Discover more from Liberty’s Highlights
327: Thoughts on AI-Art and Ethics, I'm Interviewed, China Semiconductors, AMD, AMETEK, Coal Mountain, Cloudflare, and Pro-Human Environmentalism
"It’s not getting the faces right, but the ambiance is nice!"
It isn't what people think that's important, but the reason they think what they think.
🚢 🎨🤖 Earlier this week, I installed Stable Diffusion on my Mac Studio.
This is what the text-to-image AI model gave me when I asked to show what a party on a steamboat may look like:
It’s not getting the faces right, but the ambiance is nice!
Here’s one from the blooper reel:
🚨 Don’t skip the Arts & History section today, that’s where the meat of the edition is (on generative AI)
📡🎙 I did a written interview with the great folks at Market Sentiment:
It was fun! Mostly about investing, but I couldn’t help but go on a few tangents…
🐙 Kanjun reminds us:
It always shocks me to remember that we didn’t really have street lights until the 1800s - only 200 years ago! And until then, and for long after, being out at night in cities was quite unsafe. To someone living in 1850, what electricity would enable was unimaginable.
We take it for granted, but electrical lighting was such a game-changer, inside and outside.
How many other things are almost invisible to you, but if they disappeared, your life would be radically transformed?
💚 🥃 Did you know that for the past 2+ years, most weeks I’ve worked on this newsletter project 7 days a week?
On Monday and Tuesday I work on Wednesday’s edition, on Wednesday and Thursday on Friday’s edition, and on Friday, Saturday and Sunday afternoon (whenever I can find some time) on Monday’s edition.
Add to that semi-regular podcasts that may take 1-2 hours to record and 2-3 hours to edit, get transcripts for, and publish with show notes, and other special projects like written interviews, and this steamboat is more than a full-time job.
I love doing it, but I CAN’T keep going without your support so thank you for making it happen! 🤜🤛
Liberty’s Highlights is reader-supported. To support my work, consider becoming a paid supporter. Just CAN’T do it without you ¯\_(ツ)_/¯
A Word From Our Sponsor: ⭐️ Tegus ⭐️
What makes a great investor isn’t pedigree or diplomas 🎓, it’s quality thinking and information 🧠
Tegus can’t provide the thinking for you, but on the information front, they’ve got you covered! 📚
With 25,000+ primary source expert calls covering almost any industry or company you may want to learn about, and the ability to conduct your own calls with experts, if you so choose, at a much lower cost than you’ll find elsewhere (70% lower than the average cost of ˜$400), this is the service to turbo-charge the depth and breadth of your knowledge 📈
Business & Investing
U.S. to further restrict China’s access to semiconductors equipment 🗜🇨🇳
The Commerce Department intends to publish new regulations based on restrictions communicated in letters earlier this year to three U.S. companies -- KLA Corp, Lam Research Corp, and Applied Materials Inc
The letters, which the companies publicly acknowledged, forbade them from exporting chipmaking equipment to Chinese factories that produce advanced semiconductors with sub-14 nanometer processes unless the sellers obtain Commerce Department licenses.
This is both seriously going to screw with China’s ability to keep pushing new nodes or even make new fabs for existing nodes AND force them to accelerate their own homegrown versions of these technologies.
The U.S. government seems to be making the bet that they’d rather have a slower, if less dependent, than a faster, but more dependent China….
🎧 Jay Goldberg: Semiconductors & AMD 101 (Podcast)
Speaking of semis… Nominally, this one is about one company, but it’s really a history of the semiconductors industry (80%) with a special focus on AMD (20%).
It touches on a lot of the important factors that have driven things, such as the era of integrated players, merchant silicon, fabless, general computing vs GPUs/ASICs/FPGAs/etc, the capex required to push forward Moore’s Law, Intel and AMD’s non-overlapping execution problem eras, etc.
Good stuff, Jay and Patrick did a great job 👍 AMD 0.00%↑
⛅️ Cloudflare Q2 Detail 🌏
Geo-Segment Growth Q2-21 Q3-21 Q4-21 Q1-22 Q2-22 USA +65% +53% +52% +56% +55% EMEA +65% +62% +60% +57% +54% APAC +23% +25% +29% +31% 👉+45%👈
Another ongoing worry of mine has been how poorly APAC has performed. They have been stating that its growth rates would be catching up to the other segments, and this Q continued the acceleration, rising +12pp! This is very good to see.
