145: What is your 母题?, Nvidia's Generalization, Sam Walton & Success Despite Mistakes, Intel Discrete GPU, Lithium-Ion Battery, Halo, Psych of $, Legend of Zelda DIY-Remaster

"it's easy to make things hard and hard to make things simple"

The foundational challenge in design is that it's easy to make things hard and hard to make things simple.

David Holz

📈📉 In the real world, things are rarely ever stable. Plateaus1 don’t last.

They either grow or they shrink.

Like for example, if on average over the course of a day I open more browser tabs than I close, or if I add more things to read to my “to read” file than I have time to read, eventually I have to declare bankruptcy on some of these. 😩

I hope there weren’t too many gems or life changing ideas in some of these unread tabs/links… 😔

That’s why I need a Neuralink-type tech, I need more input bandwidth…

🛀 Often when I send a text message to my wife and she's also in the house, I think about how the data is being sent to a server located probably hundreds of kilometers away, only to bounce back to the same house, then be converted to radio waves (wifi) that go to a different floor.

💉 My second-dose appointment was originally in late August, but recently I was able to move it forward to June 28th. 🤘

🤔 If you’re going to be a generalist — and I don’t mean just in the financial usage of that word — you can’t compete with specialist head-on. You have to go where they aren’t…

You have to make connections between far-flung ideas and fields, you have to be nimble and able to follow things wherever they lead and drop topics when they go cold on you and then be ready to switch back to them when something interesting is going on again.

Otherwise, you’ll have the worst of both worlds: Not as good as a specialist at whatever you decide to stick to, and not as good as true generalists because you decided to stick to something.

As a kid, there’s a lot of stuff I didn’t even begin because I knew I wouldn’t go all the way. That was a mistake. It’s fine to learn a bunch of programming concepts even if you know you won’t be a coder, a bunch of music/biology/astronomy concepts even if… etc.

You can use the 80-20 rule to leverage multiple interests into a really wide base to build on.

🛀 ⏱ Wow, I never thought I would have an ongoing riff on time-travel, but here we are (see the intros of editions #140 and #144…).

Another thing that I haven’t seen mentioned in fiction that features time-travel:

Kids are born from the fusion of the woman’s egg and of the man’s sperm cell. Because there’s millions of spermatozoa, any change in timing at all is likely to result in a different one getting to the egg first (and fertilizing it, if all goes well).

This means that unlike what we see in most time-travel fiction, almost any change at all to a timeline would result in really huge changes downstream. Imagine that traffic patterns are slightly different in a city and everybody’s schedule is off by seconds.

Different babies.

Imagine something slightly different appears in the evening news or printed in a newspaper, and it takes slightly longer to read, or makes millions of people spend a little longer or a little less time thinking about the story, etc.

There’s an endless amount of things that can change things in very small ways, but ways that are big enough to result in different babies.

But that’s not all!

Because there’s millions of gametes on the male side, but only one egg per cycle on the female side, this means that the changes would come almost entirely from a churn on the male genetic side. The female side would stay fairly stable in such a scenario.

So people in such a “slight change due to time travel” scenario wouldn’t be entirely different — after all they have the same parents — but they would be different mostly in the 50% of genes that they’re getting from their father.

Not sure this would’ve made ‘Back to the Future’ a better movie, but it’s kinda cool to think about…

🧠 🪑 If you make a living sitting in a chair all day, reading stuff, thinking about stuff, ingesting information and trying to make the best decisions that you can, etc..

Then it's crazy not to periodically make efforts to improve how you acquire, process, retain, and connect information to generate ideas.

There are endless variables to tweak here. This is partly what my recent experiments with note-taking systems were about recently, but you can also schedule your “deep work” periods at different times of the day, read at different times, on different mediums (paper vs Kindle vs audiobook), take walks without listening to anything, with instrumental music, with a smart friend. Go spend some time in a floatation tank or in the woods or at the public library. Do video calls with random people you find interesting on Twitter. Etc.

Otherwise, you're basically a runner who always trains the same way for the race, and never wonders if maybe they'd get better results if they improved the training itself.

🤔 I love this, by friend-of-the-show and supporter (💚 🥃) Lillian Li:

What is your 母题?

💚 🥃 Last I checked, we were at 4% paid supporters vs 96% free subscribers on this boat. This canoe. This Mississippi paddleboat steamer.

My goal, to feel like this is evolving into a sustainable model, is still 5% paid/95% free. But if we bust through 5% like the Kool-aid man, I may have to think up a stretch-goal and some cool bonus stuff to do.

