298: Top Gun, Constellation Software in One Graph, Amazon Robots, Andy Jassy Interview, Snowflake Apps, Anduril, OpenAI vs Minecraft, and Six Feet Under

"ominous music before stuff goes pear-shaped"

Everything that needs to be said has already been said. But since no one was listening, everything must be said again.

—André Gid

🏐🏍🍿 I bought tickets to go see the new Top Gun on Tuesday. My wife hadn't seen the first one, so we watched the first half last night (the rest tonight)...

I can’t believe how young Cruise and Kilmer look in this. When I saw the film as a teen, they seemed so adult to me, and now I’m like “they let kids fly these things?”. Funny how relative that is.

I expect I’ll want to join the Air Force by Wednesday (I know Top Gun is Navy, but I could do without the ships… isn’t the main perk of being a pilot that you can have dinner with your family and sleep in your own bed? Being stuck on a ship negates all that).

🏋️‍♂️ Three questions for you:

1. How good are you at the thing you do best?

2. If you *had* to get much better at it, how would you do it?

3. What is preventing you from putting that plan into practice right now?

🛀 🕷 I’m fascinated by characters and franchises that have a high level of virality and never seem to die (longevity).

My kids have never seen anything that features Spiderman. No films, video games, TV shows. Yet they all light up when Spiderman is mentioned, can recognize him on backpacks, stickers, and toys.

Ok, you may say, Spiderman is an active franchise with a lot of marketing firepower aimed at keeping it relevant and making sure kids know about it.

But how is my oldest kid so familiar with Sonic the Hedgehog? He doesn’t even know what a Sega is, and while there’s been a movie recently, he hasn’t seen it and it’s not exactly the biggest blockbuster or ongoing franchise…

Yet on my fridge, there’s a drawing of a blue Sonic that my kid made at school.. ¯\_(ツ)_/¯

What makes it so that some of these characters never seem to die? It can’t be because it’s easy to create viral characters, because there are very few new ones that seem to be introduced and that own mindspace with my kids.

The Ninja Turtles would probably ring more of a bell with my 8yo than whatever new character was created by Netflix or Disney in the past year, yet he’s never seen anything featuring the TMNT.

Is it a generational thing? Purely the Lindy effect?

I suspect my kid will watch some older teen playing Minecraft on Youtube, and that teen grew up with certain references (from their parents) that they pass on to my kids, and it sticks because my kid considers this teen “cool”. An unbroken chain of cool…

⏳ ⌛️ Sometimes I feel like days are too short to do everything I want to do.

If it was under my control, I think I’d have 30-hour days.

Maybe 2h before noon and 2h after noon, and then 2 more hours at night before bed, after 8 PM. I know some of you would probably add all new hours outside of working hours, but I like what I’m working on!

(Substack added a new poll feature, so let’s try it)

💚 🥃 If you are not a paid supporter yet, I hope this is the edition that makes you go:

“Hey, I think I want to support what he’s doing here.”

Thank you for that!

---

A Word From Our Sponsor: 📈 Revealera 📊

Revealera provides data and insights for investors into hiring trends for 3,500+ public/private companies + technology popularity trends for 500+ SaaS/Cloud Products.

We give investors insights into:

• Job Openings trends: Insights into a company’s growth prospects.

• Technology Popularity Trends: Insights into how widely products like Datadog, AWS, Splunk, etc, are gaining adoption.

• Vendor Sign-ups (Currently Alpha) tracks the # of companies, as well as the specific companies, that have signed up for SaaS products such as Zoom in near real-time.

Visit Revealera.com for a ✨free✨ trial/demo.

---

🏦 💰 Liberty Capital 💳 💴

The ✨ Constellation ✨ graphs to end all graphs

Back in edition #296, I posted a cool graph by Nicoper showing Constellation Software’s acquisition in the first half of 2022.

That was just the beginning!

Nicoper created two more graphs, #1 shows the Constellation of business units (I’ve lost track, but 600+), and #2 shows the company’s diversification by industry/end markets (click on the images to see bigger versions).

