Streaming, Sharing, Stealing

Supported by Greenhaven Road Capital, finding value off the beaten path.

Michael Smith and Rahul Telang of Carnegie Melon University wrote an interesting book, Streaming, Sharing, Stealing: Big Data and the Future of Entertainment.

Smith and Telang wrote this book because of a “perfect storm” in the economic entertainment ecosystem. Traditionally a few firms dominated because the best business model for the last hundred years was a return to scale. The value was in finding/attracting talent, creating a market, and distributing the product.


This model worked. This model was logical. This model was successful.

Before 1995, the search for talent “was unscientific at best” and size helped to seek talent, sell it, and ship it. The Firms did this well and movies are one example. ‘Windowing’ is the process of a theatrical release followed by DVD sales followed by digital sales followed by subscription streaming followed by cable and broadcast rights.

Each stage of price discrimination maximizes revenue. People who value movies highly see them early, people who value movies less seem them later. Streaming, Sharing, Stealing is a book now because technology has evolved and the old windows are closing.

Blockbuster was founded in 1985 and The Little Mermaid began the ‘second golden age of animation’ in 1989. Shawn Fanning created Napster in 1999. Each of these was a locally logical decision at the time. It made sense to start Blockbuster, it made sense to make The Little Mermaid, it made sense to start Napster. Each is like a mile marker on the highway of entertainment history.

Smith and Telang note that the Netflix/HBO/Amazon/Spotify/etc. state of today was a sort of perfect storm. We should be wary of simple stories like Napster ruined everything, and be more like Sanjay Bakshi and approach things with a ‘part of the reason is this’ mentality. Smith and Telang propose three things.

First is broadband internet, “It became harder to delay a product in one channel versus another channel because people found it easier to infringe.” Gatekeepers Firms had to control distribution so they could control the windowing of theater to dvd to rental and so on.

Second is digitization. “Now you had technology change happening that eroded the role of a gatekeeper at these firms.” Casey Neistat is the vlogging example. E.L. James is the writing example. The digitiziation pendulum might have swung too far:

Our third change is the rise of platforms, because, “they didn’t have limited shelf space.” Frank Underwood kills a dog yet Netflix lives on because there’s always something else on.

At this point we should pause for a moment and recall something Andrew Ng said. Ng told the story about the early days of the internet and what it meant to be an internet business. It’s the difference between being a tourist and being a resident. “What defines an Internet company is whether or not you have an architect at your organization to leverage internet capabilities to do the things that the internet does really well.”

What things does an internet company do? Internet companies, Ng said, AB test, push decisions up (not down), and have short product cycles. We’ll see a Spotify example shortly.

Michael Smith is emphatic about this because it’s a perpetual point in his presentations. The genius for House of Cards was not knowing that Kevin Spacey, David Fincher, and Robin Wright were good at their jobs.

“The advantage Netflix got was from using its data was to do something that no traditional broadcast network could replicate. Netflix’s advantage didn’t come from knowing how many fans of Kevin Spacey were in the audience. Netflix’s advantage came from knowing exactly who they were as individuals and promote content to them directly based on their individual preferences.”

That’s what it means to be a digital entertainment company.

Netflix made nine trailers for House of Cards; some for Fincher fans, some for (former?) Kevin Spacey fans, some for fans of shows with strong female leads. None were for dog fans, dog fans won’t like House of Cards.

Smith recalled,  “Michael Eisner said that if he had tried to include a similarly violent scene in an episode of broadcast television he’d be out in ten minutes.”

Why could Netflix show House of Cards but ABC could not? Unlimited shelf space. ABC has a shelf space of one. This point is made by Ben Thompson at Stratechery too, pointing out the Netflix advantage when they signed a deal with Starz. Primetime evidence exists too, as we see shows about doctors, law enforcement, families, and more doctors, more lawyers, and more families.

Netflix serves niches.

Music is both like movies and not like movies. Hal Varian noted that “music is something you want to hear over and over again. Movies you want to see on an occasional basis with one exception, kids movies.”

They’re also different in consumer preferences. For music, live is the premium experience. For film, premium means polished. No one pays to see Daniel Craig run across a stage but many love to attend Phish shows just to be surprised.

Knowing that music and movies are different and knowing that Gatekeeper Firms and technology companies are different we’ll look at Spotify’s “discover weekly (Twitter)”.

Former Spotify engineer Chris Johnson said, “Discover Weekly wasn’t handed down from the top. There are news articles about how this is the Apple Music killer and we’ve been planning it for months. No. This is purely an organic bottom-up feature that a small team of four engineers worked on.”

In his talk, Johnson explains the iteration and small steps and, in hindsight, the obviousness of it. Their ‘that’s interesting’ moment was when “We noticed users spending more time on editorial playlists than they were on their personalized discover playlists. What’s going on here?” Didn’t people leave the radio DJ for Spotify?

It was 2014 and Spotify created the first personalized playlist. “It turned out that a month later millions of users are still listening to their personalized Play it Forward playlist. We’re like, ‘Holy crap, what’s going on here?’ We did not expect that.”