You guys are at the wheel on this one, all I can do is crank out editions and let you decide if you’re getting good value and want to pay 77¢ per shot to keep it going:

💙 Subscribe now 💙

Investing & Business

One of Many Forces Reshaping the World

What happens if/when that goes to $50? $30?

‘U.S. senators propose 25% tax credit for semiconductor manufacturing’

It’s still raining:

A bipartisan group of U.S. senators on Thursday proposed a 25% tax credit for investments in semiconductor manufacturing [...]

The group did not immediately provide a cost estimate for the measure, which is on top of recent proposed semiconductor funding. Last week, the Senate approved $52 billion for production and research on semiconductors and telecommunications equipment. [...]

U.S. share of semiconductors and microelectronics production has fallen to 12% from 37% in 1990.

The senators said up to 70% of the cost difference for producing semiconductors overseas results from foreign subsidies. (Source)

Success *Despite* Mistakes

In a discussion with Patrick O’Shaughnessy, Gavin Baker mentioned a kinda mind-blowing fact about Sam Walton: He waited 18 years after opening his first store before expanding outside of Arkansas (“Niche dominance”).

It’s probably mentioned in his bio, but I had forgotten it (damn weak, biological leaky memory).

My devil’s advocate background brain-process couldn’t help but think that without a control group, it's hard to know if he wouldn't have been even more successful if he hadn't waited as long.

This isn’t really about Walton. It’s a more general point: it's easy to assign successful people's success to the things they did, but their success is in fact a mix of:

  1. Succeeding *because* some of the things they did, and;

  2. Succeeding *despite* some of the things they did.

Another thing that I think probably happens pretty often — and I’m not saying it’s what’s happening here — is that successful people make decisions based on a specific context, and people later copy the decisions without taking that context into account.

For example, let’s pretend that Sam Walton didn’t expand because it was very hard to find any risk capital at the time, and he had to self-fund expansion, and the environment was particularly uncertain with the oil shock and stagflation of the 1970s so he wanted to be extra-careful, or something like that.

A modern founder who looks back and only gets from that “you better wait a while before you expand out of your niche because Sam Walton did it” may be doing something very different from what a young Sam Walton would be doing if he was starting again in today’s different context.

Nvidia as a General Computing Company

Or I guess more specifically, an “accelerated computing company”.

Certainly not one that just focuses on the polished-sand/chip level (having more software than hardware engineers, as they often say these days).

Here’s their CFO recently:

All of these enterprises need assistance, and they need someone working to help them on that software, so the deployment and the ease of use is there for them. So when you think about NVIDIA and our work, sure, building and designing overall hardware -- hardware, silicon is top of mind. But we also have a very large focus on software today, software that is included in almost each and every single one of our platforms that we have. [...]

there is definitely a recognition NVIDIA has changed from its early days of a semiconductor overall chip company. This is something that we have moved to in a full system and essentially a computing company. Rather than looking at us just as an overall chip company. We've continued to work in terms of building out ecosystems by building out essentially anything you may need in the data center to complete that and be a full computing company. [...]

it is also important to think about how you have stitched together the ecosystem to use the overall hardware that is put together. We are in the early days of acceleration and AI workloads, many more innings in front of us. [...]

But one of the things that NVIDIA has that separates us is that development platform. The development platform is not always right front-and-center or the first thing that you see when you start the use of an overall GPU. But it is essential to keeping the movement forward in expanding the types of workloads that are using AI.

These types of new expanding workloads have expanded the use case for overall GPUs and the need for developing frameworks and stitching that together. So although it may seem that there is not a need for that underlying CUDA development platform, it is essential, everyone, in terms of as they think going forward, is jumping into CUDA to continue expand the uses of the GPU for new types of locals that are developing, that work still has to be created and the software is needed to help them advance that work.

So there will be from time-to-time an opportunity for custom chip for certain workload. But if we think about the future of overall AI, having a platform that is flexible, programmable to change with all of the advancements, we think we're well positioned for that.

The Psychology of Money

I recently finished the audiobook of 'The Psychology of Money' by Morgan Housel.

I think it's quite good, really well-written. No filler.

Probably the book I'd be most likely to recommend to non-financial people to get intro'd to the big concepts in an easy-to-understand and entertaining way.

This book does the 80-20 thing on a lot of other books and concepts, all in the same package, and that’s very valuable.

To be clear, I’m *not* saying it’s just good for newbies.

Most of the important things in finance and investing and behavioural economics and the psychology of happiness are simple but hard (to keep in mind, to put into practice consistently over time). So it’s useful — at any level — to refresh your memory once in a while and go back over the basics.

