral damage as a result of “optimization”. It doesn’t necessarily require an AI that is “self-aware” — it may just be the rampant execution of internal logic that results in major unintended consequences.
An open letter signed by Elon Musk, Steve Wozniak, and a host of luminaries in the AI world calls for a six month pause in the training of AI systems more powerful than GPT-4. They write:
Contemporary AI systems are now becoming human-competitive at general tasks, and we must ask ourselves: Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization? Such decisions must not be delegated to unelected tech leaders. Powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable.
Many researchers steeped in these issues, including myself, expect that the most likely result of building a superhumanly smart AI, under anything remotely like the current circumstances, is that literally everyone on Earth will die. Not as in “maybe possibly some remote chance,” but as in “that is the obvious thing that would happen.” It’s not that you can’t, in principle, survive creating something much smarter than you; it’s that it would require precision and preparation and new scientific insights, and probably not having AI systems composed of giant inscrutable arrays of fractional numbers.
Without that precision and preparation, the most likely outcome is AI that does not do what we want, and does not care for us nor for sentient life in general. That kind of caring is something that could in principle be imbued into an AI but we are not ready and do not currently know how.
Absent that caring, we get “the AI does not love you, nor does it hate you, and you are made of atoms it can use for something else.”
Of course, there are those that plain disagree, saying that AI is nowhere near the stage which we should worry about (e.g. we’re still far away from building an artificial general intelligence), or that the worry-warts have ulterior motives (they want the additional time to catch up to Open AI), or that in practice there’s no way to halt AI development, so why bother?
The latter school of thought is emblematic of today’s world in which lack of expertise in a field isn’t a barrier to expressing a strong opinion. Strong opinions are provocative, and that is a means of getting attention. In reality, we just don’t know what’s going to happen with any certainty. Even the creators of generative AI models can’t predict what their own creations will do as they develop. But when the potential risks are astronomical, then even a far-fetched and fanciful probability that they may come to light means those risks deserve thoughtful introspection, humility, and caution.
Given the potential for wide-ranging harm that generative AI has, I do think we need to get some tentative regulations and protocols in place, or at least start a robust discussion about it between all the major players. Had we known that social media would have the side effects it does today (misinformation, teen depression, etc.), perhaps some things may have turned out differently?
The idea of the Great Filter is one theory why we haven’t found intelligent alien life yet. The theory goes that there is some universal, insurmountable step that prevents civilizations from expanding into the galaxy. If it exists, we cannot know whether we have already passed that step, or whether it lies ahead of us. Nuclear war and climate change are traditional candidates for a future Great Filter that are in play. Runaway AI is another one that has been theoretical to date, but it feels more real today than ever.
More Issues.
Dystopia. Even if AI turns out not to be an existential threat, I don’t think anyone disputes that it will eventually cause massive societal change. The short story Manna presents a dystopic and utopic view of AI. The dystopic threat doesn’t look like Terminator robots, or a Skynet-type overlord, but simply a bunch of machines optimizing situations without any regard for humanity. And consider the court system. Courts are the final avenue of appeal for disputes, including those that result from automated decision making (e.g. your insurance claim got denied). Now, it doesn’t take a big leap of imagination to cut out human judges and start having computers render judgments in smaller forums, like small claims courts. After all, computers have the ability to synthesize the entire history of jurisprudence to figure out a ruling. And then, keep sliding down that slippery slope. (There’s a reason why the GDPR requires human intervention to be available where an individual disputes an automated decision based on their personal data.)
Bias & Misinformation. Generative AI has issues with bias — including political and racial — and being confidently incorrect. Therefore, there are real concerns about models having the ability (or some would say being intentionally imbued with the ability) to persuade and misinform, and the ability to help with misinformation campaigns. Additionally, Twitter believes that the challenge of detecting bots is much harder with AI in play, since sorting out who’s human and who’s not just got more difficult. In response, they are moving focusing on paid subscriptions as a way to help verify identity.
Criminal Enterprises. This one is going to be a real big problem. Think of all the scams you can improve with AI. Related to deepfakes.
Legal Issues. There’s a lot here.
The biggest one relates to copyright infringement. AI models are trained on a huge amount of data, and almost all of this data is owned by someone else. Is use without permission infringing, or is there a fair use argument?
AI makes mistakes, and so people may get into various sorts of trouble if they blindly rely on what AI says. You can put all the disclaimers you want, but people aren’t going to be bothered to double check what the computer says a lot of the time.
Information ingested by AI models also raises issues relating to compromising the confidentiality of any sensitive information obtained. The information itself isn’t a discrete component of a model, so it’s hard to extract.
The Chinese Government. Using AI models requires a great deal of computing power. Queries are therefore sent off devices for processing in the cloud, but it’s foreseeable that devices will be able to do that processing on-device at some point. There’s a hypothesis that Apple, with its impressive homegrown chips, is maneuvering into a position to be able to do this (and it’s also in line with its emphasis on privacy). It’s tougher to censor what is happening on a device, so this represents a challenge for the Chinese Government’s censorship efforts. I imagine this will lead to the technology and any hardware that runs it being heavily regulated, or even banned. It is of course possible to build models that return answers that toe the party line (OpenAI has been accused of injecting liberal bias into their models, and they also censor results that are deemed harmful), but AI is tougher to directlycontrol because it’s a bit of a black box. Travel blogger Gary Leff writes: “Perhaps AI and Large Language Models represent the next best hope for an end to repression in China. If China wants to compete in this space, they’ll have difficulty doing it behind the Great Firewall. Their tools won’t be as strong as the ones from the West, with access to more knowledge to train on. Is there a stable equilibrium where their AIs can train on unrestricted content, but answers from the AI remain restricted? What about when everyone has AI chatbots on their phones, rather than central servers?” I’m unfortunately not optimistic about this, as in earlier times people thought the internet couldn’t be tamed, but the Chinese Government has done a pretty darned good job of it.