That’s when Johnson and his team realized, “The same content as the Discover Page but…a playlist.” People wanted, expected, used playlists more than pages. They tested and tweaked. They released internal versions. They changed the cover art of the playlist.

Then Apple pushed the U2 album onto iPhones. Uh oh. That’s kinda what Spotify was doing, pushing playlists onto the app. Johnson included a survey to Google Docs asking for feedback. Only three people out of a thousand brought it up so they kept building. “Users know best. AB test everything.”

Another part-of-the-reason Discover Weekly works is that it’s human curated. Not one person editorially but a person sporadically. “All the models we’re building are built using human data, on what users are streaming or curating. We’re just mining through the human element.”

In another talk, Mathew Ogle said “Every song on Discover Weekly is there because at least one other person on Spotify added it to their playlist. It’s not there because a machine thought it sounded good.”

Using Ng’s definition, Spotify is definitely a digital music company. Like Netflix, they’ve made it easy to pay for and listen to music and have provided added value using data and AB testing to the users.

But Spotify has other problems – “For Spotify the royalty payments also crush its gross margins.” – that Netflix does not. Marc Andreessen asked Ted Sarandos about sports and Ted said, “It’s okay for us not to be doing everything…The reason I don’t get tempted by major league sports is that the pricing power all belongs to the leagues.”

Netflix and Spotify exist because consumer preferences changed. In How Music Got Free we see part of the story about how this happened. For example, CD’s were over-engineered. “Where the (CD) sales literature promised ‘Perfect Sound Forever,’ (Dieter) Seitzer saw a maximalist repository of irrelevant information, most of which was ignored by the human ear. He knew that most of the data from a compact disc could be discarded—the human auditory system was already doing it.”

And the industry dominated distribution. “Consolidation in the radio industry also helped, creating a homogenous nationwide listening environment that could propel an album to a Platinum status almost instantly on the basis of a single hit. Controlling the airwaves was critical—if Limp Bizkit could go forty times platinum, then literally anyone could.”

In 1996 the compact disc retailed for $17, cost $1 to manufacture, and had approximately 1.8 good songs. Music was ripe for Christensen’s disruption. It was true that CD’s were easier to buy, sounded better, and had more ways to listen. But the MP3 was better in certain ways and the shortfalls could be improved.

It was technological advances that led to 1,000 songs in your pocket.

It was analytical advances that led to Discover Weekly.

It feels like circa 2019 there’s more room to improve analytically than technologically. Netflix uses analytics for strategy but not scripts. Ted Sarandos, “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.”

Smith quotes Frank Underwood in House of Cards, “Power is a lot like real estate. The closer you are to the source, the higher your property value.” Netflix is darn close to the data source.

This approach is appropriate because Netflix has unlimited shelf space. Smith notes that they can invest in properties with niche audiences like those for Woody Allen, Adam Sandler, and Arrested Development. Netflix wants to offer one thing you love at any given moment. ABC wants to offer something most like at one particular time. “Netflix didn’t face the same scarcity that traditional broadcast networks do. In a traditional broadcast network, you can only deliver one show at a time so that show has to appeal to as many viewers as possible. But a Netflix subscriber who was offended by Kevin Spacey’s actions could choose from thousands of other shows on Netflix.”

“Netflix doesn’t have to find more viewers to watch the individual programs. Netflix has to find more programs that appeal to individual viewers.”

The stakeholders matter too. “Netflix wasn’t pursuing an advertising-supported business model so it didn’t have to worry about offending advertisers by including a potentially controversial scene.”

This trio of broad band, digitization, and platforms is the media subset of what Michael Munger writes about in Tomorrow 3.0. In the book, Munger explains the three T’s of transaction costs that prevent economic exchange; triangulation, transfer, and trust.

In other words, I’ve got to find what I want, I’ve got to get what I want (and pay the person for it), and I’ve got to trust the other person. That’s what the digital behemoths have done. Netflix make it easier to find what we want – even though we didn’t know we wanted it – because they make nine different trailers for a tv show Netflix is able to get us what we want legally by offering an automatic payment system. Netflix creates trust that they’ll provide something because they have so much.

Teaching this is fun, says Michael Smith. Students are digital natives and they get this.

“With higher education for example, we have nothing to worry about when it comes to technology changing our business and our power. The students do exactly what you all just did, they nod their heads then go ‘Wait a second!’.”

How so? “There’s a lot of parallels between what happened in the entertainment industry and what might happen in higher education.”

A transition from gatekeepers who succeeded thanks to the collection of talent, the creation of markets, and the distribution of content to digital companies with platforms and unlimited shelf space. Austen Allred has one theory of education. Our XMBA post is a looser approach.

Digitization changed some parts of the economy more than others. Information goods like newspapers, movies, and music all became non-rival as copies were cheap to make and share. Thanks also to network effects, ideas went viral and sharing was easy.