Trivia: If you’re wondering about the name of that kind of graph, it’s a Sankey diagram.

One of my favorites is this one by Lawrence Livermore National Labs that shows energy flows in the US.

Amazon showing off some new robots and warehouse technologies 📦📦📦📦 🦾🤖

The green light where the robot is “looking” is a good idea. It makes it much easier for humans to know where it’s going. The eyes are cute, and I guess we’ll anthropomorphize robots forever, because of how our brains are evolved to look for faces in things.

There are a few other videos with more detail on this page.

The machine-vision demo is pretty cool. They’re using 120 FPS high-resolution cameras to scan packages in the hands of employees in a fraction of a second, with no need for a scanner gun or to put the box somewhere special. I’d guess this is Cognex technology, but I’m not sure.

Every time I see one of these public demos, I can’t help but wonder about all the stuff they don’t show yet. I mean, they probably can’t deploy truly sci-fi stuff too widely without having it leak just because they have so many employees, but I’m sure they have really interesting stuff in the lab or in small-scale testing that is more advanced than what we know about.

Interview: Andy Jassy, Amazon CEO

Speaking of Amazon, here’s a recent 26-minute interview with Andy Jassy:

• 🎧 Amazon CEO Andy Jassy (Studio 1.0)

Not a ton that is new in there, but I like getting to know him a bit better.

h/t my friend MBI (💎🐕)

❄️ The puck towards which Snowflake is skating ⛸ 🏒 🥅

Sorry about that headline, I cringe too…

Friend-of-the-show Peter Offringa (💿) wrote a very good, detailed summary of Snowflake’s recent summit where they announced new features and roadmap direction.

There’s one aspect that I found very interesting:

• Support programmability – bring the applications to the data, not the data to the applications.

When customers are ready to utilize a third-party application, they provision it to run in their Snowflake instance with direct access to their data. This “sharing” of application context is critical because it prevents the developer from having access to the customer’s underlying data.

The current model of SaaS applications assumes the SaaS vendor will maintain a copy of the customer’s data. While this architecture has existed for many years, it is inherently wasteful and insecure. The customer has to import their data from disparate SaaS applications through ETL pipelines in order to consolidate it into one place for analysis. This approach creates two copies of the data and a lot of overhead in managing data pipelines.

It just tickles the engineering part of my brain — it’s inherently more efficient and secure!

if the SaaS vendor experiences a data breach, then the customer’s data is exposed. Similarly, if the customer decides to terminate their relationship with a particular SaaS vendor, they have to trust that the vendor deletes all their operational data, including all back-ups (which really isn’t feasible).

Snowflake’s vision is to bring the applications to the data, not the data to the applications. […]

displacing duplicate data storage from SaaS providers should help offset the incremental cost of Snowflake. A new flock of application vendors may emerge with better unit economics by building their software services directly on Snowflake's Data Cloud

Makes a lot of sense!

It remains to be seen how much traction they get, and if their implementation works well in the real world, but they have a good track record of building effective and useful products, so the odds are pretty good.

---

🧪🔬 Liberty Labs 🧬 🔭

🚁 Interview: Anduril CEO and co-founder Brian Schimpf 🪖

Another good interview by Patrick O’Shaughnessy (☘️).

(the above may be the sentence that I’ve written the most in this newsletter 🤔)

Aunduril is a company that works on hard digital + atoms problems in the defense industry, trying to disrupt the more traditional “cost plus & give us a decade to work on this” approach in the space.

• Brian Schimpf - Anduril: Building The Future of Defense

Their main product right now is the Ghost autonomous drone. It’s like a miniature helicopter that can be carried in a suitcase and does surveillance and reconnaissance.

You can see what else they make on their website (they have anti-drone stuff, they even have a submarine drone).

When I was younger, I had that reflexive: “war = bad, so defense industry = bad, weapons tech = bad”. But now that I’m older and, well, older, I now realize that dictators and aggressive regimes around the world aren’t going to stop spending on weapons, so I’d much rather that democracies had overwhelmingly stronger militaries.