Miscellaneous Stuff.
It has been said that the internet makes information globally available, so gaining an advantage is less about who has access to information, but who can analyze it the fastest and execute on the actions needed to capitalize on it. Now, with AI helping to analyze information, and potentially being able to act on it, where does that leave humans? A lot of valuable content will now consist of real-time reporting and opinion pieces, since research and summarization is something that AI is pretty efficient at.
I’ve also tried Google Bard and Bing Chat, but ChatGPT is the best (and it can write code).
ChatGPT uses a “prove you are human” Captcha, which I thought was ironic. Chaptcha’s days are numbered.
When I joined a law firm after graduating from law school, a second year solicitor was assigned to show me the ropes (we both reported to the same partner and were on the team that did work for Microsoft). That person is now the Chief Responsible AI Officer for Microsoft. I haven’t spoken to her in the 15 years since I left the firm, but that job must really be something!
Further Observations
Trump got indicted on over 30 counts on Thursday. The counts are supposed to become public next week. Interesting times ahead.
The CTFC is going after Binance. Binance’s response is a bit lame, and there’s a lot of stuff in the CFTC complaintthat really doesn’t look good for Binance. One example: “Binance’s senior management, including Zhao, knew the Binance VPN guide was used to teach U.S. customers to circumvent Binance’s IP address-based compliance controls. In a March 2019 chat, Lim explained to his colleagues that ‘CZ wants people to have a way to know how to vpn to use [a Binance functionality] … it’s a biz decision.’ … in a July 8, 2019, conversation regarding customers that ought to have been ‘restricted’ from accessing the Binance platform, Lim explained to a subordinate: ‘they can use vpn but we are not supposed to tell them that … it cannot come from us … but we can always inform our friends/third parties to post (not under the umbrella of Binance) hahah.’”
Inside the Dying Days of Tibet’s Caterpillar Fungus ‘Gold Rush’ (Huffington Post) My mum used to regularly make us soup from cordyceps. I said it looked like a caterpillar. They said no, it was a fungus, and I was cool with that. I guess they were technically correct, and knowing it was actually a freaking fungus-infested caterpillar wouldn’t have made things any better.
Court transcript from Georgia v. Allen (Superior Court of Floyd County, Georgia) The court transcript to end all court transcripts. Here is but one excerpt: – Mr. Allen: Take off the cuffs. – The Court: How many hands do you have to have to do it. – Mr. Allen: Take off the cuffs. – The Court: Come on—no. Jack off. – Mr. Allen: This is kangaroo court. – The Court: Come on. Jack off. – Mr. Allen: This is—this is a kangaroo court. – The Court: Jack off right now.
Movies & TV
John Wick: Chapter 4 You know what you’re getting, and there’s 2 hours and 49 minutes of it. And Keanu only speaks about 360 words in the whole movie. I was thoroughly entertained.
The Fed raised interest rates by 0.25% on Wednesday to a range of 4.75-5.00%. The upper range of that bound is now 20 times what it was a mere 12 months ago. Rapidly rising rates were always going to put pressure on various sectors of the economy that have benefited from cheap money for the last decade plus (that’s kind of the point). We’ve seen it with startups who scored cheap equity financing a couple years ago that are now having trouble raising up rounds. We’ve seen it with families no longer being able to afford mortgages that are as large (with monthly interest payments almost tripling), resulting in a housing market slowdown. We’ve seen it with the regional bank runs that declines in the values of banks’ bond portfolios have sparked.
The question is now which sector is going to be the domino that tips the economy over the edge into a recession? Commercial real estate seems to have arisen as the most likely candidate, with falling tenancy rates combining with a lot of maturing loans that are coming up for refinancing at today’s much higher rates appearing as storm clouds on the horizon.
We had our third prolonged power outage this year. Again on a Tuesday. An extra-tropical cyclone parked itself just off the coast of San Francisco causing more fallen trees and localized flooding. Power was out for us for about 30 hours, and we ended up spending yet another night in our hotel. We now also have a great view into our neighbor’s back yard because most of our fence is gone. And next Tuesday the forecast is for another period of “excessive rainfall with localized flooding”. Sigh.
Last week, I wrote about the fallout from a protest incident at Stanford Law School. This week, SLS Dean Martinez penned a detailed explanation of why she did what she did. It’s a well written missive, and I agree with almost all of it. However, I had two observations. The first is the practical impact of deciding not to sanction anyone for their behavior, ostensibly because it was too hard to figure out who was breaking policy and who was not. Everyone is instead going to have to sit through a mandatory training. Will this mean that a future protest tactic is to flood a talk with bodies because it’s just too hard to identify who was doing what? The second is that the letter only mentions in passing something about respectful discourse. The letter also expressly notes that asking vulgar and provocative questions when you’re given the microphone is ok and not disruptive. That may be true from the point of view of free speech, but as lawyers, respectful argument is part of the job and—normally—a positive value to encourage. (Try spewing profanities at a judge in court and see how far that gets you.) One protester allegedly called for the judge’s daughters to be raped.