Ben Carlson observed this:

“Why 45 million ppl watched Bird Box on Netflix: holidays & ppl need entertainment, slow TV season before new episodes/shows come out in Jan, no good new movies released. So ppl basically had no other choice but to watch this I Am Legend/World War Z/Quiet Place reboot.”

Entertainment (and the economy) isn’t better or worse, only different. It’s like a hike where the terrain has changed but the direction towards good enough has not.

Sources: Smith and Telang gave a very nice talk at Google (, the book goes into more detail ( and this lecture at UC Irvine about teaching this ideas also has some nuggets (

For Spotify, Mathew Ogle spoke in 2015 about music discovery ( Chris Johnson gave a talk in 2016 in Austin about Spotify’s Discovery Weekly history that was insightful (

Greg Lindsay

Supported by Greenhaven Road Capital, finding value off the beaten path.

Greg Lindsay’s talk, Innovation Doesn’t Happen at Your Desk is an attempt to get people to work together and better. I think he’d agree with Rory Sutherland in that many of the logical questions have been answered and we need to get a bit weird.

A more positive term for weird is serendipity.

Lindsay tells of a natural experiment where a building needed asbestos treatment. The  teams there had their offices shuffled about, and when their (innovation) results were measured, “Randomness was the best strategy for increasing innovation for these teams.”

Put different people near each other, let them talk, and watch the good ideas flow.

Part of the explanation is The Allen Curve, an observation that people talk more to those nearby, even when digital tools allow us to talk to anyone anywhere. Dr. Thomas Allen said, “Broadband communication isn’t a substitute for face to face.” That is, people communicate with those they see.

If proximity leads to communication then communication leads to serendipity. Examples include the Nike Waffle Shoe, Spotify’s Discover Weekly Playlist, and Reed Hastings (who got lucky serving coffee to a certain computer lab).

Lindsay said, “All three major types of artificial sweeteners came from lab accidents where the researchers then picked up their lab bench and stuck their fingers in their mouths.”

Jenna Fischer agrees with action as a solution, “…you’re much more likely to be in the right place at the right time if you’re busy doing showcases, plays, and taking classes. Chances are you won’t be in the right place at the right time if you’re spending your days eating Lucky Charms your couch. Trust me, I tried.”

There are four “black boxes” for serendipity Lindsay said:

  1. You
  2. The office
  3. The city
  4. The network


Building on ideas from James Lollies and Joey Ito, Lindsay said, “You know the classical line, the true sound of scientific discovery is not Eureka but ‘that’s interesting’.” And “Do you have the latitude to chase something?” It takes an internal curiosity and external circumstances.

That’s interesting moments surround us. Peter Rahal wondered why a Crossfit sold t-shirts but not protein bars. Eric Maddox wondered why certain people were still hanging around an area. Brian Koppelman wondered why Americans celebrate entrepreneurs. Jim O’Shaughnessy wondered if personalities weren’t the perfect proxy for well-performing investments.


Cubicle offices are too much like tree farms, orderly and great for optimizing harvests. Another way to approach Lindsay’s idea is via Christensen’s disruption model; optimized offices are best for sustaining innovations while open offices (and mandates) are better for disruptive technology.

“Google started a beekeeping club so that engineers who are interested in beekeeping might meet each other and actually have discussions about unrelated subjects.”

Pervading the office furniture is the office culture, and perks are not culture. Ben Horowitz said that culture is what people do when you don’t tell them what to do. Does your boss encourage serendipity? Do you?


Lindsay cites Geoffrey West that cities get better as they get bigger.

He also (seems) to like the work of Jane Jacobs and mentioned a Sante Fe panel/study along with this; Being Nicely Messy. While YOU and THE OFFICE are bonsai-like cultivatable, the city is more emergent. “Cities are the greatest serendipity environments of all so Google and Facebook are trying to design these environments that are more walkable and allow all these sorts of spillover effects.”


Org charts often differ from communication networks.  Albert-Laszlo Barabasi has done similar work to what Lindsay cites. Barabasi noted that the people best connected in a network are rarely the decision makers, it’s the safety management supervisor who has to visit each office and likes to talk.

Other examples from the talk include:


Thanks for reading and happy serendipitying to you.

Brynjolfsson and McAfee

Supported by Greenhaven Road Capital, finding value off the beaten path.

For a long time, there were only small changes in human existence. Then James Watt was asked to fix a steam engine. Watt fixed, tinkered, thought, experimented, and (is credited) with the invention of the double acting piston. Then the industrial revolution took off. Watt’s invention is a GPT (general purpose technology). Christened to remove water from mines, it was implemented in many other places. Subsequent GPTs were railways, the internal combustion engine, and electricity.

Technology becomes a GPT and not just a T when it’s applied to adjacent domains. In the early days of electricity, factories kept their layouts but changed their fuel. Electricity became an OG GP with the conveyer belt.

From there we get the Model-T, which benefited from another GPT, the combustion engine. Then we got the internet. As Andrew Ng noted, malls with websites weren’t Amazon competitors. What makes an internet company said Ng, is acting like an internet company. One example is Spotify and the Discover Weekly Playlist. AI may be another GPT, because Brynjolfsson said, “The core AI breakthroughs have applications in just about every part of the economy.”

