111: U.S. Semiconductors, Mindshare, TSMC, Microsoft's ARmy, Nvidia, Tiny Gyroscopes, Johnson & Johnson, Valve's Half-Life Dev Process, and Robert Heinlein's Paper Tech
"ok, augmented reality jokes just can’t be funny"
You can’t learn if you think you already know.
—Ryan Holiday, Ego is the Enemy
🤘 I think growing up listening to more underground types of music and being part of those sub-cultures can be a big asset for this.
When you grow up listening to death-metal, hardcore punk, or hip-hop (before mainstreaming), conforming & being like others is a dirty word.
Not that there isn’t the mirror issue of reflexively doing the opposite of the mainstream, or disliking things just because they are popular, which is just as unthinking as doing the same thing as others and liking things because they are popular, or that there isn’t pressure to conform within the sub-group, but at least it gives some practice at going against the grain.
🛀 I read somewhere a while ago that with intelligence, you can only see a couple levels above yourself.
Too much higher, and you can't separate the super-genius from the hyper-genius, it all blends together because the differences can appear subtle to the untrained eye but still profound.
I think it probably works similarly with other things, like certain skills.
For example, to the best curling player in the world, small nuances in technique probably really stand out between the #1 best player in the world and the #5, but to the average player, if you didn’t know their rankings and statistics, and just watched them play, they’d probably all seem pretty close.
Where else does this apply?
(yes, I took curling as an example — felt like Basketball probably had too many associations, so it’s harder to look at it as objectively. It’s like the old trick of never giving examples using current political parties, but instead use historical references like Roman Empire factions or whatever)
🚧 Those of us who live in places where the winter destroys roads every year need automated, roboticized trucks that drive around 24/7 and fix hundreds of potholes per day, just by parking over them and having everything under the carriage to dump some hot asphalt and flatten it, with flashing hazard lights and arrows to direct traffic around it while it operates.
They could be self-driving or have a human driver, but the actual road-work should be as close to automated as possible, with all kinds of cameras with ML image recognition and LIDAR to detect and correct problems in roads.
The thing would drive back to the station when it needs more fuel and asphalt, and then rinse & repeat forever.
Or we could just have a bunch of humans with regular old low-tech equipment and just get it done, but clearly that’s not working because the roads look like civil war, so I guess we need to invent some cool sci-fi shit ¯\_(ツ)_/¯
A Canadian can dream, can’t he?
🦠 Where I am, there’s been a spike in COVID19 cases, and the area is going into lockdown next week. No school for the kids, most businesses closing, curfews, the whole nine yards.
I’m so ready for this pandemic to be over. Let’s find an extra gear for those vaccine factories!
If the newsletter is a bit weird next week, it’s probably because I have a kid talking to me about Zelda as I write…
💚 🥃 *whispers in your ear*
Investing & Business
Distribution… in the Mind
A couple friends were discussing this Alex Rampell quote:
The battle between every startup and incumbent comes down to whether the startup gets distribution before the incumbent gets innovation.
It made me think about how there’s the more visible aspects of distribution (warehouses, stores, etc), but there’s also the more intangible ones.
Sometimes it's mindshare.
When your potential customers automatically think of you to do X, that's very powerful (and drives way down your CAC).
I guess you could use different terminology for this, call it a brand that lowers search costs, etc. The words matter less than the underlying principle that this is effectively a kind of distribution, a way to reach your customers with your products and/or services.
U.S. Semiconductor Industry is getting… Stimulated (🤭)
From the White House’s 2 Trillion plan:
Invest in R&D and the technologies of the future:
Advance U.S. leadership in critical technologies and upgrade America’s research infrastructure. U.S. leadership in new technologies—from artificial intelligence to biotechnology to computing—is critical to both our future economic competitiveness and our national security. Based on bipartisan proposals, President Biden is calling on Congress to invest $50 billion in the National Science Foundation (NSF), creating a technology directorate that will collaborate with and build on existing programs across the government. It will focus on fields like semiconductors and advanced computing, advanced communications technology, advanced energy technologies, and biotechnology. [...]
The President also is calling on Congress to invest $50 billion in semiconductor manufacturing and research, as called for in the bipartisan CHIPS Act.
