52: Substack Having a Moment, Fred Liu's 164% YTD Letter, Okta CEO Interview, AmazonBasics in India, more Apple ARM, COVID19 Vaccines, Unreal 5 Engine, and the Coolest Thing Intel Made Lately

"DCFs are great at valuing businesses in the optimization phase"

Why do you go away? So that you can come back. So that you can see the place you came from with new eyes and extra colors. And the people there see you differently, too. Coming back to where you started is not the same as never leaving.

—Terry Pratchett

I’m filing this essay for later, when I’ve said something really stupid on this newsletter and I can’t reach in everybody’s email inboxes and edit it out/throw it in the memory hole:

Found it here, where there’s discussion about it.

✰ I guess enough of you guys & gals watched Sentieo’s free research presentation last month, because they are sponsoring a few editions again, starting today. Every time I have a sponsor, I feel just a little bit like a media mogul (but without the dynastic power struggle among my heirs and multiple ex-wives).

Investing & Business

A Word From Our Sponsor

An earnings season like no other: Q3 2020 was at the confluence of the Covid pandemic, unprecedented government actions, and the U.S. presidential elections.

On Thursday, November 19 at 2pm ET join fundamental research platform Sentieo for a live recap of Q3 earnings, including selected Q3 trends across sectors, overall market and sector performance, as well as an update on their election ideas. 

Register here to attend live or to receive the recording after the event.

Ben Thompson & James Allworth on Apple’s ARM Transition, Hardware vs Software Differentiation, Intel, etc

I enjoyed this episode of the Exponent podcast with Ben Thompson and James Allworth. It covers a bunch of stuff I’ve been writing about recently, so the topics will seem familiar to those of you that haven’t skipped over those parts of the recent editions.

Their conversation’s starting point is this recent piece published by James:

The causal mechanism behind disruption that Grove so quickly understood was that even if a disruptive innovation started off as inferior, by virtue of it dramatically expanding the market, it would improve at a far greater rate than the incumbent. It was what enabled Intel (and Microsoft) to win the computing market in the first place: even though personal computers were cheaper, selling something that sat in every home and on every desk ends up funding a lot more R&D spend than selling a few very expensive servers that only existed in server rooms.

Similarly, Apple’s initial foray into chips didn’t produce anything that special in terms of silicon. But it didn’t need to — people were happy to just have a computer that they could keep in their pocket. Apple has gone on to sell a lot of iPhones, and all those sales have funded a lot of R&D. The silicon inside them has kept improving, and improving, and improving.

Finally, today marks the day where, for Intel, those two lines on the graph intersect. Unlike the last time the two lines intersected in the personal computer market, Intel is not the one doing the disrupting. And now, it’s just a matter of time before the performance of ARM-based chips continues its march upmarket into Intel’s last refuge: the server business.


I’ve covered some of that in my past coverage of ARM (and Nvidia, who’s going to accelerate ARM’s roadmap and expand their IP in the data-center, if the acquisition goes through). There’s some of it in edition #39 and some in edition #26, among other places (I wish I had a way to link directly to a sub-section of a past newsletter — Hey Substack? Can you build that feature?).

Substack is Having a Moment (he writes on his Substack…)

Matt Yglesias writes:

I’m looking forward to hitting the ground running at Substack. Today’s my last day at Vox, and my first post is already up. Check it out and enjoy!

Welcome to Slow Boring

Vox co-founder leaves Vox for Substack because he has more freedom there 🤔

I guess getting the product & business model right creates a huge tractor beam pull and lots of stuff happens quickly.

Top journalists/writers have historically been underpaid vs their impact. They just didn't have control over the distribution bottlenecks in the old world.

There will soon be many writers making multi-millions/year that were making $80k-200k before.

Mostly Borrowed Ideas and I had a conversation in this thread about Substack, here are some highlights:

MBI: What if he loses his freedom on substack at some point?

Me: At least there he owns the email list and can just import it to his own self-hosted site running on Ghost or whatever...

But I think the main difference is he's now clearly just writing in his own name and not the name of a large organization. [so doesn’t have to self-censor/follow the org’s editorial line]

Plus it's not a bad deal for him, he probably still owns Vox equity (never looked into that, but guessing) and now will probably have a separate large income stream that isn't correlated with that. Nice, if you can get it.