“Our world needs climate action on all fronts—everything, everywhere, all at once,” said UN Secretary-General António Guterres in a speech on Monday. He was referencing an IPCC report which once again is warning we’re nearing the point at which it will not be possible to avoid a rise in the global average temperature by the end of the century of a magnitude that will cause severe human misery. “Unless nations adopt new environmental policies — and follow through on the ones already in place — global average temperatures could warm by 3.2 degrees Celsius (5.8 degrees Fahrenheit) by the end of the century, the synthesis report says. In that scenario, a child born today would live to see several feet of sea level rise, the extinction of hundreds of species and the migration of millions of people from places where they can no longer survive.” It’s going to be a slow burn, but wild weather is going to be increasingly common, and difficult to adapt to in our lifetimes, regardless of where you live in the world.
For those who have the fortune of being able to decide where they want to live and work in the world, I think climate factors are going to rise to one of the top considerations within the next couple of decades. Not just whether the weather is “nice”, but whether that cliffside house or oceanfront view is such a good idea, or whether 100 year flood zones really are 10 year flood zones, or whether water scarcity might be a real problem.
Deal Alert: We use Doordash a lot. If you live near a Lucky Supermarket, they have a promotion right now where you can buy six $50 Doordash gift cards and you’ll get 6000 rewards points, which is good for $66 of store credit (and which can be applied to buy more gift cards). Buy with an Amex Gold card and you’ll also get 1200 Membership Rewards points.
How can AT1 bonds be zero'd when they're more senior than equity which is getting paid 2bn, you ask?
Because bondholders agreed they could be.
Here's the section in the $CS 9.75% AT1 prospectus (don't worry if you weren't aware, half the holders prob haven't read this either): pic.twitter.com/Q5FbnOMRM4
Eyes on Tuesday for (again) excessive rainfall with localized flooding issues per @NWSWPC A rapidly developing surface low off the NorCal coast will also bring another round of gusty wind to the region as well. Best to use the relatively quiet weather through Sunday to prepare… pic.twitter.com/al2GNcJCEb
On March 9, Stanford Law students heckled and shouted down a 5th Circuit Judge who was giving a talk hosted by the school’s student chapter of the Federalist Society. Among the grievances of the heavily progressive student body was that the judge had refused to use, in a 2020 opinion, the preferred pronouns of a transgender person who had been convicted of several offenses related to child pornography. (“In conjunction with his appeal, Varner also moves that he be addressed with female pronouns. We will deny that motion.”)
The judge was reportedly unable to complete his prepared remarks, and during question time was subjected to invective, such as the question, “I fuck men, I can find the prostate. Why can’t you find the clit?”
Apparently, when one of the school’s diversity deans arrived to “restore order,” she ended up criticizing the judge, while other administrators failed to tell protesting students to allow the judge to speak without being interrupted.
Student members of the Federalist Society were further subjected to a name and shame campaign, and allegedly “encircled” and abused at the event after federal marshals escorted the judge away.
The judge later remarked, “Don’t feel sorry for me. I’m a life-tenured federal judge. What outrages me is that these kids are being treated like dog shit by fellow students and administrators.”
On March 11, the Dean of the law school, Jenny Martinez, and the University President, Marc Tessier-Lavigne, sent a written apology to the judge:
We are very clear with our students that, given our commitment to free expression, if there are speakers they disagree with, they are welcome to exercise their right to protest but not to disrupt the proceedings. Our disruption policy states that students are not allowed to “prevent the effective carrying out” of a “public event” whether by heckling or other forms of interruption.
In addition, staff members who should have enforced university policies failed to do so, and instead intervened in inappropriate ways that are not aligned with the university’s commitment to free speech.
Dean Martinez followed that up with an email to alumni clarifying the law school’s stance on free speech.
In response to the apology, “hundreds of student protestors wearing masks and all-black clothing lined the hallways outside Stanford Law School Dean Jenny Martinez’s classroom” where she was teaching a con law class. She “arrived to find her whiteboard covered in fliers ridiculing Duncan and defending those who disrupted his speech. The fliers echoed the opinion of student activists and some administrators who claimed hecklers derailing Duncan’s talk was a form of free speech.”
“They gave us weird looks if we didn’t wear black” and join the crowd, first-year law student Luke Schumacher said. “It didn’t feel like the inclusive, belonging atmosphere that the DEI office claims to be creating.”
As a law school alumnus, I found this behavior incredibly embarrassing.
You only have to flip the roles to highlight how ridiculous it is. What if it were a left-leaning circuit judge being yelled at and insulted by a motivated group of right-leaning students? Would the same progressive students who yelled down Judge Duncan support the right’s right to do so then? I doubt it.
I am not familiar with the judge’s jurisprudence, but to be clear, with respect to the viewpoints of his that were reported by the media, I do not agree with them. But that should not matter here.
I can’t think of any other country which has a more expansive right to free speech than the U.S., and it is a fundamental enough right that it is enshrined in the constitution. However, if you seek to wield that right, you wield a double-edged sword. The same right that allows anyone to speak out on political topics without fear of government prosecution, allows other people to picket (at a distance) the funerals of gay murder victims with hate speech. But that is kind of the point. As a result, censorship is something that runs contrary to free speech values, and not letting someone talk by shouting over them en masse is a form of censorship. And, at some point, that kind of disruption can cross the line into unlawful speech—harassment, threats, slander, and the like.