Meanwhile, at TSMC
“TSMC expects to invest USD$100b over the next three years to increase capacity to support the manufacturing and R&D of advanced semiconductor technologies,” the company said in a statement responding to local media reports. [...]
already planned a record capital expenditure of as much as $28 billion this year, but recent trends and developments have pushed for even more capacity [...]
In a letter to customers obtained by Bloomberg News, TSMC Chief Executive Officer C.C. Wei wrote that the company’s fabs have been “running at over 100% utilization over the past 12 months,” but demand still outpaced supply. Thousands of new employees are being hired and multiple new factories are under construction, he added, and TSMC will suspend wafer price reductions for a year from the start of 2022. (Source)
‘Microsoft wins U.S. Army contract for augmented-reality headsets, worth up to $21.9 billion over 10 years’
The Pentagon announced that Microsoft has won a contract to build more than 120,000 custom HoloLens augmented-reality headsets for the U.S. Army. The contract could be worth up to $21.88 billion over 10 years [...]
It follows a $480 million contract Microsoft received to give the Army prototypes of the Integrated Visual Augmented System, or IVAS, in 2018. The new deal will involve providing production versions. [...]
The standard-issue HoloLens, which costs $3,500, enables people to see holograms overlaid over their actual environments and interact using hand and voice gestures. An IVAS prototype that a CNBC reporter tried out in 2019 displayed a map and a compass and had thermal imaging to reveal people in the dark. The system could also show the aim for a weapon. (Source)
The ARmy. Amirite? (ok, augmented reality jokes just can’t be funny)
Plus it’ll be great to play Minecraft during downtime in the barracks, and voice-chat on Discord if MS buys it… Synergies!
Interview: Bill Brewster (yeah, I mean, being interviewed)
I can’t do justice to the first part, so I’ll let you listen for yourself, but they also cover some very interesting ideas, and most importantly, raise some interesting questions without feeling like they have to prescribe the answers.
Often, asking the right questions is plenty enough to move things forward, especially when few others are asking those questions.
Some choice highlights:
[Jim, on Mimetic copying] When someone is getting stoned to death, the one you want to study of the one who threw the first stone, because the others just copied him.
[Jim, on speculation and betting money when the odds are against you:] You always want to be the house... Losses did not pay for all that marble at the Bellagio.
‘LinkedIn confirms it's working on a Clubhouse rival and will begin beta testing soon’
Ok, the rapid copycatting has officially gone too far.
There are Worse Things Than a Little Inflation
Friend-of-the-show Andrew Walker was spitting some 🔥 in his always-excellent end-of-month post (and I’m not just saying that because he had some kind words for the club here):
I know plenty of people are worried about the combo of low interest rates, record deficits, and stimulus leading to some inflation. I tend to think those who are worried about inflations are missing the forest for the tress; sure, it's a risk, but COVID just shut our country down for a year and still killed more than 500k Americans (and counting). Does it really matter if we get a little too hot in the recovery and inflation goes to 3% or 4% instead of 2%? In the grand scheme of things, probably not?
Science & Technology
How Gyroscopes/Accelerometers Fit into Smartphones
I have a basic understanding of how gyroscopes work, but I never looked too deeply into them. When I thought about them, I mostly pictured in my mind the mechanical model above…
Recently I was looking at my smartphone and wondered, how do we fit the equivalent of that in there. There has to be a modern, miniaturized, mostly solid state way to do it, but how does it work?
It’s actually really cool and clever.
Rather than try to explain with words, I’ll point you to a couple of videos that do a good job (they’re a few years old, but the principles should still apply):
3-axis MEMS gyroscope (“Two examples of MEMS gyroscopes are discussed. The first example is the very first MEMS gyroscope made by Draper Labs in 1993. The second example is an 3-axis MEMS gyroscope inside iPhone.”)
Oops, Johnson & Johnson Edition
Workers at a plant in Baltimore manufacturing two coronavirus vaccines accidentally conflated the ingredients several weeks ago, contaminating up to 15 million doses of Johnson & Johnson’s vaccine (Source)
How many people will die because of this mixup?