MBI: Right, I get it. And given customer convenience of subscribing, substack does seem appealing.

But I'm guessing some of these writers will generate >$100K bookings in future, and if they start importing email list to avoid paying >$10k to substack annually...

Me: Big writers like him must have side-deals with Substack and aren't paying the regular fees. I know Substack has been building features just for some of the big names, even offering them perks like health insurance (iirc..?)..

He's a halo customer they're using to attract others

MBI: That's the only way it will make sense for them to stay on substack. Otherwise, substack will lose its best writers on a consistent basis.

The tension will rise as cost of substack is tangible but the benefit (e.g. customer convenience) is intangible (but pretty real).

Me: I feel like Substack's main benefit is ease of use vs self-hosting. Big deal for the 99% of the curve that is right of the power law peak. And the fact that readers know what they're getting and will more easily hit "subscribe" vs an artisanal website where uncertainty is higher.

So far I haven't been too impressed with their development velocity, but I know they've been hiring... If they can step on the gas and keep improving the back end tools and fix the bugs, they'll lose fewer big people to self-hosting.

Hayden Capital: Fred Liu’s Q3 Letter (up a mere 164.1% YTD…)

So far, we have produced +164.1% for our partners year-to-date, while the S&P 500 returned +5.6% and the MSCI World at +1.7%. This quarter’s results bring our annualized returns to +30.0% per year since inception [...]

In the field of decision theory, there’s a paradox called the Ellsberg Paradox (named after Daniel Ellsberg). The theory suggests that people prefer to take risk in situations where the odds are known, rather than a scenario where the odds are unknown – even when the latter scenario has a guarantee of a positive outcome
(it’s just that the magnitude of the outcome is unknown)

Or in other words, people prefer to take risk in situations that have a lower expected value, but where that expected value can be calculated with certainty – rather than where the expected value is guaranteed to be higher, but by how much exactly is unknown. It’s often used to illustrate how people have an aversion to ambiguity. [...]

DCFs are great at valuing businesses in the optimization phase – they already have a product and new product launches are unlikely to materially affect the trajectory of the company. As such, most of the future value creation comes from optimizing the existing product / service (growing number of customers, raising prices on existing customers, streamlining operations to cut costs and improve margins, etc.). These are linear and gradual changes…

By contrast, knowledge-based industries such as technology derive a significant amount of value in the optionality – creating a new product or service using existing internal capabilities, that if successful, will significantly change the trajectory of the company (for example, Apple launching the iPhone created shareholder value multiples the size of Apple’s previous state). And the earlier-stage the company, the larger the embedded real options value drives the valuation of a company, as opposed to the value of current business in its existing form today. [...]

market leading businesses tend to have “first look” at new opportunities, given their resources and position of strength. Market leaders tend to be in more areas and thus collect more information than smaller competitors. Because of this, they have an informational advantage into potentially interesting new fields

Fred’s the right mix of smart and humble and effective. You should follow him on Twitter.

Here’s a link to Hayden’s Q3 letter. Here’s the link to a 34-page separate case study presentation on Afterpay (ASX: APT).

AmazonBasics is Selling Fridges in India

This is kind of a step up from AA batteries…

Looking around the Indian Amazon site, I also saw dishwashers, microwaves, air conditioners, freezers, washing machines…

On the US site, I could find toasters, coffee grinders, ricer cookers, portable induction cooktops, etc. But nothing quite as big.

Source. h/t Ragunath Jayabalakrishnan

Interview: Todd Mckinnon, Okta CEO (Podcast)

Another good episode from Patrick O’Shaughnessy’s special series interviewing founders & operators:

Todd started Okta in 2009 after realizing that enterprises would need a robust solution for identity management in a world where everything was quickly moving to the cloud and today counts over 7,000 enterprises as customers. Our conversation focuses on how Todd decided to leave Salesforce to start Okta, the painful early years of growing the business, how companies can create and define a new market, the different roles he’s had to play as the company grew and went public, and the frameworks he’s put in place to continue to innovate and test new things as public business.