This doesn’t mean the end of work. “It’s too soon to worry about the end of work…it may be that someday we will be able to make machines that do the full spectrum of what humans can do but that’s not the challenge we face today.” Ng said to think of tasks, of “anything a typical person can do with less than a second of thinking we can probably now assume automate.” Ajay Agrawal said much the same, “AI doesn’t do workflows, it does tasks,” and “Think of the recent advances in AI as advances in prediction; better, faster, cheaper prediction.”

Brynjolfsson isn’t worried about jobs so much as returns from work. Median income is flat since 1999 and though the economy has grown, the top twenty percent (based on wealth) earned more than 100% of the gains. They are the Apple iPhone of the economy.

What’s allowed this? People do more with less. Facebook bought Instagram when there were fifteen employees. When photography was a chemistry problem, Kodak occupied the lord’s estate and 145,000 serfs kept the kingdom running. When photography became an arithmetic problem Kevin Systrom & Mike Krieger stormed the gates with a band of merry men. Though it wasn’t only men, it was Systrom’s wife, neé girlfriend, who contributed an essential idea.

Instagram’s success is emblematic of the digitization bull that’s gone through the economic china shop. Along with easy copies are zero marginal cost distribution and the rise of the individual. Steve Jobs got this wrong, noted Brynjolfsson’s co-author, Andrew McAfee. Jobs mistakenly wanted the app store to remain closed. But the iPhone’s ascent accelerated when it opened, creating what Alex Moazed calls a Modern Monopoly.

Though we do more with less, “it would take the average American only eleven hours of labor per week to produce as much as he or she produced in forty hours in 1950,” this doesn’t mean the end of jobs is nigh. The Second Machine Age is an optimistic book about work.

AI, like other technologies, can supplement humans. Thanks to algorithms we’ll stick with base rates rather than have overconfidence in our abilities. McAfee said, “The single biggest failure mode that I see when I talk to smart people is that smart people tend to have exaggerated version of a failure mode, to be too confident of their own judgment.”

Then there are the jobs that won’t be too affected. In a study of 964 O*Net occupations, most had 20-30 tasks and most of the tasks were things algorithms won’t or can’t do. “When you look at all the tasks of a particular occupation, some of them were suitable for machine learning but many others were not. There was no case where machine learning just ran the table and was able to do the whole set.”

We talk about truck drivers being automated but truck drivers do a lot. Finn Murphy wrote that he was not only a driver but a manager, counselor, and in one case, an honorary Native American. Hal Varian makes a similar point when he points out that only one job has been automated away, the elevator operator. But, the tasks from that job have shifted to other areas, receptionists, concierges, and hostesses all do more.

Education might help. “There’s an increasing need for interpersonal skills…It’s not that work is disappearing but that there’s a whole bunch of tasks that only humans can do and we need to shift our skillset into those and then labor income is likely to go up.”

In another talk Brynjolfsson explains, “Right now and for the next ten years training and education are probably at the top of my list and most economists…I don’t think our schools are doing a very good job teaching those, or worse yet, many of them are actually crushing them.”

Seth Godin wants schools to teach students how to lead and how to ask interesting questions. Bill Burnett at Stanford ( said school cripples our creativity, “Something happens, mostly our educational system which promotes a massive fear of failure and a search for only one right idea.”

Rory Sutherland noted that the opposite of a good idea may be another good idea, which is contrary to the one-right-answer attitude in some schools. Pat Dorsey praised the liberal arts approach to education; “In a way equity investing combines a lot of different fields in that it is a set of ever-shifting problems to be solved. It’s not electrical engineering where there’s a right answer to everything. There is never a right answer to what the company is worth, or what is the competitive advantage to a business, or how much cash will they generate in five years.”

Which brings us to Erik Brynjolfsson and Andrew McAfee’s, Machine, Platform, Crowd. They wrote this one because “People who run companies kept approaching us in the hallways of places like Davos and kept on saying, ‘I believe your story, now what do I do?’ In some cases, I got the impression that there was desperation behind the questions,” explained McAfee.

Brynjolfsson and McAfee’s solution is in the title; collaborate with machines, harness platforms, and use the wisdom of crowds.

They wrote the book because “The narrative these days is that technology is killing jobs and it’s not too far from that narrative to ‘therefore technology is a bad thing’. We think that’s a really harmful road to go down.” Paul Daugherty agrees, “I’ve started using the term ‘collaborative intelligence’ rather than ‘artificial intelligence’. The problem with AI is it scares the public and it leads to the wrong discussions.”

Another stumbling point is making GPTs general. Malls with websites aren’t Amazon and record stores online aren’t Spotify. The issue is that “Tech progress rewrites the business playbook.” McAfee said, “I see a lot of companies underestimating the power of machines a lot of companies underestimating the power of platforms and a lot of companies not doing a great job of tapping into the crowd.”

Okay, so how can someone use these ideas?

