182: AirPods Requiem, Cloud 100 Infrastructure, Nvidia's AI Dominance, Right vs Wrong Biz Simplification, Twitter, Lab-Grown Meat, Cryptopunks, and Polymetric Rapping
"There’s a reason why phone numbers aren’t longer"
TEMPUS EDAX RERUM
(Time, devourer of all things — including AirPods)
🪦 After years of heavy daily use, I think my AirPods are finally nearing the end of the road… 🚧 The left one is now only connecting about 50% of the time, and that rate is decreasing by the day.
This is a requiem for my AirPods.
Since getting AirPods in 2019 (they’re the second generation, but technically, the update was more like a 1.5), I must’ve literally listened to thousand of hours of music, podcasts, and audiobooks on them.
I gotta say, they’re one of the highest-value-for-money pieces of electronics I’ve ever owned.
They fixed a problem I didn’t realize I had, and now I couldn’t go back to using wired headphones — like an animals. By reducing friction at every point in the process, they’ve made me listen to a lot more audio content than I would’ve otherwise.
And for me, for my life priorities, something that gets me to learn more things, explore more ideas, and listen to more good music has got to be very high on my 💙 list.
If I think back to the introduction of this product in 2016 (has it been that long already?), I’m still impressed at the engineering. These batteries at tiny, and it was brilliant to put a bigger battery in the portable case to keep the earpieces charged, the use of magnets, etc.
Most ‘wireless’ headphones at the time had a wire running between the buds — remove it like Apple did and you run into a lot more potential desynch problems, and you now have to worry about two different radios needing a good signal *through your skull and body*, which thankfully isn’t metal (except this guy), but it’s still not entirely radio-transparent (you’ve gone bones in there, and plenty of water..).
If you carry your phone in your jeans pocket, the low-energy bluetooth signal has to go through your whole body diagonally and up to the opposite ear… That’s not the easiest path.
But they made it work through feats of miniaturization, and what a great experience it has been. Now I just have to somehow figure out how to wait until the new AirPods, because I hate buying something just before a new model comes out…
Maybe I’ll just have to buy AirPods Pro, as I don’t think these are about to be updated ¯\_(ツ)_/¯
🎓 I feel like one advantage autodidacts have is that they learn *because they want to learn*, they actually try to understand to scratch an itch. As with everything, motivation matters a lot.
In school, you learn on someone else's schedule because you have to to pass a test, so you do whatever you have to to get that grade, but may not truly learn much, understand, or retain for long. A lot of it is more about “guessing the teacher’s password” than learning something well enough that you could teach it to someone else who doesn’t know it.
📝 Feels like a while since I gave an update on my use of Obsidian.
I still love it.
💚 🥃 The price of a couple coffees or one alcoholic drink isn't a bad trade for 12 emails per month (plus 𝕤𝕡𝕖𝕔𝕚𝕒𝕝 𝕖𝕕𝕚𝕥𝕚𝕠𝕟𝕤) full of eclectic ideas and investing/tech analysis. That’s 77¢ per edition.
If you make just one good investment decision per year because of something you learn here (or avoid one bad decision — don’t forget preventing negatives!), it'll pay for multiple years of subscriptions (or multiple lifetimes).
As Bezos would say of Prime, you’d be downright irresponsible not to be a member, it takes 19 seconds (3 secs on mobile with Apple/Google Pay):
Investing & Business
Which ☁️ do the ‘Cloud 100’ use as infrastructure?
Reader and supporter (💚 🥃) Brent Muio shared with me some scuttlebutt he did on the Forbes Cloud 100 list (y’know, one of these clickbaity lists that the media loves because it exchanges some prestige/publicity for access).
He looked at each company on the list and tried to figure out which of the big clouds they used as infra (sometimes a single one, sometimes multi-clouding it).
Disclaimer to make it clear: This is based on public information and sometimes best guesses based on things like job postings and such, no promises of accuracy.
Here’s what he found:
AWS -78 use AWS exclusively -87 use AWS in some capacity Azure -3 use Azure exclusively -6 use Azure in some capacity GCP -2 use GCP exclusively -8 use GCP in some capacity Alibaba -1 uses Alibaba exclusively -2 use Alibaba is some capacity Huawei -1 uses Huawei exclusively On-Premise -5 use on-premise solutions
Really shows how AWS is still king with younger businesses and startups, despite the huge growth at Azure, which has to more largely come from older and more mature businesses.