Stanford is a private university and is not legally required to uphold free speech values. However, these values share much in common with the notion of academic freedom, and so it is unsurprising when Dean Martinez writes, “Freedom of speech is a bedrock principle for our community at SLS, the university, and our democratic society. … The way the event with Judge Duncan unfolded was not aligned with our institutional commitment to freedom of speech.”
It’s not clear in the reporting whether the protesters were law school students, or students from other parts of the university (events like these are normally open to all to attend). If they are law school students, what are they going to do when they become lawyers and get in front of a judge they disagree vehemently with, but still need to present a case to?
(Sidenote: Interestingly, unlike the U.S., a few countries like England and Australia, practice the “cab-rank rule” which obliges barristers to accept work from any client as long as they are competent enough to handle it, and regardless of any personal distaste the barrister may have for their client’s reputation, character, etc. “Without the cab-rank rule, an unpopular person might not get legal representation; barristers who acted for them might be criticized for doing so.” So in these countries, you not only need to have the ability to present a case respectfully in front of a judge you may personally hate, but you may also need to do it on behalf of a client you find repugnant. Not easy but, in my opinion, an important part of the justice system.)
Further Observations
Last week’s newsletter about SVB produced the highest number of views out of all my past newsletters. The drama continued this week, with regional banks under pressure. One of the most notable among these banks is First Republic, a San Francisco-based bank that serves a lot of high net worth individuals. First Republic has been experiencing an outflow of deposits which has led to efforts to shore up its balance sheet. In addition to obtaining a $70 billion credit line, a consortium of large banks deposited $30 billion on Thursday. First Republic was reportedly also looking for an acquiror. However, those efforts failed to calm the markets, and the stock closed down for the week, reflecting skepticism that First Republic will be able to weather the storm without going into receivership.
On the other side of the Atlantic, UBS has agreed to buy Credit Suisse for about $3.2 billion in an all stock deal which is expected to close by the end of this year. That’s 0.50 Swiss francs per share, which is about a quarter of CS’ stock price at market close on Friday.
Everything Everywhere All at Once won seven Oscars last Sunday. Unusually, EEAAO took out most of the top shelf awards, including Best Picture, Best Director, and three acting awards, despite being a sci-fi flick. Also unusual was that two Asians from the movie landed two of the acting awards: Michelle Yeoh won Best Actress in a Leading Role, and Ke Huy Quan won Best Actor in a Supporting Role. (Jamie Lee Curtis landed the award for Best Actress in a Supporting Role.) Yeoh is the first Asian winner of her award in Oscars history. Harrison Ford presented the awards, which was a nice touch seeing that Ke Huy Quan last appeared on screen with him in Temple of Doom as Short Round. EEAAO is an absurdist film that I found difficult to follow at times, but it’s quite a spectacle with some memorable scenes, including one featuring butt plugs which people are doing their best to sit on.
Strong winds caused another power outage at our home on Tuesday. It lasted 12 hours for us, but 48 hours for our kids’ preschool. Due to the recent rains saturating the ground, a lot of large trees were uprooted, leading to widespread blackouts throughout the Bay Area. We had to spend a night at a hotel again.
In other news, Tesla has finally started selling its Powerwall on a standalone basis again. Due to supply constraints over the last couple of years, Tesla would only sell Powerwalls bundled with solar panels. The recent blackouts have pushed us towards buying a pair.
Never Split the Difference (Chris Voss) A very interesting book written by a former hostage negotiator for the FBI. It highlights the shortcomings of a more traditional “principled” approach to negotiation, where rationality rules the day, and focuses more on the emotional and human aspects of negotiation. Voss asserts that compromising (”splitting the difference”) is a cheap way out that often leads to sub-optimal results for both parties. Sometimes there’s a way to get all of what you want, even when it appears you have little or no leverage, and without blowing up the relationship.
Hotels
Residence Inn by Marriott San Mateo – San Francisco Airport (San Mateo) Unfortunately, the nearer (and nicer) Residence Inn we stayed at during the last power outage was all booked out. This hotel is older, but still pretty well equipped and pretty good to work from. We got upgraded to a two bedroom, two bathroom suite split over two levels.
Charts, Images & Videos
On Twitter
Danger! Workers are doing too well! Call the Fed to shut that down! A conversation with former Treasury Secretary Larry Summers. Watch the full interview in our latest episode, now streaming on @appletvplus. pic.twitter.com/B6GtO0QLnw
— The Problem With Jon Stewart (@TheProblem) March 17, 2023
You guys know what happened to the stock market the last two times this happened? pic.twitter.com/j0GvICdBB6
1/10 The USG response has likely stemmed contagion. However, $SIVB saga has effectively forced the mkt to price the entire Fed Funds rate into the funding cost of the regional banking system. An acceleration in lending contraction is here.
Silicon Valley Bank is, relatively speaking, just down the road from me. As one the top 20 largest banks in the U.S., it is the banker to half of the startup industry but, fortuitously, not to my employer.
At the start of this week, SVB held about $170 billion of deposits from its customers. On Thursday, it fell victim to a bank run. SVB’s customers, within the space of about 24 hours, withdrew $40 billion—a quarter of its deposit book. This is a tremendous amount, and SVB did not have enough cash on hand. In fact, at the end of the day, it had a cash shortfall of almost $1 billion dollars. When a company is unable to pay its debts as and when they fall due (and on Thursday, $40 billion suddenly fell due), the company is considered insolvent and cannot continue business as usual.