For machines, we need to get away from HiPPO decisions ( and move to Geek decisions. This is sabermetrics in baseball like the Oakland A’s. This is running 10,000 experiments a day like at Google. This is who to market new shows to like at Netflix.

For platforms, the key is the interface. How can an organization create an easier way for customers to find X? How can an organization create easier ways for producers to create X? Successful platforms solve for Michael Munger’s three T’s; triangulation, transaction, and trust.

For crowds, the key is trusting weirdos. McAfee said that the core can be bested by the crowd. “I honestly just mean these hundreds of millions or very soon billions of complete stranger weirdos out there available on the internet that you can tap into where you can access their knowledge and their tenacity and their energy if you do it correctly.”

Yet again we’ll remind ourselves that technology is a tool like a shovel not a panacea like a silver bullet. It’s still up to the user not to hammer their thumb. Machines, crowds, and platforms are wonderful compliments for decision making. McAfee said, “The failure mode among really smart people is to trust themselves too much.”

No one knows how fast work will change. We’ll get glimpses of pieces first. The duo gives examples like a smart grid in a series of factories or an autonomous shipping route from Dallas to Los Angeles. While Brynjolfsson wants more training, McAfee wants changes to the entrepreneurship policy and physical infrastructure.

Neither seems to think work is going to go away in the next thirty years, and McAfee points out work is more than a job. “I’ve become a fanatic about work, about the value and importance of something like a job. The evidence is overwhelming that when work leaves a community bad things happen.”

Roger Lowenstein wrote about Warren Buffett, “(he) understood that most people, regardless of what they say, are looking for appreciation as much as they are for money.” Voltaire concluded, “Work keeps at bay three great evils: boredom, vice, and need.” Adam Smith suggested that man wants to be loved and be lovely. Andy Grove wrote this:

“I felt the frustration that comes when the things that worked for you in the past no longer do any good.”

“Businesspeople have emotions, and a lot of their emotions are tied up in the identity and well-being of their business.”

“In many instances, your personal identity is inseparable from your lifework.”

Change, especially when your personal identity is at stake, isn’t easy. Brynjolfsson and McAfee cite the work of Angus Deaton and Ann Case who coined ‘deaths from despair’. “I think about which of those social problems will be fixed by a magical check from the government showing up every month,” McAfee said, “and my answer is basically none.”

Jobs give more than money.

We don’t know how the future will be different only that it will be different. In that spirit we’ll take our final word from Andy Grove:

“You need to plan the way a fire department plans: It cannot anticipate where the next fire will be, so it has to shape an energetic and efficient team that is capable of responding to the unanticipated as well as to any ordinary event.”


Thanks for reading.

Final Jeopardy

Supported by Greenhaven Road Capital, finding value off the beaten path.

Final Jeopardy by Stephen Baker is the story of IBM Watson, the Jeopardy champion who defeated Brad Rutter and Ken Jennings.

In December we looked at AI with posts from Andrew Ng, Paul Daugherty, Hal Varian, and Avi Goldfarb and Ajay Agrawal. The book Final Jeopardy is a polaroid for AI. It’s a precursor to those blog posts. A sort of this is how things were.

In the late 1970’s IBM crested the technology wave and then big blue began a descent as Microsoft rode the new wave of PCs. IBM’s mainframes were a vertically integrated business model and the era of PCs, horizontal integration ran roughshod over a prostrate Big Blue.

By 1992 the company was saddled with a 5B$ loss and their shift to software and consulting began. In 1997 IBM’s Deep Blue defeated Kasparov in chess and in Harry Friedman was promoted to Jeopardy producer and began to include more popular culture and current events in the show.

He thought he was going to figure out the computer. That BBC video demonstrates something Rory Sutherland rails on; that creativity means solving problems in new ways. Kasparov resigned after 19 moves becuase the computer solved the problem in new ways.

Aside from Trebek shaving his mustache (2001) and SNL Celebrity Jeopardy, Jeopardy, America’s favorite quiz show stood stolidly with Wheel of Fortune as American staples.

In 2003, the five-show limit was lifted and Ken Jennings took the game by storm, drawing 15 million viewers – according to some – igniting a question at IBM. Could we beat him?

Watson was part curiosity, part marketing, part interestingness. Baker wrote, “IBM’s biggest division, after all, was Global Services, which included one of the world’s largest consultancies. It sold technical and strategic advice to corporations all over the world. Could the consultants bundle this technology into their offerings?”

Internally there were some high hopes. They just had to make it and that proved damn difficult.

“The biggest obstacle (early on in AI) was language.” In addition to that, there was no consensus on how to approach trivia questions. Some researchers want general AI, others craved specific. Reading a book written in 2011 about events in 2009 in 2018 gave one perspective but it helps to remember that whatever is being done at the time doesn’t seem easy to the people doing it. We all Google all day, but Baker does a good job articulating all the issues the team faced.