Why we *have* to simplify businesses in our minds…
Human working-memory is very limited. There’s a reason why phone numbers aren’t longer…
Whether we want it or not, we tend to focus on just a few things at once, even if we’ve learned a lot of them. We can retrieve stuff from long-term memory, but we can only hold so much loaded up in ‘RAM’ at once.
Some people seem to think that the split between those who are mostly right and those who are mostly wrong is that the right ones are just smarter and can keep track of way more things and details.
I think the explanation is a bit different.
The main demarcation I see is that amateurs and unskilled operators will focus on a handful of the wrong things while the most talented/experienced/wise players will focus on a handful of the right things (or at least on righter things — life is more about being less wrong than about being right — more a direction than a destination you ever reach).
It doesn’t mean that the high-performers haven’t parsed through more pieces of data and thought about things for way longer and more deeply over time, or that they aren’t also smarter, but most of this serves to train the implicit pattern-recognition models that help narrow it down to better KPIs.
So it probably doesn’t mean that they can really play with all this stuff in their heads in real time, more that they gathered the ingredients and then cooked the recipe. Even the more complex recipe leaves you with digestible food at the end of the process…
(we’ve all got models of reality that are kind of between analog NTSC and 720p, maybe some are 1080p. But nobody gets all the way to 3D 8K HDR VR)
It’s a bit like chess grandmasters. Non-chess people think that they’ve got to be able to just instantly rattle through possibilities like a computer would, but when you listen to them explain how they do it, it’s a lot more about recognizing large categories of patterns and training their intuition through endless iteration of deliberate practice and study.
This is why Buffett can give a ‘yay or nay’ on a deal in a very short amount of time.
He’s seen all the pieces on the board so many times, he’s been playing the same kind of games for so long… He can just focus on a handful of the most important thing and that gives him a high enough batting average so that over decades, his left almost everybody else behind.
h/t friend-of-the-show and supporter (💎🐕) Mostly Borrowed Ideas (I had the idea for this thing during a conversation with him a couple days ago)
What Could Have Been, Twitter Edition
Friend-of-the-show Terminal Value:
Reminder that Twitter could have been TikTok. They bought Vine and shut it down. In blown opportunity terms, that's drummer-who-got-replaced-by-Ringo level. That's Ron-Johnson-selling-his-Apple-stock-in-2007 tier
Heck, Twitter could’ve been what Twitter could’ve been, and that’s a big miss too…
(Ok, I realize facts shouldn’t get in the way of a good joke, but… Yeah, Vine wasn’t quite Tiktok, it didn’t have the algo-fu that Bytedance brought to the format to take it to the next level. But let’s say that instead of shutting Vine down, Twitter had kept improving it and had been a nimble product/engineering organization, they could have been fast-followers of what Bytedance was doing in China and could possibly have been the #1 player in the space outside of China. Maybe.)
Nvidia Market Share for “AI Processors”
It’s not breaking news, but I don’t pretend to play the media game here, just looking for interesting stuff… So here are some numbers on Nvidia’s market share for AI processors/accelerators (which includes more than GPU — they also include ASICS and FPGAs):
Leading graphics processing unit (GPU) supplier NVIDIA Corp. maintained its dominant position in the global market for artificial intelligence (AI) processors used in the cloud and in data centers in 2020, with an 80.6% share of global revenue, according to Omdia.
NVIDIA generated cloud and data center AI processor revenue totaling $3.2 billion in 2020, up from $1.8 billion in 2019 (Source)
The graph on top shows a forecast up to 2026. Who knows what reality will be like exactly, but the trend is clear, and if I had to guess on a surprise upward or downward, I would guess that we’ll be surprised upward, because innovation tends to add new ways to consume/use compute in unforeseen ways, and it’s hard for people to model these innovations, so forecasts tend to be based on business as usual continuing.