On Friday, SVB was placed into receivership. A federal government agency called the Federal Deposit Insurance Corporation (FDIC), took over management of SVB and immediately closed it for business. Anyone still with money in SVB was now unable to get it out. “Anyone” turns out to be mostly startups, who are now facing an uncertain and very stressful few days.
But how did we end up here? Why did customers suddenly want to pull out $40 billion? And if took in $160 billion in deposits, why did they have not enough cash to pay $40 billion back?
What happened?
We don’t normally think about things in this way, but a deposit in a regular bank account is basically a loan to the bank that you can ask the bank to repay at any time (which essentially happens when you go to an ATM and withdraw cash, or ask Venmo to send money from your bank account to a friend). In exchange for lending money to the bank, they pay you interest (normally). Banks use your money to make money. Typically, this is done by lending those deposits to other people, at higher interest rates. So, for example, you might lend the bank $100,000 and they pay you 1% interest, but then someone else might borrow that $100,000 from the bank to buy a house (i.e. a mortgage loan) and pay the bank 3% interest. They pocket the 2% difference.
But the loans that banks make aren’t “on demand”. If you have a 30 year mortgage, you only need to repay a certain amount, plus interest, each month. The bank normally can’t force you to pay more than that. On the other hand, the money you lend to the bank in a regular savings account is “on demand”. This is sometimes referred to as “borrow short and lend long” and is just how banks work.
So, given that timing mismatch, how can the bank lend any money out? After all, it’s no good if you go to the bank one day and want your $100,000 back, only to be told “sorry, we lent your money to someone else and we have to wait 30 years before we get it all back”.
Enter “fractional reserve banking”. In reality, people rarely ask for all their money back at once, and never does everyone ask for all their money back at once, so banks only need to hold back a fraction of their deposits as cash, and can lend the rest of it out (or invest it in other things that are expected to produce a positive return). The fraction that banks need to hold in reserve is fittingly called the reserve requirement, and is set by banking regulations.
There are situations in which unusually high amounts of withdrawals may exhaust a bank’s reserves, but there are usually also facilities available under which banks can borrow money (typically from other banks) on a short-term basis to fill any holes while they scramble to convert their other assets (loans and investments) back to cash.
SVB was in a slightly different situation, but the basic principles are the same. The main difference is that their client base is heavily composed of tech startups. Most tech startups are not profitable or cash flow positive, so they finance their operations by raising equity financing (giving up a piece of the company in exchange for money), rather than debt financing (paying interest in exchange for money). Missing an interest payment on a loan can be deleterious, so when you’re not reliably making money as a startup, debt can be dangerous and is usually avoided. As a result, SVB didn’t have a lot of avenues for lending out the money it had received from its customers to other customers, so it needed to find another place to invest that money to earn a return.
For this, SVB invested a lot in debt in the form of U.S. treasury bonds and mortgage backed securities. When you buy a U.S. treasury bond, you are lending the U.S. government money, and they pay you interest. Treasuries are considered “risk free” in the sense that the U.S. government will always be able to pay you back. They can do this because they can just print money, if they need to. (Let’s leave to one side for now the game of chicken that politicians play every few years with the debt limit.) So from a creditworthiness perspective, treasuries are a very conservative investment vehicle. They are also a very liquid asset, which means there is a deep and active market that lets you buy and sell a lot of bonds quickly.
Because the short-term interest rates have been near zero for so long, SVB decided to invest in $90 billion worth of longer-term bonds that returned a little less than 2% of interest per year, and the full amount of principal after several years.
During the pandemic years, investment in tech startups blossomed, with equity financing pouring into companies at ever increasing valuations. Consequently, because of how concentrated SVB’s client base is in tech startups, their deposits trebled from about $60B at the start of 2020 to almost $200B just a couple years later.
In 2022, after over a decade of near-zero interest rates, rates began rising sharply, to around 5% today. SVB started to find itself having to pay more interest on its deposits, increasing the need for SVB to earn a return on all that money that its customers had lent it. However, most of its money was locked up in those long-term bonds, which was one problem.
The other problem, is that due to increasing rates, the funding environment for startups tightened right up—instead of investing in startups, more people were now investing in treasuries, which were yielding more than they had for over a decade (and remember, treasuries, unlike startup investments, are considered risk free). On the other hand, tech company valuations were getting slashed. This meant that as startups spent their money, no new funding was coming in to replace it. SVB’s deposits started to fall by billions of dollars as startups withdrew money to pay employees and vendors.
SVB needed to start converting some of its investments back to cash so that it could pay those customer withdrawals. Here’s where the problem started.
Most of SVB’s assets were treasuries. If those treasuries are held to maturity, SVB gets all its money back. But if it needs the money now, SVB needs to sell those treasuries now. Unfortunately, if SVB paid $100 for a treasury bond that only pays $2 in interest a year, no one today is going to buy that bond for $100 because the U.S. government is currently issuing bonds that pay $5 in interest. So if SVB sells those bonds, they are going to have to sell them for something less than $100.
This was a problem for SVB, because let’s say it sold some bonds for 95% of what they bought them for. Accounting rules require SVB to consider that its entire bond portfolio is now only worth 95% of its original value. For a $90 billion portfolio, that’s a $4.5 billion decrease. In reality, it’s reported that the bond portfolio had actually lost $15 billion in value. So if they sold some of those bonds, they would have a massive $15 billion hole in their assets that they’d have to plug somehow (remember, that’s $15 billion they no longer have to cover the deposits they’ve taken in). So those bonds were effectively untouchable in the short term.