I liked Todd’s answer to the question about what is he seeing ahead, by looking at so many companies that are “building the future” with their tools:

Todd McKinnon: Maybe this is a silicon valley perspective, but we think the world is further along technologically than it is. There’s still a lot of value to be delivered to companies just doing basic things.

We have a product that adds another layer of security by doing multi-factor authentication, so not only passwords, but also checks a 6-digit codes, or sends you a push notification.

For us it seems like, man, doesn’t everybody have that already? Isn’t that on every application? But there’s just so much out there that doesn’t have basic security like that.

The thing that keeps hitting me is how much opportunity there is with basic steps helping companies be more secure… Helping them get online. We live on the internet all the time and think everybody is online, but there’s still many businesses that have a lot of work to do getting online and the potential there is tremendous.

(my own hand-rolled, organic & artisanal transcript, so any errors are mine)

Science & Technology

Interview: ‘Paul Offit, M.D.: An expert perspective on COVID-19 vaccines’

This one just came out this morning, so I haven’t had time to listen to the whole thing, but 49-minutes in, it’s really good, and you can always trust Peter Attia to do a good technical interview on medical matters:

Paul Offit is a pediatrician specializing in infectious diseases and an expert in virology and vaccine development. He currently serves on the FDA committee overseeing and evaluating COVID-19 vaccines. [...] Please note that this episode was recorded 11/05/2020, prior to the 11/09/2020 news release of the Pfizer vaccine update.

While it was recorded pre-Pfizer update, and pre-this-morning’s-Moderna update, they do talk about both of the mRNA vaccines in the part that I heard (and probably more in the second half of the podcast).

The Coolest Thing Intel has Fabbed Recently…

is this sticker. It’s really clever to show the processor architecture on the back.

Plot twist: the sticker is printed using the 10nm process

I should make it clear that I enjoy making a bit of fun of Intel, but it’s still a great company full of amazing engineers, it’s an institution that is very important to the US and the world, and while they’ve been having troubles in recent years because of technical and strategic problems, their problems only stands out as much as they do because they were so dominant for so long.

As I wrote about in the sub-section on Intel and AMD in edition #48:

As a customer, what you want is generally for the big players to be neck and neck and be trading the 1st place prize back and forth. That’s what leads to the biggest generation-over-generation jumps in performance, to the most innovation, to better prices, etc.

Over the years I’ve seen it over and over again in various places (web browsers were never as bad and as slow to evolved as when Internet Explorer finally crushed the competition). We’re not there yet with Intel vs. AMD, but we should probably all hope that Intel can get back on track (and not only for this reason — it’s strategically important to have cutting edge semiconductor manufacturing be dispersed around the world and not be all concentrated in Taiwan and South-Korea).

So I’m hoping that they get back on the horse and become highly competitive again. But that doesn’t mean that I won’t make fun of them once in a while…

Really Cool Demo of Unreal Engine 5 on the Playstation 5

I had seen a similar video last Spring, but it didn’t go as in-depth as this one. Lots of really cool stuff in there.

h/t Rishi Gosalia (⬅️ Denzel)

U.S. Election Probably Screwed Up Your Sleep

I mentioned that I wear an Oura ring sleep-tracker before. They have analytics on all their wearers that they can aggregate to find out interesting patterns. Election night definitely was an outlier!

  1. As a nation, we lost 138,833,045 hours of sleep.

  2. Our hearts were racing, and our average resting heart rates shot up 1.4 bpm—much higher than the predicted average we’d expect for a ‘normal’ Tuesday.

  3. The night was stressful, and we felt it. In the Oura app, there was a sharp 15% increase in the usage of the “Stress” tag and a 9% increase in “Anxiety.” 

  4. The weekend before the election marked the end of daylight savings time, and we were given an extra hour of much needed, and deserved, sleep—but that was short lived.  Even with that extra hour, the nation’s Sleep Scores plummeted on election night to 76 (out of 100).

My own sleep that week. Source.

Real-World Example of Software Using Apple’s M1 Neural Engine Cores

The image editing app that I use, Pixelmator Pro, just pre-announced its 2.0 release in a few days. It'll support Apple's M1 ARM chips, including their neural engine cores. They're seeing a 15x (!) speed improvement on ML tasks vs Intel:

A 15x real-world speed up is legit. Graphic artists and photo editors and amateurs who just like to try to make their stuff look half-way decent will all feel that, and not just a little.