It’s a cliché, but the best way to avoid war is often to prepare for it and to deter aggressive acts forcefully (ie. What if there had been a much more painful blowback on Putin when he invaded part of Ukraine in 2014? The current war may never even have happened… How much death, suffering, and destruction could’ve been avoided?).

⛏ OpenAI trained an AI to play Minecraft by watching Youtube videos

Minecraft is open-ended and each map is different because it is procedurally generated, so the AI model can’t just “learn the game” as a sequence of events and actions like it could with linear games.

We trained a neural network to play Minecraft by Video PreTraining (VPT) on a massive unlabeled video dataset of human Minecraft play, while using only a small amount of labeled contractor data. With fine-tuning, our model can learn to craft diamond tools, a task that usually takes proficient humans over 20 minutes (24,000 actions). Our model uses the native human interface of keypresses and mouse movements, making it quite general, and represents a step towards general computer-using agents.

The AI only has the knowledge that a human would have while playing the game, seeing what is on screen and interacting through the regular interface (this is different from in-game AIs that have access to more than a regular player, and so all kinds of shortcuts can be used to make them appear smarter than they really are).

Trained on 70,000 hours of IDM-labeled online video, our behavioral cloning model (the “VPT foundation model”) accomplishes tasks in Minecraft that are nearly impossible to achieve with reinforcement learning from scratch.

More detail here. There are some videos showing the AI’s play progression.

Speaking of OpenAI, Sam Altman gave this update:

Little openai update:

gpt-3, github copilot, and dall-e each have more than 1 million signups!

took gpt-3 ~24 months to get there, copilot i think around 6 months, and dall-e only 2.5 months.

Not only is the size of the models growing exponentially, but the demand for them too!

Dr. Peter Attia on nutrition, disease prevention, sleep, and more 🏋️‍♂️ 🥗🍔 🛏 😴

This is a kind of retrospective, sharing things that he has learned over the past 100 episodes of his podcast, things he’s changed his mind and updated on in the past couple of years, etc.

• 🎧 #202 – Peter on nutrition, disease prevention, sleep, and more — looking back on the last 100 episodes

It’s wide-ranging and a good summary on many topics for those who don’t follow all of the longer full episodes.

Some of the topics covered:

• Cancer: the promise of immunotherapy

• Cancer: how aggressive screening for gastrointestinal cancers could save lives

• Cardiovascular disease: how early and aggressive lowering of apoB could change the course of ASCVD

• Alzheimer’s disease: genes that modify risk associated with the APOE4 variant

• Time-restricted feeding: where the benefit comes from, and when this practice can be problematic

• The common problem of protein underconsumption

• The tremendous impact of exercise on lifespan and healthspan

• Psychedelics: a powerful therapeutic tool in the right setting

• Sleep: updated thoughts on blue light and a remarkable drug for aiding sleep quality

---

🎨 🎭 Liberty Studio 👩‍🎨 🎥

Six Feet Under — That's My Dog 🐕 🥺

*Spoilers for Six Feet Under season 4, episode 5*

*Spoilers for Six Feet Under season 4, episode 5*

*Spoilers for Six Feet Under season 4, episode 5*

I don’t know if the statute of limitations has run out on spoilers for this since it came out 18 years ago, but I was recently thinking about it — and about how I need to re-watch this show at some point.

I think more shows should have a random episode like this, where it starts just like a regular episode, but then things go really off the rails in a totally unexpected way. In this case, a total psychopath kidnaps and robs the main character and toys with him like a cat with a mouse, traumatizing poor David with a near-death experience.

It doesn’t have to be this brutal — in fact, it was hard to watch — but I like a curveball once in a while, especially if the show has a well-established formula. It keeps you on your toes and makes it seem more like real-life, where you don’t necessarily get a big warning sign and ominous music before stuff goes pear-shaped (my friend just almost randomly drowned, which is probably why I’m thinking about this 😮‍💨).