I’ve long wondered why APAC was so far behind, and thought that maybe the new office that they launched in Singapore a little while ago had something to do with energizing their efforts in the region.
I hope what we’re seeing now is that bet paying off. NET 0.00%↑
Overview of AMETEK 🔩⚙️
Years ago, I studied this company pretty closely, but I hadn’t been paying much attention since.
The Compounding Capital Review did a good job both of looking back (far back to the 1930s!) and bringing me up to date on this serial acquirer conglomerate:
The Coal Mountain 🏔
Coal looks like a lot of stock charts for SaaS stocks in the past couple years…
Alec Stapp: “Imagine predicting in 2007 what would happen to the US coal industry over the next 15 years”
Science & Technology
Michael Shellenberger: Pro-Human, Pro-Abundance, Pro-Nuclear Environmentalism
I enjoyed this interview, both for its context about the past, telling the story of what actually happened to nuclear power, and a vision of the future that can replace the de facto misanthropic one that has taken a lot of space in the public discourse in recent decades.
‘Why AI is not a Moat’ ✋🤖 🚫
Wow there’s a lot of AI/ML stuff in this edition (don’t miss what’s in the Arts section below)… So why not more?
Friend-of-the-show Shomik has a good piece about what it takes for AI to be a moat, and how to approach the tools and techniques from a startup’s POV:
The Arts & History
👩🎨🖼 More Thoughts on AI-Generated Art 🖼 🎨🤖
This is a follow-up to edition #326.
First, I want to state up front that I believe that this unfolding revolution will create both winners and losers, and will be a rough time for many illustrators, photographers, and visual artists.
Supply & demand is a powerful force, and if you increase supply much faster than demand, prices tend to fall.
But today I want to discuss a different dynamic:
There are a lot more people in the world who have interesting visual ideas than there are people who can *execute* those ideas with a paintbrush in hand (so to speak — whatever the medium, digital or analog).
The ratio may be 10,000-to-1, I have no idea, but it’s hugely lopsided.
If, like me, you believe that art is an extremely valuable thing for humanity, then having more people create it, seeing more variety of ideas executed… That’s very good!
I’m defining “art” a certain way here. I don’t want a semantics debate. If you define art as “something a human made to express inner feelings”, then yeah, none of this AI stuff is art.
But if you define art more like “something that makes you feel something, or go 😲 or 😳 or 🤯 or or 🤔 or 🧐 or 🤨”, then a lot of this stuff qualifies.
I have a folder full of AI-made images, many of which I’d consider masterpieces if I had randomly seen them in a museum with a little plaque under them…
I find that some of them are very *original*. Things I had never seen before (that’s the cue for someone to go “akhtually, this underground DeviantArt account did something similar in 2009…”).
I mean, I haven’t seen everything, everywhere, from all eras (who has?), and few things are truly **100%** original. But on the spectrum of originality, I’m seeing a lot of high-scoring pieces come from this nascent scene…
🤔 Ethical Questions on AI Art ⚖️ 🧐
This is important.
I keep seeing objections like these being raised these days:
Here are three key ethics questions around this new software explosion, courtesy of Baio:
"Is it ethical to train an AI on a huge corpus of copyrighted creative work, without permission or attribution?"
"Is it ethical to allow people to generate new work in the styles of the photographers, illustrators, and designers without compensating them?"
"Is it ethical to charge money for that service, built on the work of others?"
I think there are two aspects, one technical and one philosophical.
The AI models don’t contain the images that they trained on.
They “learn” from billions and billions of images, petabytes and petabytes of data, but the resulting model may only be on the order of gigabytes in size.
What’s distilled in it are mathematical relationships and atomized characteristics of various aspects (edges and color palettes and geometrical relationships and who knows what else), but it’s a one-way relationship.
Once you’ve made the whisky, you can’t get the barley out of it anymore. 🥃
Do we apply these standards to human artists?
Can artists learn from the copyrighted work of others?
Can they make new works using techniques and ideas that they learned from these works (but without directly copying them)?
Can they learn from the style of other artists and make new works in the styles of these artists?
Seems to me like we allow all of this, and the world would be much worse if artists couldn’t do anything based on techniques, styles, and ideas that they saw elsewhere (in fact, most couldn’t do *anything* under those rules, as most creativity is remixing and standing on the shoulders of giants).
Most musicians start out imitating and doing covers, it’s how we learn and find our own style…