Science & Technology
‘No, Vaccinated People e Are Not ‘Just as Likely’ to Spread the Coronavirus as Unvaccinated People’
This misunderstanding, born out of confusing statements from public-health authorities and misleading media headlines, is a shame. It is resulting in unnecessary fear among vaccinated people, all the while undermining the public’s understanding of the importance—and effectiveness—of getting vaccinated.
So let me make one thing clear: Vaccinated people are not as likely to spread the coronavirus as the unvaccinated. Even in the United States, where more than half of the population is fully vaccinated, the unvaccinated are responsible for the overwhelming majority of transmission. [...]
this framing missed the single most important factor in spreading the coronavirus: To spread the coronavirus, you have to have the coronavirus. And vaccinated people are far less likely to have the coronavirus—period. [...]
for those instances of a vaccinated person getting a breakthrough case, yes, they can be as infectious as an unvaccinated person. But they are likely contagious for a shorter period of time when compared with the unvaccinated, and they may harbor less infectious virus overall. [...]
recent data from New York City that show that more than 96 percent of cases are among the unvaccinated. Only 0.33 percent of fully vaccinated New Yorkers have been diagnosed with COVID-19. [...]
As an emergency-medicine physician, I’ve seen firsthand the vaccines’ dramatic role in reducing severe outcomes from a virus that flooded my emergency room early in the pandemic. And as a member of one of the first groups vaccinated in the rollout, the shots kept me safe while I cared for patients and prevented me from bringing the virus home to my family. (Source)
‘Lab-grown meat is definitely going to be a thing’
There’s a fairly long piece here that goes through all kinds of liner and exponential technologies, showing the difference (coal vs solar, Moore’s Law, etc). I’ll skip to the juicy bit (ugh), because I assume that if you’re reading this, you already know about these exponential curves. So here’s the stuff:
The first data point we had for a kilogram of lab-grown meat was just under $2.3 million. The second data point, just three years later, was just under $40,000. The third, a year after that, was $80. Last year we got down to $22.43.
Cultured meat has gone from $2.3 million dollars per kilogram to just over twenty dollars per kilogram—a 99.999% drop in price—in just seven years.
Last year, cellular agriculture company Mosa Meat announced they had reduced the cost of their cell culture media (the nutrient broths that feed the cells so they can proliferate) by 88% in 10 months [...] This year, the same company revealed they’ve been able to reduce the cost of their fat differentiation media by 98.48% in 18 months.
Look at that beauty. Note that the Y axis in the chart above is logarithmic.
To me it’s not even a question that eventually 90%+ of the meat consumed by humanity won’t come from animals, it’s just a question of when.
And because the current system is pretty inhumane and polluting and inefficient and unhealthy in major ways, making this “when” happen as fast as possible is an imperative on economic, environmental, health, and moral grounds.
I’m hoping that more and more talent and R&D dollars and companies are founded in the space to accelerate things. Like research into the diseases of aging, this has huge leverage to benefit humanity and reduce unnecessary suffering on this pale blue dot.
P.S. My call for finding a good name for this is still unanswered. I’m looking for something better than “lab-grown meat” or “clean meat” or worse of all, “green meat”.
The Arts & History
When Form Follows Function
Why are so many NFTs made with 8-bit-style pixel art?
I think it’s because it’s easier to mass-produce this type of artwork (there are 10,000 ‘cryptopunks’ like the above, many of which are the same head foundation with glasses, cigarettes, or hats cut & pasted on top, or skin tone changed, etc).
Commercial constraints have long affected the artistic side of things, so this isn’t new, but it’s interesting to think that a certain aesthetic could become popular in a certain community because of it.
Rapping in 5/4 Time Signature & Polymetric Flows
I enjoyed this short demonstration of musical time signatures and why 5/4 isn’t used much in hip hop, but a very nice demonstration of how it’s totally possible:
This one about polymetric time signatures and the different ways to resolve where the beats match was also very cool:
The artist is Mazbou Q. I like the way he explains things, he’s obviously a very talented educator. And his bio is cool, he’s an eclectic man after my own heart:
Mazbou Q is a UK-born, Tāmaki Makaurau based Nigerian artist and producer. Stemming from his background in classical, heavy metal and West African Highlife, Mazbou Q's particular brand of Hiphop challenges the very notion of genre and captivates those with a deep appreciation for musicality.