This week's case is free, and about a related hotel family: the Pritzkers of Chicago, who to this day largely own Hyatt Hotels. In reality, the story of the Pritzkers is about capital allocation and effectuation on a grand scale.
The capital allocation piece is easier to understand. Jay Pritzker was a particularly good deal maker, and acquired a whole bunch of under-valued companies. He then tossed them to his brothers — Robert, who was an engineer, and Donald who was a lawyer.
Robert took the various manufacturing companies that Jay bought and fixed them, putting them together in a conglomerate by the name of Marmon Group. This is the more straightforward capital allocation + operational excellence story that we've seen before. Eventually, as they got a feel for what could be done with the various companies they bought, Jay only acquired companies that Robert thought would be fun to run (or fun to turn around).
The effectuation bit comes from the hotel side of the house. In 1957, Jay Pritzker stumbled onto a remarkably well run airport hotel by the name of Hyatt House just outside LAX. He asked to meet the owner — Hyatt von Dehn — and asked for some operating numbers. He then made an offer. von Dehn called back a few days later and agreed to sell him the hotel.
Reading between the lines, I think Jay's original idea was that "hey, this is a well run airport hotel, and it's making money — let's expand on the concept and see how it goes".
So by 1961, just four years later, Hyatt was a chain of six airport hotels. And then Donald graduated from law school and took it over.
Donald made a huge bet in 1967. He took over an unfinished hotel with a radical architectural design in Atlanta that year. The centrepiece of the architectural design was the atrium, conceived by architect John Portman Jr; the attraction of this particular opportunity was that all the financing had been lined up. The problem? Every other hotel company had passed.
Hyatt was a bit player in the hotel industry at the time; they were not the first hoteliers Portman Jr had approached. Conrad Hilton (of Hilton hotels) dismissed the building with "that concrete monster will never fly." And there was good reason for this: after all, an atrium design meant that you were building an empty building — instead of filling it with rooms, you carved out this vast, cavernous void — albeit one that made the lobby feel grand and drew the eye upwards.
Donald had no idea if it would work. But the financing had already been lined up, the construction was under way ... it seemed worthwhile to try.
It worked.
Over the next few years, just about every hotel that Hyatt opened was built around this atrium concept, which drew business and boosted occupancy. It became so popular that for a while, every hotel chain — both in the United States and internationally — copied it. The atrium was what made Hyatt the hotel company it is today.
The Hyatt Regency San Francisco, another atrium hotel from that initial expansion.
I read a story like this and think to myself — this is only possible through effectuation, through taking action to generate information. First of all, there's no way I would have sat in a lovely hotel cafeteria the way Jay did and thought to myself, "Huh, I should buy this." There was also no way Jay could have known that Hyatt would become a global hotel company; he bought the first Hyatt to build nice airport hotels. Finally, there was no way Donald could have known that the atrium concept would be the thing that propelled Hyatt into the big leagues. In fact, he left town on the opening of the Atlanta hotel because he wasn't sure the concept would work.
This is what entrepreneurship looks like. We've talked before about effectuation — this idea that the best entrepreneurs take lots of affordable loss bets, improvising into the future, embracing what cannot be known through analysis alone. They start from who they are, what they have, what they know, and who they know, and then they:
take bets with affordable loss
build partnerships to decrease the probability of said loss
exploit contingencies — that is, unexpected good things (or bad things!) — that happen over the course of execution.
By definition, contingencies cannot be predicted in advance.
The Atlanta hotel was was certainly one such bet. And so this style of acting is what differentiates good managers, in existing companies, from entrepreneurs who expand the space of what is possible. It's just rare that you get to see both modes — managerial excellence with Robert, and entrepreneurial excellence with Donald (and I guess investing ability with Jay) — all in the same family, within the same generation. As a result the Pritzkers are one of the wealthiest families in America today.
💡 The Commoncog Membership isfor the discerning businessperson. Get full access to paywalled articles, curated events, a rich and growing case library, and an exclusive, members-only forum.
The Commoncog members-only forum is a private place for sensemaking on business and markets.
Here are a couple of members-only discussions I'd like to draw attention to:
Are we in a recession? — astute readers would have noticed the undercurrent of 'how you deal with a recession?' — running through many of Commoncog's latest cases. In this thread, I'm asking the forum to make sense of whether we're seeing a recession across the multiple countries that the Commoncog membership lives in.
Thinking Clearly as a Form of Expertise — one of the best threads last week. Members debate whether clear thinking is a domain-independent skill or a form of expertise that requires deep contextual knowledge, with one member sharing research showing that the ability to think critically depends on specific knowledge rather than general reasoning skills.