If you’re curious what the 2.0 looks like, they made this short promo video that shows off a lot of the interface and features. I used to use Photoshop, but now I do it all with this and GraphicConverter (another great piece of software, made by a single person).

(they apparently pay attention on Twitter)

Interview: Apple’s Greg Jozwiak and Craig Federighi

While we’re on the topic of the ARM Macs and the M1, may as well go fully down that rabbit hole.

Two Apple VPs did an interview with The Independent, here’s some highlights:

Usually, a major advance in computing performance might add 20 or 30 per cent faster processing speed – but the new [Macs] multiply that number by 10x, with numbers showing that the computers as much as three times more powerful generally and up to 11 times faster at some tasks. [...]

We overshot,” says Federighi. “You have these projects where, sometimes you have a goal and you're like, ‘well, we got close, that was fine’.

“This one, part of what has us all just bouncing off the walls here – just smiling – is that as we brought the pieces together, we're like, ‘this is working better than we even thought it would’.

We started getting back our battery life numbers, and we're like, ‘You're kidding. I thought we had people that knew how to estimate these things’.” [...]

That’s a high-quality problem to have!

that might sound like wild hype, and Federighi knows it. “We are so eager for everyone to get these in their hands and have the same experience that we all had because I think it will speak for itself but we are legitimately positively surprised with the outcome of our work, and really happy we're able to give this to customers,” he says. [...]

Federighi jokes that the the “ordering website was overwhelmed with Apple employees”. “No one is worried about the [version 1.0] of this system.” [...]

I bought a 5k iMac 27” in 2019 (6-core, 32 gigs of RAM).

My last desktop computer before that was a 2008 Mac Pro (8-core, 2.8ghz, 24 gigs of RAM, 1-SSD and 3-hard drives… Expensive up-front, but I used it 11 years, so it ended up being pretty cheap per year thanks to the performance headroom and enterprise-grade durability. It still works fine, btw, I just wanted a 5K P3 wide-color-gamut screen). Anyway, the point I’m trying to make is I tend to keep computers a very long time.

But when they release an ARM 5k iMac, I may sell the 2019 Intel iMac and upgrade to it because it’s just such a leap.

The thing that is really holding these chips back is heat: as you give them more cooling to play with, they become even faster. [...]

If [a fan is] the difference between the MacBook Air and the MacBook Pro – where even in the bigger computer everything is still tightly packed – imagine the kind of performance that might be possible in a computer like the Mac Pro, a large tower with three vast fans. [...]

That’s the thing to understand. Thermals matter a lot. A lot of CPUs and GPUs can peak very high, but then they get throttled. So a quick benchmark and a real-world task that lasts a long time can lead to very different speed profiles.

Even the battle for speed and processing power – surely the one measure on which the bigger computers should be reliably better – was being won by the phones. Apple didn’t show off about it much, presumably because it embarrassed the Mac, but for many tasks its iPhones have been faster than its laptops. And it does all that while remaining cold like an iPhone.

It was funny to see Apple dance around this, never comparing the A13 and A14 to competing laptops too much because that meant indirectly comparing them to their own Intel-based laptops.

That can continue to happen in the future, Apple says. Now that the chips are related, any new technologies can be added into both, and they’ll each improve the other. [...]

All of the different chips are lined up now so that they can be developed together, and so it’s far from ridiculous to speculate that the Mac could now be on the same kind of annual, very aggressively improving release schedule that has marked out the iPhones.

That’s the most interesting part of this to me. The story continues, it’s not a one-time event.

The Arts & History

‘Wow, you love reading books’


VFX Showcase and Dystopian Sci-Fi Mini-Story (4 Minutes)

If you liked it, there’s also this one on a similar theme (it was actually made before the one above, but it’s a bit less elaborate).

This was made by Corridor Digital. They have a lot of other stuff on their channel.

And if you’re curious about how they make these visual effects, they have a second channel with some making of/behind the scenes stuff. Here’s the one for the second robot video.