Instead, SVB sold substantially all of its other investments for cash—$21 billion worth—it also tapped out some other lines of credit it had. As a forced seller, it incurred a $1.8 billion loss on the sale because those investments had declined in value. To plug that hole, SVB tried to do what all of its startup clients do—raise $1.75 billion equity financing. In the press release for the equity raise, the $1.8 billion dollar loss is mentioned in the last paragraph, almost as an afterthought.
People noticed, and this news spooked people. People started to wonder why SVB decided to liquidate $21 billion at a significant loss, and rumors started to fly about whether SVB was in trouble.
Bank Run
Silicon Valley is a pretty interconnected community and news spreads fast. For example, my CEO was plugged into his CEO and VC network and hearing what was going down. Our head of finance was talking to other heads of finance at peer companies. I was reading buzz from a mailing list with hundreds of head of legal on it. Word started circulating that companies were getting their money out of SVB, just in case. It became an echo chamber.
The other thing about having tech startups as clients is that they are tech experts, not finance experts. A large number of tech startups are run by talented under-35 founders who have not experienced a financial crisis in their professional lives, nor are they finance experts. As a result, they logically turned to their VCs for advice on what to do when the rumors started. After all, VCs are “the money guys” and they should know much more about financial management since that’s the industry they effectively operate in. So when one VC says “get your money out,” that causes a whole bunch of their portfolio companies to do just that.
Having seen how fast things unravelled in 2008 (“Bear Stearns is fine!”), my initial reaction to hearing the news was: Get your money out of SVB right fuckingNOW… if you can. As in, drop whatever you’re doing and get those wire instructions in. If it’s nothing, you can always move the money back.
If you didn’t have a second bank account, or you had a loan with SVB that contractually required you to keep your cash with them, things got a bit trickier, and you then had to make calls like whether to move the cash into a founder’s personal bank account, or ignore your loan covenants, and then worry about any legal ramifications afterwards.
The next morning, and $40 billion in withdrawal requests later, SVB was dead. The FDIC announced that SVB was both insolvent and failing its regulatory liquidity requirements.
There has been some blowback against some VCs for fanning the flames of a bank run that might have been avoidable, but I don’t think VCs are to blame here. As a company, you have to look out for your own employees and business and, in this case, taking your money out if you could was the right call. If it’s a false alarm, there’s little downside. But if it’s real, then you don’t want to be stuck in a tough place… especially if it was avoidable. The herd mentality is a powerful driver of financial markets and you want to at least be part of the stampede—not be crushed by it.
What now?
Through the FDIC, the U.S. government provides deposit insurance at all FDIC-insured banks. If a bank collapses, the FDIC will ensure that depositors will be able to get at least $250,000 of their funds back. (This limit used to be $100,000 but was raised after 2008 Great Financial Crisis.) This is very helpful for the average person, who is likely to have less than $250,000 cash lying around in a bank account. It is not so helpful for a company, that is likely to have much more than that. Apparently, about 96% of SVB depositors were not fully covered by FDIC insurance, compared to about 38% at Bank of America.
This insurance kicks in very quickly. On Friday, the FDIC created a new bank called the Deposit Insurance National Bank of Santa Clara (DINB). The insured portion of all SVB bank accounts was then transferred to the new bank, and depositors will have access to those new bank accounts on Monday. The uninsured portion of those bank accounts (anything over $250K per depositor) was then effectively frozen, with the FDIC issuing a “receivership certificate” to each depositor that represents an unsecured claim on the remaining assets of SVB up to the amount of the uninsured funds.
Next, the FDIC’s job will be to sell off SVB in a way that maximizes the money they receive. That money is then distributed to a list of people, starting with secured creditors, then depositors, then other unsecured creditors, then subordinated debt holders, then anything remaining goes to SVB’s stockholders.
As the FDIC sells off SVB, it will issue a dividend to receivership certificate holders. It’s rumored that the FDIC has been hard at work flogging off tens of billions of dollars’ worth of SVB’s assets and will issue an “advance dividend” of about 50% of the value of a certificate sometime in the next week.
The best outcome now is for another bank with a strong enough balance sheet to buy SVB and assume all the deposits in one fell swoop, which will allow accounts to be unfrozen. I think that is a likely outcome. If that doesn’t happen, then SVB will be sold off in pieces, and uninsured depositors will get their money back over time. I think that depositors will get most, if not all, of their money back, but it will take months, or even years.
In the meantime, the timing uncertainty and inability to access funds is incredibly stressful to affected startups. Payroll is due next week. Failure to pay wages is one of the things that can “pierce the corporate veil”, meaning that liability for that failure spills beyond the company, and may impact officers, directors and stockholders personally. However, one would think that if employees get paid a few days late, they’re going to be understanding and not litigious.
This weekend, startups are trying to find sources of immediate-term funding, ranging from loans from VCs, bridge loans, credit cards, and selling their receivership claims at a discount to opportunistic investors.
And teams of bankers, lawyers, and bureaucrats are spending a sleepless weekend trying to figure out what to do with SVB before the markets open on Monday.
I think the majority of SVB startups will be fine at the end of the day. There may be a severe liquidity issue for startups with short runways (e.g. if you had a 12 month runway and only get 50% of your funds back next week, you now have a 6 month runway and no idea when the remainder of the money will come back, and you’re now thinking about whether you need to lay off people to conserve cash).