To balance it out, here some good news:
Pfizer is shipping its doses ahead of schedule, and Moderna is on the verge of winning approval to deliver vials of vaccine packed with up to 15 doses instead of 10
Geeking out on Nvidia’s A100 GPU (from May 2020)
I was DM’ing Dylan Patel about some stuff and ended up checking out this interview that he did with James Wang and Paul Teich (I have to mention that “Teich” is a perfect name for someone working in tech) around the time Nvidia released their A100 Ampere chip last year:
It’s good stuff and goes into enough detail that I learned a few things even though I’ve been following this architecture since release.
They also discuss the positioning of Intel and AMD, the custom silicon coming from the hyperscalers and how each player is positioning itself to try to accrue value to its part of the stack. There’s also talk of the Mellanox roadmap. As I said, good stuff.
I gotta say, what a pleasure it is to get quality hardware that feels solid and well designed. The microphone alone weights two pounds and is all metal. There’s no kidding around with this.
I’m still waiting for the mic stand that I ordered separately to arrive, but I’ve already had some fun playing with the new shiny. Hopefully my next Mini-Podcast sounds better than the first three (here's #3).
AVShop.ca even put some free candy in the box. Robert Cialdini would be proud!
The Arts & History
Valve’s Half-Life Theory of Fun
Great anecdote about the development of one of the best and most influential games of all time, Half-Life, via Byrne Hobart ($ sub required):
the development process for Half-Life was itself recursive: at one point, the game was close to complete, but playtests showed that it wasn't especially fun. So Valve started over: they built a single level with every fun and distinctive element of the game, then developed a Theory of Fun based on what made the level appealing, and only then iterated to build the rest of the game. Take the atomic unit of the product, make it perfect, and then crank out more similarly-perfect atomic units and chain them together, measuring every new element against the perfect-level standard. It's hard to develop a good theory of how to make a fun game in the face of either no game or a not-very-fun one, but relatively easier to do just enough practical experimentation to get the theory right, and then figure out which parts scale.
There’s more detail on this in this piece:
The first theory we came up with was the theory of "experiential density" — the amount of "things" that happen to and are done by the player per unit of time and area of a map. Our goal was that, once active, the player never had to wait too long before the next stimulus, be it monster, special effect, plot point, action sequence, and so on. Since we couldn’t really bring all these experiences to the player (a relentless series of them would just get tedious), all content is distance based, not time based, and no activities are started outside the player’s control. If the players are in the mood for more action, all they need to do is move forward and within a few seconds something will happen.
The second theory we came up with is the theory of player acknowledgment. This means that the game world must acknowledge players every time they perform an action. For example, if they shoot their gun, the world needs to acknowledge it with something more permanent than just a sound — there should be some visual evidence that they’ve just fired their gun. [...] if the player pushes on something that should be pushable, the object shouldn’t ignore them, it should move. If they whack on something with their crowbar that looks like it should break, it had better break. If they walk into a room with other characters, those characters should acknowledge them by at least looking at them, if not calling out their name. Our basic theory was that if the world ignores the player, the player won’t care about the world.
A final theory was that the players should always blame themselves for failure. If the game kills them off with no warning, then players blame the game and start to dislike it. But if the game hints that danger is imminent, show players a way out and they die anyway, then they’ll consider it a failure on their part; they’ve let the game down and they need to try a little harder.
As an aside, it’s impossible to find a screenshot of the original Half-Life that doesn’t look like crap to modern eyes (I’ve tried), yet at the time, the graphics blew everyone’s minds.
Always interesting how that stuff works, even if you know about it. When you don’t see a game for long enough, you have this imagine in your mind that collides with the reality when you look up what it actually looked like.
Robert Heinlein’s Analog Auto-Responder
Heinlein engineered his own nerdy solution to a problem common to famous authors: how to deal with fan mail. In the days before the internet, Heinlein’s solution was fabulous. He created a one page FAQ answer sheet – minus the questions. Then he, or rather his wife Ginny, checked off the appropriate answer and mailed it back. While getting a form letter back might be thought rude, it was much better than being ignored, and besides, the other questions you did not ask were also answered! Indeed, it is both remarkable and heartwarming that Heinlein replied at all to most mail. Can you imagine other great authors doing the same — even with a form letter? Heinlein’s form is very entertaining to read because you are forced to reconstruct the missing requests. (Source)
Very clever, and a good use of time.
Graph Says it All: