Being (even more) Bayesian

Bayesianism has become my favorite math-idea-that-doesn’t-involve-math. It’s three simple steps. Step 1: have an idea about a thing. Step 2: observe the thing. Step 3: have a new idea based on the observation. (repeat)

There are two tricks to make this work for you. The first is how much to update. Being Bayesian means changing your mind in proportion to the change. Try the expression, “I’m slightly more sympathetic to X,” for example. Saying this acknowledges the new information and massages the ego.

The second trick is where to start (Step one), and we have to start somewhere.

“By not taking advantage of the accumulated knowledge that we have as a scientific community, we are artificially leveling the playing field. We are giving theories with no basis in scientific fact too great of a chance to prove themselves through the data.” – Aubry Clayton, The Conversation, August 2021

Clayton’s context is Covid19, but he touches on a larger point too. How much coordination and decentralized command a system allows.

A decentralized command iterates quickly. From the front lines of fast food to fashion to fights. If an organization wants to move fast, the decentralized command structure works better than coordination.

But while individual agents may be fast, the whole may be slow. Why? No coordination. The scientists in a medical research lab will do more experiments with no oversight or collaboration but they may not make more progress.

Coordination and decentralized command apply to both knowledge and people. Having accurate base rates and priors means coordinating our existing knowledge with the accumulated.


Bayesians even frame things beautifully. It’s not “changing your mind” bur rather it is “updating your beliefs”.

They’re not waiting on a phone call from Paris

One strength of L’Oreal is a focus on cosmetics. Rather than a breadth of products like their competitor Unilever, L’Oreal has a tighter focus. That’s not all they do well according to Aoris Stephen Arnold:

“Another part is how they have organized internally, the company works very hard to have a decentralized model. The people running the Lancome brand in Australia or the people running the Redken brand in Brazil own it and are in control of it. They’re trying to get the benefits of their size and scale but have the flexibility, agility, and entrepreneurial feel of a small company and it is hardwired into their DNA, it is how they have operated for a long time. It has benefitted them as markets change, as they have in the last year, because they aren’t waiting for a phone call from Paris to tell people what to do.” – Aoris Stephen Arnold, Australian Investors Podcast, July 2021

Decentralized command is the idea that problems should be solved by the person closest to them. Sometimes Penn Jillette and his partner Teller are hired for private events. These events are good, but Jillette writes, his agent will warn the group ahead of time. “They’ll get a better show if they let us make all those decisions (what jokes, what volunteers, etc.), which is true and they usually understand that. But if they insist, we give them what they ask for, and they’re happy with that, too…but we know we weren’t as good as we could have been if they had just let us do our jobs.”

Hire people, advised James Mattis, who can “unleash their initiative”.

The decentralized command is not a complicated idea but to see it in an organization as far-ranging as L’Oreal can inspire other organizations. It’s worked on Seinfeld and The Simpsons. It worked for Alice Waters at farm-to-table Chez Panisse and for every-meal-is-the-same-at-McDs founder Ray Kroc. Spotify’s discover weekly came from a decentralized command. So did the iPad keyboard.

The decentralized command approach isn’t a panacea, but for organizations pointed in the right direction it can sure make them better.


A couple other L’Oreal figures. The company accounts for 12% of the beauty market globally, and turns up in 50% of the beauty queries on YouTube. L’Oreal also spends 60% of its advertising dollars online.

The Art of Decentralized Command

“Our whole business is based on giving our artists and designers complete freedom to invent without limits.” Bernard Arnault (HBR)

For a long time, we advocated for a decentralized command. It just made sense that as Ray Kroc wrote, the person closest to the problem should solve the problem. But reading Good Strategy, Bad Strategy reveals a wrinkle in this advice.

Richard Rumelt opens the books with an example from his classes. Rumelt asks his MBA students why Walmart went on to succeed where Kmart did not. The shiny pre-MBAs “are willing to throw anything into the bin, and I don’t stop them,” he writes. There will be boxes and lines and obvious answers, but where was the competition? Why wasn’t this obvious to everyone? 

And early on we get a central idea to Rumelt’s work:

“Whenever an organization succeeds greatly, there is also, at the same time, either blocked or failed competition.” 

Rumelt sees a synthesis between Walmart’s command and logistics. They weren’t just better, they were the only one doing it. It was Zero to One. With large stores connected to integrated supply lines and digital record management, Walmart was the largest buyers (at the lowest prices) and most efficient distributor. 

One part of their strategy fed another which fed the original.

Had Sam Walton let his store managers dictate purchases, those supply and price advantages would have evaporated. However, store managers orchestrated local marketing. Walton encouraged just about anything to bring people to the stores, often a popcorn or ice cream cart was involved. It was okay that individual stores varied in applications that weren’t central to the synergy.

Part of the reason this blog has focused so much on DC is that we write about business in the earlier and smaller stage. For nascent businesses action trumps coordination.

The existing businesses which maintain a more bottom-up structure have a more ‘collection of parts’ vibe. Ted Sarandos said about Netflix’s non-notes: “We’re way better off taking someone’s creative vision and putting it through the service than us trying to go in and retool it. At the end of the day if the creator says, ‘That’s my show.’ we put it up.” 

It doesn’t matter to Netflix that documentaries are shot a certain way or that some subjects are covered while others are not. What data could have, would have, suggested Tiger King? 

When John Galliano created dresses from newspaper, Arnault wasn’t worried. 

“I don’t have alarm bells when it comes to creativity. If you think and act like a typical manager around creative people—with rules, policies, data on customer preferences, and so forth—you will quickly kill their talent. Our whole business is based on giving our artists and designers complete freedom to invent without limits.”

Why didn’t it bother Arnault?  What-if all of LVMH’s 75 brands did this? The don’t. Each of the ‘newspaper dresses’ is an experiment, it’s outcomes are (expensive) limited runs, and the mass-market has some of its DNA. 

Finally we have the (maybe old) difference between Toyota and GM. Known as Andon manufacturing, the Toyota assembly system included a chord anyone could pull when they saw something wrong. GM, did not

There are lots of ways to run a business. Arnault says just this. However for most business, some ways are better than others. Netflix’s hands-off-ness helps creators make great work and share the word about Netflix being hands off, which leads to better creators working with Netflix. This is what Rumelt teaches. That’s the art to it.