[UPDATE: It’s over; startups can feel relief. 30 minutes before this post was scheduled to go out: ‘The Federal Reserve, Treasury and Federal Deposit Insurance Corporation announced in a joint statement that “depositors will have access to all of their money starting Monday, March 13. No losses associated with the resolution of Silicon Valley Bank will be borne by the taxpayer.” The agencies also said that they would enact a “similar systemic risk exception for Signature Bank,” which the government disclosed was closed on Sunday by its state chartering authority.’ (per the New York Times) “Separately, the Federal Reserve announced that it was creating a new lending facility for the nation’s banks, designed to buttress them against financial risks caused by Friday’s collapse of SVB.” (per the Washington Post).]
Bail outs?
Several prominent VCs and investors have been screaming for the federal government to come in and backstop deposits in full—not just $250K. In other words, the government should effectively insure everything, and then get reimbursed as FDIC sells off SVB.
This is not a bailout of SVB. SVB is no more—its board and executive team will find themselves looking for new employment (if not already), and the equity holders will likely be wiped out.
But it is a bailout of depositors, and that has other people screaming that the U.S. taxpayer should not have to do that either. Let capitalism take its course. (And, as usual, the whole situation has been politicized.)
The issue is more nuanced than that. As I mentioned above, although depositing money at a bank is essentially lending the bank money (and therefore a form of investment), people don’t really see it like that. And you’re not really thinking that the 16th largest bank in the U.S., which has been around for 40 years and is regulated, is a credit risk. The average person on the street certainly isn’t going to be thinking about that. They’re just looking for a place to put money that isn’t under their mattress. During my time in the U.S., I have had personal bank accounts at a lot of different financial institutions (more than half a dozen), and creditworthiness was never something that crossed my mind.
So is it fair or desirable that depositors should suffer—particularly when what we are talking about here are small innovative businesses that employ thousands of people between them?
But then should we just make FDIC insurance unlimited for everyone going forward? If not, why not?
Some suggest that a failure by the government to backstop depositors here will catalyze a chain reaction that leads to catastrophic bank runs at other (small) banks. The argument goes that why would anyone put money in a smaller bank that is at risk of a bank run? People will just move money into the biggest 4 banks in the U.S., which will cause further bank runs, kill small banks, and lead to more concentration and less competition in the banking industry, which is bad for everyone. I’m not very convinced by this argument. I think we are in a specific situation exacerbated by the uniquely concentrated customer base that SVB had (only 3% retail clients!) which produced a high proportion of uninsured deposits.
This was also a liquidity issue, not a situation where SVB plowed billions into FTX stock that is now worthless and they now can’t cover the hole. SVB apparently has enough assets to cover its deposit liabilities—it just needs time to sell them off. People probably aren’t going to lose a lot of money, unlike when your crypto exchange goes belly up.
As for the question why anyone will now deposit in small banks if there’s no backstop—in most banks, the existing backstop covers most depositors. Secondly, why do people bank with smaller banks today? Because their product is positively differentiated in various ways. I’m skeptical that the perception of a potential threat of a bank run happening to a smaller bank is going to outweigh all the other reasons that smaller banks exist in a way that leads to an existential crisis for them. But let’s see what happens when the markets open on Monday.
Another thing that is clear to me is that we now have a generation of workers who have been exposed to a financial crisis for the first time. To be sure, it’s mostly localized to tech startups (for now), but for almost 15 years now, money has been cheap and free-flowing and the last 9 months have been a huge shock to the system. It will be intriguing to see how these interesting times shape the psyche of Gen Z and younger millennials, just like how the financial habits of the Silent Generation were shaped for a lifetime after growing up through the Great Depression.
More to come
I still think the worst is yet to come for the economy as a whole. If interest rates are sustained at current levels (and it’s starting to look that way, as employment is still strong and inflation remains elevated), we’ll start to see them bite into the parts of the economy that are the most sensitive to rate rises, and that will cause a domino effect. I’m not sure where that is, but I could see, for example, companies that have maturing loans getting into trouble when they have to refinance at much higher market rates. Or unrealized losses that bondholders being forced to be realized due to liquidity needs.
If you spend much time on AI twitter, you might have seen this tentacle monster hanging around. But what is it, and what does it have to do with ChatGPT?
It's kind of a long story. But it's worth it! It even ends with cake 🍰
The failure of @SVB_Financial could destroy an important long-term driver of the economy as VC-backed companies rely on SVB for loans and holding their operating cash. If private capital can’t provide a solution, a highly dilutive gov’t preferred bailout should be considered.
The gov’t has about 48 hours to fix a-soon-to-be-irreversible mistake. By allowing @SVB_Financial to fail without protecting all depositors, the world has woken up to what an uninsured deposit is — an unsecured illiquid claim on a failed bank. Absent @jpmorgan@citi or… https://t.co/SqdkFK7Fld
Warren Buffett’s annual shareholder letter was published as part of Berkshire Hathaway’s annual report last week. This year’s was quite brief, but as usual, very accessible and a repeated reminder of the simple themes that have served him well over the decades.
After the growthy period of 2019-2021, Berkshire returned to outperforming the S&P index, returning 4.0% in a year where the market fell 18.1%.
“We are understanding about business mistakes; our tolerance for personal misconduct is zero.”