In 2005 IBM starts considering the challenge in earnest and one engineer is tasked with an off-the-shelf approximation. We see this advocated in companies from IBM to IDEO, there’s huge value in prototyping. Before jumping in, assemble a simple solution. IBM discovered that their first internal version was much better than off-the-shelf software. They moved on.

The IBM team used stuff they had worked on but also invented new ways to parse data. They also created new ways to work together. One key to Watson was parallel inquiries. Once the system figured out what a question asked; the sausage celebrated every year since 1953 in Sheboygan Wisconsin, it would try to come up with a collection of answers.

Ideally, some of the answers would be the same and quantity lent credence to quality. It was a sort of wisdom of crowds, what does the collective algorithm suggest? And the team needed to work this way too. “Ferrucci decided to take the same approach with his team. He would cluster them. He found an empty lab at Hawthorne and invited his people to work there.” They had to be together.

This might be true beyond teams like the Jeopardy engineers. What if schools work so well not because of economies of scale but network effects? I give you a good idea and you give her a good idea and so on. What if we should prioritize the gestalt rather than the CBA.

Vicinity also kept values aligned. Small groups don’t need mission statements so much as communication. The Watson team didn’t necessarily want Watson to Win. They wanted Watson to demonstrate IBM’s prowess. That included not looking bad, the team worried that “a humanoid Watson might frighten people.” Watson needed to represent IBM well, and not embarrass them. It could be funny, but not profane. Early answers examples like “This is the type of diet grasshoppers eat – Kosher” brought laughs. But in the category of Just Say No, ‘This is a four-letter German word.’ Watson’s gave the answer ‘Fuck’.

Oops. IBM staffers chuckled and began work on a profanity filter.

Straight facts were the easiest questions. “There were factoids, each one wrapped in the most helpful data for Watson: hard facts unencumbered by humor, slang, or the cultural references that could tie a cognitive engine into knots.”

But Jeopardy is full of puns and ambiguity. “At low levels of confidence, I think we’ll just have it say it doesn’t know, sometimes that sounds smarter,” said Chu-Carroll. Watson needed to look like a helpful took, not an unhelpful HAL.

What’s clear from the book is that Watson took a lot of work to work well in a limited domain. What’s also clear since then is that technology has gotten quite good at working well in many different domains. From grammar to adaptive cruise control, it seems like technology mostly works and mostly makes life a lot better. One step along that way was Watson.


Thanks for reading.

Brian McCullough


Supported by Greenhaven Road Capital, finding value off the beaten path.

Brian McCullough spoke at Google and podcasted with Chris Dixon about his book How the Internet Happened. McCullough said he wrote the book  because “I feel like anybody would want to know who did this” and “It just sort of bothered me that no one had done a book about the internet going mainstream and I just got tired of no one writing that book.”

These glitches in the Matrix where something confuses us, enlightens us, or bothers us are where a lot of products begin. It’s the catalyst for products like RX Bars, Panera Bread, and Spotify’s Discovery Weekly. “So like any other startup I’ve done, I was like, ‘Well someone’s gonna do that so why not me?'”

McCullough found some interesting things at the library, like Al Gore’s contribution. During interviews, he also noticed some media manipulation. “To a person, all the Mosaic/Netscape engineers made the point that this sort of work hard, sleep under your desk, startup culture was from the media. These were corn-fed Midwesterners.”

The media distorts things all the time. Ed Catmull noted the caricature of Steve Jobs. Annie Leibovitz wrote about the manipulation of scenes for war zone photos.  Andre Agassi never planned on saying image is everything.

But maps distort things. So do stories. Memories too. Nothing is a perfect replication.

As countermeasures, we can adopt Tyler Cowen’s approach and avoid the philosophy of once-and-for-allism and heed this advice from Jason Bateman:

“Baked into the cake is the assumption that the people reading this are reading this for fun. You kinda know that with those magazines there’s more than a grain of salt, more like a kernel of truth to it, but its super hyperbolic to get the magazine sold or the click-bait on the computer.”

Netscape’s browser was the first major software to be distributed digitally. McAfee and Brynjolfsson and Smith and Telang researched why this shift was important. McCullough reminds us, “Before Netscape, you would still put software in boxes, shrink wrap it, and put it on a shelf.” I remember getting Warcraft in a box and thinking, that’s a lot of space for one disc and a manual.

Netscape was also the first internet company in the sense they acted like – A/B testing – an internet company. Andrew Ng said, “What defines the internet company is whether or not you have an architect at your organization to leverage internet capabilities to do the things that the internet allows you do really well.”

Part of the reason Netscape did this was out of fear. “The research was a surprising reminder to me how much everything in the 90s was in relation to what Microsoft might do.” Read anything about technology in the 90s and Microsoft comes up. Tim Wu compared Microsoft to Kronos, eating the young. Jerry Kaplan wrote “If we so much as threatened to sue, it would rain lawyers on our office like the plagues of Egypt.” “Googlers told me that when AdWords finally took off they didn’t tell anybody until they were far enough along so that Gates can’t catch us,” noted McCullough.