The Halo Effect

This tweet by the ever-wise Ed reminded me of a book that I think should be more widely read, ‘The Halo Effect’ by Phil Rosenzweig.

I’ve recommended it in a past edition, but hey, they say that the more often you hear something, that more likely you are to believe that it is true, so may as well use that power for good!

I like to say that this is the companion book to ‘The Outsiders’2 by Will Thorndike, and that both books taken together are more powerful than each individually (2+2 = 5.7 — yes, why not be overly precise for no reason?).

💙 Subscribe now 💙

Science & Technology

Why Short-Haul Flights are Going Electric

Neat video, makes a pretty compelling case. Seems more a question of “when” than of “if”.

Software is Eating the World that the World is Eating

Jahmy Hindman, chief technology officer at John Deere, says that “John Deere employs more software engineers than mechanical engineers now”.

Why? Here’s Hindman’s vision:

I’m going to paint a picture for you: It’s this idea of enabling each individual plant in production agriculture to be tended to by a master gardener. The master gardener is in this case probably some AI that is enabling a farmer to know exactly what that particular plant needs, when it needs it, and then our equipment provides them the capability of executing on that plan that master gardener has created for that plant on an extremely large scale.

You’re talking about, in the case of corn, for example, 50,000 plants per acre, so a master gardener taking care of 50,000 plants for every acre of corn. That’s where this is headed, and you can picture the data intensity of that. Two hundred million acres of corn ground, times 50,000 plants per acre; each one of those plants is creating data, and that’s the enormity of the scale of production agriculture when you start to get to this plant-by-plant management basis. [...]

we’re putting GPUs on machines today to do vision processing that would blow your mind. Nvidia GPUs are not just for the gaming community or the autonomous car community. They’re happening on tractors and sprayers and things too. (Source)

h/t Friend-of-the-show and supporter (💚 🥃) Brad Slingerlend

Intel Discrete GPU Benchmark Leak

Intel makes a lot of integrated GPUs, but the discrete GPU game is owned by AMD and Nvidia.

Intel wants in that game, and they’re getting close to release (late 2021? 2022?) what they call the Xe-HPG, with multiple variants with more or less RAM and execution cores.

Some benchmarks have leaked (though you never know if these are real), and they’re pretty impressive for a debut, putting the Intel with 448 EUs less than 10% below AMD’s RX 6700 XT and Nvidia’s RTX 3070.

Of course, raw horsepower isn’t the only thing that matters. It’ll be interesting to see how well Intel’s card plays with the software.

Pulse Secure VPN App Breached by China

A cyberespionage campaign blamed on China was more sweeping than previously known, with suspected state-backed hackers exploiting a device meant to boost internet security to penetrate the computers of critical U.S. entities.

Basically, they hacked the software that is used to connect to encrypted VPNs. It’s like getting access to a secure building by hacking part of the security system itself.

The Associated Press has learned that the hackers targeted telecommunications giant Verizon and the country’s largest water agency. News broke earlier this month that the New York City subway system, the country’s largest, was also breached.

Security researchers say dozens of other high-value entities that have not yet been named were also targeted as part of the breach of Pulse Secure, which is used by many companies and governments for secure remote access to their networks. (Source)

That’s not good, right?

‘Biased humans’

h/t Janelle Shane

💙 Subscribe now 💙

The Arts & History

Turning the Original Legend of Zelda to 3D

This is really cool, and shows the power of the tools available to individual creators and artists today.

If you’re around my age (I’m 39), you probably still have the overworld theme from the original 8-bit Zelda stuck in your head, and spent countless hours as a kid playing and daydreaming about this game (though I was a bit young to fully get the original Zelda… My most iconic instalment of the franchise is Zelda 3 (‘A Link From the Past’, on Super-Nintendo)..

CodyCantEatThis turned the original 2D game into a 3D game, and the video above shows the process and the design decisions that he had to make. If you just want to see some gameplay, this other video shows more of that.

But trust me, if you watch the video that explains how he rebuilt the game in 3D, coded simple AI for the enemies, made the models, etc, you’ll learn a ton.

This type of project has so many little problems to solve and interesting “eureka” moments. This type of stuff was what I loved about making maps for Doom, Quake, Starcraft, and a few other games as a teen.


As an aside, “plateau” is a French word, and the plural is “plateaux”, so it looks weird to me to write it with a “s”. But in the context of an English sentence, “plateaux” doesn’t feel right either. So it’s no-win situation for me here…


Investing hipsters love to crap on that book because it was really popular for a while, but it’s a great book. It just doesn’t tell the whole story — which isn’t a unique flaw for a book — and only gives you some pieces of the puzzle.