Berkshire buys businesses and also invests in public companies with the expectation of holding each indefinitely and therefore looks for enduring businesses with trustworthy managers. He notes that private, controlled businesses are almost never available at bargain valuations.
He notes that they make, on average, one “truly good” decision every 5 years, and that the performance of their portfolio of businesses includes “a large group that are marginal” but many that are “very good” and a few that are “truly extraordinary”.
For example, he highlights the performance of a 28-year holding of Coca-Cola. Cost base was $1.3B. As of 2022, it threw of $704M in cash dividends (53% annual yield on the original investment!) with a market value of $25B (19x capital gain).Put another way, if my parents had invested $100k in Coke when I was in primary school, that investment would today be almost $2M and generating $53k in dividends. Makes me think about what I should do for my kids to leverage the power of compounding. They are each currently, as Morgan Housel puts it, “time billionaires”.
Last year, Berkshire acquired another property-casualty insurer, growing its insurance float from $147B to $164M (the premiums it holds).
Berkshire holds a “boatload of cash and U.S. Treasury bills” and avoids behavior that could result in uncomfortable cash needs at inconvenient times, including financial panics and unprecedented insurance losses. Seems like a good personal approach too.
He rails against the politicization of stock buybacks. As a pure matter of mathematics, stock buybacks, when performed at a good (undervalued) price, are beneficial to all remaining shareholders.
His formula: retaining earnings in its businesses (Berkshire doesn’t pay dividends) + compounding + avoiding major mistakes + what he calls the American Tailwind: “America would have done fine withour Berkshire. The reverse is not true.”
Berkshire’s annual corporate tax bill of $31B is 0.1% of the entire American tax base.
He shares a bullet point list of some things Charlie Munger said on a podcast, including the value of being a patient, long-term investor, the dangers of leverage in wealth destruction, that you “don’t need to own a lot of things to get rich”, and that investing requires adapting as the world changes.
In reference to Charlie Munger: “Find a very smart high-grade partner — preferably slightly older than you — and then listen very carefully to what he says.”
Phones at Work
When I was younger, I generally disliked talking on the phone (there was a measure of anxiety attached to it), so the advent of SMS, emails, and IMs felt like a blessing to me, and more so as the younger generations pushed adoption of that socially, and then in the workplace. However, especially in tech workplaces, I now find myself missing phone calls.
To contact someone now, people typically reach out via Slack and then wait. Or schedule some time on their calendar and then wait. Or if it’s really urgent, they’ll send a text and then wait. It’s less intrusive, but also there’s no way to guarantee a quick answer for small things — even if the person is completely available, they just might not see your message for a while. And for a lot of these calls you simply don’t need video. You just need to get some shit done.
When I was a junior lawyer at a firm, we had these Cisco IP phones on our desk. People would just dial each other up by punching in 4 digit extensions. It was spontaneous, marvelously tactile, and it wasn’t a big deal. If the other person was unavailable (in a meeting, on another call, or just under the gun getting something out), they would ignore the call or send it to voicemail. (And if it was super urgent and you weren’t answering, they would physically show up at your door.) No one really worried about interrupting anyone. Clients would call out of the blue as well.
For a period I had an office next door to my supervising partner (I still remember his 4-digit extension, 15 years on), and he would still call me despite the physical proximity. I was thankful that he didn’t scream through the wall, even though that would have been marginally quicker.
These calls were a great way to deal with things in a minute that can now take quite a lot of minutes and relative effort to resolve over chat. Also, it’s a great way to ask a bunch of questions in a way that feels natural but, when done over a textual medium, can feel interrogational and even aggressive.
No one does this anymore, and I think you lose a valuable communication method and tool.
Further Observations
This week I learned that the U.S. Supreme Court opens each term with the pronouncement “Oyez, oyez, oyez!” Oyez is an Anglo-Norman word that means “Hear Ye!” It’s kind of weird that it took me so long to learn that given that I work in law and also started a blog called Hear Ye! over 20 years ago.
A 15-lawyer Canadian litigation law firm tried the 4-day work week for 2 years and published its results (Wednesday was their off day). They are pragmatic and interesting:
“Sometimes you’d have to work 5, 6, or 7 days a week and that is OK; that’s just life.” In other words, at a normal law firm you sometimes have to work weekends. In this case, the weekend is 3 days long, and you may have to work weekends. Particularly since clients expect availability and the courts have their own schedules. “In my interviews with our lawyers, they said they have had to work at least 1 hour or more on over 50% of Wednesdays.”
“You can attract the wrong crowd by advertising 4-day work weeks and need to get your message right and clear.” They put in a 3 month probationary period that requires new hires to work 5 days in the office until they get an idea of their work ethic.
“We have way, way less turnover.” This makes sense. And this suggests people chuck sickies quite often: “Our sick day requests have been reduced by over 80%.”
They still managed to grow revenue over that period (more than 2x over 2 years).
Drive to Survive (Season 5) “It’s not a documentary. It’s closer to Top Gun than a documentary.” —Toto Wolff. Great way to whet the appetite for the new F1 season, which started this week!
There's an easy way to measure Google's moral temperature at any given time: how closely the ads resemble the search results. pic.twitter.com/QZ15aOa0Ak
This is the online journal of Stuart Loh.
Originally from Sydney (8:23pm), Stuart is currently based in California (3:23am). New to this site? Hear Ye! has been online since 1998. View the
Archives, browse the
Categories, run a
Search or get the
RSS Feeds.
9:00pm 