Gates’s mistake was in thinking TV was the portal and broadband was the means. “The thing that Gates missed was that the internet was good enough.” McCullough told Dixon, “The lesson of the early internet is that sometimes just good enough technology is good enough.”

This was true for Twitter and the fail whale. Instagram too. Kevin Systrom said, “To be clear, there is no reason we should have succeeded. The server was down every other hour and people just kind of forgave us. They came back and they would share their photos. At the time mobile networks weren’t that great either, so people would blame their connection and not us – which was great.”

McCullough’s talk and book include many other companies; Facebook, AOL, Yahoo, and Amazon among others. eBay too, which did three things.

  1. “Taught normal people to trust faceless strangers across the country online.”
  2. Created “masses that are self-organizing,” and introduced a rating system.
  3. “They were the first company who succeeded with the business model of the platform, of whatever the users are doing.”

McCullough says that newspapers were interested in buying eBay but they had no inventory. Chris Dixon noted that what the publishers thought was a bug was actually a feature. Related was Alex Rampell’s Tivo problem.

At one point, 7% of all eBay listings were for Beanie Babies, and Zac Bissonnette wrote about the toy’s importance, “In order to lure people out of their comfort zone and into the idea of e-commerce—and, even scarier, peer-to-peer e-commerce—eBay needed to offer consumers something that was easy to ship, couldn’t be found anywhere else, and elicited passion among the people who were looking for it.”

More recently Michael Munger summarized the economics, noting that platforms reduce the transaction costs. He said, “The best proposition is where transaction costs are high but reducible and excess capacity is high.” Before eBay people used newsletters and local shows to trade toys.

McCullough also addresses the idea of a killer app. “I think the reason mobile didn’t take off for so many years was because there wasn’t anything for normal people to do with it. The iPhone comes out in 2007 and six months later Facebook opens registration to everybody.” The ‘opening’ was something Steve Jobs nearly got wrong.

Ben Horowitz addressed this idea too, “When asked ‘What’s the killer app?’ No one ever gets it right.” Horowitz didn’t give an answer but he did offer an approach. “If you think about the smartphone it was much worse than the PC. It had a tiny screen, it was far less powerful, etc, etc. But it had a couple of properties that weren’t in a PC. It had a GPS and it had a camera, so you could now build things like Lyft and Instagram.”

McCullough supposed, “Maybe you had to have Facebook to go mainstream for the iPhone to go mainstream and vice-versa.” There had to be ‘an app for that.’

Screenshot from iPhone 3G commercial.

We study history, said Michael Ovitz “Because past is prologue.” It also helps said McCullough, “how much can you learn by looking back at dumb ideas that will always be dumb ideas (inversion) or dumb ideas that were only dumb ideas at that time (timing matters).”

This a trope that this time is different. That’s true on some levels but not others. What never changes is an exchange, an X for Y. That’s the essential economic question. What always changes is the means, reasons, and explanations.


Thanks for reading.

Peter Rahal

Supported by Greenhaven Road Capital, finding value off the beaten path.

Peter Rahal and Jared Smith founded RXBar in 2012 and sold to Kellogg in 2017 (for 650 million dollars). Rahal shared his story on the How I Built This Podcast and no podcast is the antithesis of 28 Lessons from Startups that Failed as this one.

Peter Rahal didn’t know he was dyslexic until late in his academic career. “From second grade to age twenty-two I just thought I was not successful.” Malcolm Gladwell suggested dyslexia was an advantage but I think it’s something different. Successful entrepreneurs are prosocial deviants. They want to dent the universe. They want to shape the future. They want to do something their way because they believe it’s better. It might be why many founders don’t like formal school.

Josh Wolfe does one better than ‘not liking formal school’.

There are ranges of tendencies that don’t help in K-12 school. In that setting they’re weaknesses, but for each weakness has a strength. Ben Thompson said, “I’m a strong believer that for anyone, their strengths are their weaknesses.” Think about where you live, Thompson explained. There are advantages to be in SF or DC or USA but there are disadvantages too, like insularity or biases.

Rahal guessed that dyslexics trend towards entrepreneurship because they tend to see the whole better and, “A lot of entry-level jobs require sequential thinking and I think that’s why you see a lot of entrepreneurs who are dyslexic.”

What does someone do if they can’t do some of the Level One job?

Rahal bounced around after college. He worked for a family friend. He went to Lebanon. He said about the time there, seeing his family’s history, “I had to be there. You gain a lot of empathy from seeing how people live.”

In early 2012, Rahal attended a class at his CrossFit gym. While there he wondered why they sold water and t-shirts but not protein bars. He told, “The (bar) category was so competitive in places like Whole Foods,” but not at CrossFit.

He went home and tried a batch. The RXBar recipe includes dates and figs but was made with anger, hunger, entitlement, and a short temper. Peter Rahal was not complacent.  He pitched his friends on the idea. He wanted to see what they thought, but he also needed help. “One of the advantages of being dyslexic is knowing very early in your life what you’re not good at. I knew I couldn’t do it by myself. I couldn’t do a lot and Jared (Smith) could balance me out.”

