Similar to Shoe Dog ā and different in the way that Nike is different from Trader Joeās.
Three things I want to take away from this book:
Joe was incredibly structured in how he thought about problems. He wanted to have a retail store where he paid people well which required him to have goods that had a high price per amount of space they took up. He was willing to cycle through lots of weird ideas (including things like gun ammo) as long as they met this criteria.
Discontinuities. Trader Joeās would specifically target little edges in product categories. As an example, they would become experts in the regulations for say cheese or butter to build limited edition products. In particular they did this on the product side, carefully understanding product categories, and on the regulatory side, carefully reading the fine print to find edges that others didnāt have. An example of a discontinuity is being willing to sell coffee in non standard container sizes or for a limited period of time.
The concept of double entry retailing, which is another way of saying that decisions are interconnected. As an example, paying people more reduces shrinkage.
I didnāt know whether or not to laugh or cry when he said that Trader Joeās target customer is overeducated and underpaid.
This book helped me better understand how retail, goods, and media are interconnected. The transition from network tv to cable tv happens at the same time as Trader Joeās is shifting away from homogenized consumer packaged goods to the more varied assortment we see today. A similar version of this happened with Facebook and the DTC brands of the 2010s.
My guilty pleasure on YouTube right now are videos claiming Ancient Egyptians had access to advanced technology that allowed them to machine vases out of hard stone. Iām agnostic as to whether or not this is true, but I canāt look away! A second thing that makes these videos delightful is that they all pit themselves against mainstream archeology which just cracks me up. Who are these mainstream archeologists? What are they doing to hinder this message? I see the evidence for advanced manufacturing but these mainstream archeologists seem like a mythical species.
Why is Switzerland so rich? This is good, but I think it misses a couple of things. First, Switzerland was spared the physical and human losses of both World Wars. Second, thereās a cultural element that the post doesnāt speak to. Switzerland is both highly individualistic and highly communal, a mix of live-and-let-live and weāre-all-in-this-together that I believe allows it to make more pragmatic decisions, the benefits of which compound over time.
Someone told me this week that in France they say that there are six reasons someone will pay for something: Security, Pride, Novelty, Comfort, Money, Friendliness.
Dead Framework Theory - the idea that LLMs are freezing frameworks like React into the internet. I thought like this at first, but I no longer think that this is true and I actually think LLMs will make it easier to bootstrap new frameworks provided those frameworks have real advantages over what theyāre replacing because LLMs make it so much easier to adopt new tools.
The actual question was much funnier. My 5 year old made a piggy bank at church, causing my 3 year old to ask, āDaddy, do pigs have banks?ā As I think about this, it gets even more puzzling, because I'm not sure he's ever been to a bank.↩
Person Do Thing is on Amazon. Youāre here so you know Uri already, but Iāll just say that my family loves this one and that it makes a great gift for the person in your life that loves games.
The universe as an evolving organism. I have no idea whether or not this is true, but I really enjoy this style of conversation about black holes and space and what we know and what we donāt. There should be more of this.
"Life is 10 per cent what you make it, and 90 per cent how you take it" āIrving Berlin. Sometimes I think the quotes at the end of The Browser our aimed directly at me. I promise you my kids will grow up with this one memorized.
A little bit of SSP
I was on the Demystifying Cosmetics Podcast talking about what I learned making high performance biomaterials from kelp.
If thatās not enough, weāre doing a live show of --dangerously-skip-permissions on Friday at 2 pm ET. Come and hang out.
(I have to be the only person putting out a podcast on beauty ingredients and coding with AI the same week)
The implication behind all of this is that slop is the fault of the LLMs or their creators.
But for a moment, I want to ask you to considerā¦ is this true?
I donāt dispute that weāre seeing a lot more AI generated slop than we were 2 years agoā¦ but I do wonder what itās displacing. When I see this AI video of a woman jumping through a glass bridge, I wonder what human content it is replacing in those users feeds. What is the quality of that content? Would it qualify as slop too?
Here is what I believe to be true.
Making something excellent takes care and focus. Sometimes that care and focus is the years you spent prior to the morning you make something excellent and the final thing hops out of you almost fully formed. Sometimes that care and focus is the years you spend refining something until it is excellent.
Most creative work, including mine, isnāt particularly high quality. If you want to be uncharitable, you could call this slop.1 Internet platforms have made it easier for people to create and display their work. LLMs have made it easier to create. This means that we see a lot more slop.
There will be some Sora posts that will be funny, wonderful, even beautiful. There will be many that are slop.
Mike Judge has a great piece poking at the AI hype where he asks essentially, āIf these tools are so great, where is the explosion of AI created stuff in the world?ā
The whole piece is worth a read, but one of the most interesting things to me about it is the data he brings to bear on the question.
He looks at:
iOS app releases
Android releases
Domain registrations
Steam releases
Public GitHub repos created
And then concludes from these that AI coding tools are ābullshitā ending with the call for people who claim that they are now 10x software engineers because of AI, to show the receipts.
First, I want to concentrate on what I love about this. āIf this is so great, where is it in the data?ā is absolutely the right question to be asking.
And there is definitely a dog that isnāt barking here. The data that he cites arenāt perfect (more on this in a second) and yet really impactful things tend to move really obvious metrics. The gains in life expectancy between 1870 and 1970 are really easy to see.
So on one hand, I love this challenge. On the other, I think he goes too far in calling it all bullshit and saying that it doesnāt work.
Iāll hold myself out as the example that Mike asks for. While I wonāt claim to be a 10x engineer, I had never completed a meaningful software project in production before GPT-4 launched. Since then, Iāve coded this blog / portfolio site myself, launched an AI based local news site that has hundreds of weekly readers, and I have a third unreleased prototype that I think could be a real product.
Analyzing myself against Mikeās charts:
I havenāt launched an iOS app
I havenāt launched an Android app
I have bought 3 domains
I havenāt launched a steam app
Iāve created 1 public repo, unrelated to my AI coding work
Now Iām open to the idea that Iām the exception rather than the ruleā¦ but I also too humble to think that Iām a unicorn on this dimension.
Thereās plenty of room for middle ground here. Itās totally possible that:
1. AI tools are net negative for most software engineers
2. AI tools are transformative for people like me
3. People like me are a minority
Intuitively, I doubt that this is true and yet I donāt have hard data beyond my personal experience to bring to bear on this question. Itās certainly something Iāll be thinking about over the coming months. A more likely explanation in my view is that we havenāt unlocked the right combination of values, tactics, organizational design, and training to unlock AI software productivity at scaleā¦ but I canāt prove that at this point.