Sensemaking the AI Security Thing — four different frames about the OpenAI Hugging Face incident. Probably the most interesting frame here is from Steven Sinofsky, which implicitly accepts that AI alignment will never happen, and design zero trust execution environments accordingly.
AI experience report: writing ability improvement — A member shares their year-long journey of writing a 150,000-word interactive fiction game, learning the skill of fiction writing in the process. (I vouch for this. I read bits of it as the member was working on the project over the past few months, and their writing really did improve.)
More excerpts from The Everywhere Millionaire. Founders can build transferable effectuation skills within a domain, using the example of a tanning salon owner who pivoted to waxing salons when health concerns shifted consumer tastes.
A member recommends Rick Rubin's book The Creative Act — this is in the context of how Rick Rubin actually has genuine expertise, despite AI bros portraying him as a zero-skill vibe producer.
But more importantly, adaptive expertise is not a skill, no more than critical thinking is a skill, or creative thinking is a skill, or problem solving is a skill:
As I have contemplated how to adapt TDGs, I have been thinking about the role of the map, as it is so central to the experience. So central that even if a domain does not happen in a specific geography, we still try to find a proxy for the map.
But why exactly? What makes this particular training technique so dependent on maps when other training techniques (e.g., Deliberate Practice) do not require them? And what is a map anyways? These are important questions if we want to train expertise in new domains.
As I was pondering these things, I happened to be in a meeting where Gary Klein mentioned something offhand that I should have known, but hadn’t truly internalized. Gary’s offhand remark was this: there is no general decision-making skill, nor a general situational awareness skill. There is no way to train your general ability to make decisions or become situationally aware.
That may sound extreme to some, but as I dug into that claim I realized that Gary’s claim was actually far too narrow. The literature is pretty clear that there is also no critical thinking skill, no creative thinking skill, and no problem solving skill. None of these are skills.
This line of reasoning led to a cascade of insights that started with ‘adaptive expertise is not a skill, it is an uncommon common sense’ and ended with ‘and that is why you need a map’.
Berkshire's ‘hidden’ mega-business: What is Marmon Holdings? (archive.is link) — In 2008, the Pritzkers sold Marmon — the industrials holding company that Robert and Jay built from scratch — to Warren Buffett, as a solution to the fourth generation's inheritance dispute. This article covers what Marmon does, from the perspective of a Berkshire subsidiary.
Confessions of an Unrepentant Slob Snob — This was linked in the Commoncog forums a couple of weeks back, but I didn't read it because I was heads down with the QA for the upcoming launch of Commoncog audio cases. Anyway, I think the valuable thing here is that Charity Majors outlines a framework for why AI-generated writing pisses people off (versus when it's acceptable).
Note, however, that this piece is written to resonate with readers, and that's why it has spread so successfully. A member comments that you can actually just get caught up with the resonance ("yes! y'all are idiots for using AI writing!") and conclude the wrong thing.
(That specific member is actually saying something a lot more subtle, but I'm not going to repeat their point here — click through and read the full thing!)
There are actually two possible interpretations of the ideas in this essay. The first frame is: "this is why sending me AI-generated prose will make me freaking mad, and if you don't realise this, everybody should judge you harshly." The second frame is that if you're experimenting with AI-generated writing, you can actually experiment even more aggressively now, because there's a framework that explicates the guardrails for when you should and shouldn't use it.
I think that second frame is actually the more interesting one. Reading Charity Majors' piece through that lens should encourage you to try to do more with AI-generated writing, not less.
As mentioned earlier, I've been heads down doing QA for the audio version of the entire Commoncog Case Library, along with the rest of my team. If you're a member, you'll be able to listen to any case in the library as a private, custom feed in your own podcast player. We're nearing launch now, so expect an announcement next week.
Working on this project really ... drives home the fact that reality has a lot of detail. Generating audio using AI seems straightforward — the models available are incredibly good; the specific models we use are amongst the most expensive on the market, and therefore we get some of the most realistic voices you can imagine. All the AI audio model providers are trivially easy to get started with.
Making sure all the cases sound right, though, is another matter entirely. We're talking about over 100,000 words across more than 100 cases here. When you're dealing with that volume, it turns out there's quite a sizable need for human QA to ensure that there are no audio artefacts, there are no mispronunciations with foreign names, and there are no weird voice inflections.
Maybe some day we'll write up the details of the system. Though it's likely most folks aren't interested in those details!
I'll see you next week.
Warmly,
Cedric
⛪️ The Commoncog Membership: "Like church for business nerds" Get full access to paywalled articles, curated events, a rich and growing case library, and an exclusive, members-only forum.