The RXBar team (neé Peter and Jared) started their market research. WholeFoods didn’t stock anything as simple, nutritious, or delicious. So Rahal and Smith fired up a red Cuisinart blender and got to work. It wasn’t any more difficult than making cookies.

The duo tinkered in Rahal’s parent’s basement and invested five thousand dollars each for supplies and raw materials. Their low overhead was essential, especially for a business with inventory. William Thorndike concluded, “There’s an apparent inverse correlation between the construction of elaborate new headquarter buildings and investor returns.”

Then there was the advice they got from Peter’s father. “I remember asking my dad, ‘Do you know any rich people you could connect me with?’ He’s old school, super traditional. He told me, ‘You need to shut the fuck up and sell a thousand bars.'”

So they did.

“I remember showing up at the CrossFit gym I was a member of with a Tupperware of this stuff all the time. I did this all the time. We collected data all the time. It was; ‘Do you like this? What about this?’ We just asked a ton of questions.”

Talking to your customers is essential.

Fellow HIBT guest, Ron Shaich did almost the same thing. Before he started with bread, Shaich sold cookies and like Rahal he beat the street asking people what parts they liked and what parts they didn’t. Listening to customers led to Panera Bread.

Tiffany Zhong said, “Anyone who does stuff in product, design, engineering, or wants to start their own company should know how to talk to and interview users or potential users.” Jenn Hyman bought one hundred dresses from Bloomingdales and hosted a pop-up event at Harvard. “The idea behind this was to see if: Women will rent dresses. What will they rent? How much will they pay?”

At CrossFit, people liked the bars. Rahal’s idea sprouted in August 2012 and with the first sale in March 2013 the seedling broke through the surface. Now it just had to survive.

They kept their costs low, designing packaging with PowerPoint. They assumed, hoped, and neglected the stability of the product. “In the early stages of business you can’t break the laws but you have to break the rules a little bit. We were eating it and we were fine.”

They kept selling the bars at CrossFit. They WANTED to sell the bars at CrossFit. “I think a lot of people look at niches and (think) it’s not big enough for them.” The strategy was to be in a CrossFit in Chicago before being in a grocery store in California. The early days were a combination of sales hustle and ingenuity to keep costs low. “Instead of moving to a kitchen we moved to my parent’s basement. Instead of going to a designer we used PowerPoint. Instead of going to a printer we went to Staples. Instead of a third party website designer, we went to Shopify.”

RXBar succeeded because technology allowed them to scale. All of the intangible parts; the software, website, marketing, etc. were enabled by software. Sam McBride was employee #4, (in 2018 he was the COO) and he said their early marketing was him laying in bed at night and sending Direct Messages to Instagram users asking if they’d ‘influence’ the product.

In his Talk at Google, McBride restated the simplicity of RXBar’s early business model. “The business was very simple; make bars and sell bars and that was basically it. In between, there hopefully was enough cash to buy raw materials to make more bars the next time.”

Sidenote: this situation is why Albert Wenger is so bullish on cryptocurrencies. When asked about the necessity of financing, “There’s one and only one answer that’s the correct answer, the timing of cash flows. You need financing if you need to pay people before you get paid.”

McBride had an unconventional path to RXBar. He knew Smith and Rahal but he joined the company because he sold his home security business in rural Kansas. Really? “This has become a filter for me. If any kind of early-stage company or product is not met with enough people that don’t get it or it’s not kind of uncomfortable or embarrassing to explain what you’re doing, then probably too many people are already doing it.”

Howard Marks suggested investors (and entrepreneurs) be different and be right. It’s easier to be different if you’re in a boring business. Michael Ovitz/CAA survived their early days thanks to their unglamorous, though lucrative, daytime television contracts.

Businesses succeed when they make something people want to buy. McBride said, “We were so close to the consumer; selling directly, taking their phone calls, emailing when they had problems, we knew that the packaging would work.”

This is why CrossFit was so important. People don’t call the 800 number on packaging but people will tell you between bites when something is delicious. As RXBar grew they included a phone number and the rule was one ring only said, McBride. Paul English did this at Kayak too.

Through luck and skill, RXBar did a lot of things right for the conditions of the day. Rahal said, “It was an interesting time in CrossFit because the idea of retail was just beginning.” Danny Meyer saw this too, “I would also have the good fortune of entering the restaurant industry during its fertile period of revolutionary change.”

Another helpful trend was online sales. Rahal explained, “When you’re selling online you’re collecting that cash.” McBride noted this too, saying that rather than net-60 terms they had net-1, immediate payment.

Another was online marketing. McBride’s home security business was direct to consumer and explained, “We couldn’t use influencers in the home security business but we could use influencers really effectively for food.”

But they still had to make the bars. They couldn’t keep up with demand from the kitchen basement so they moved to a co-packer. The first one didn’t work out and they had to dump $300,000 of inventory. They kept growing, selling, and making bars. Eventually, they outgrew the American date supply.

In November 2017 Kellogg purchased the company. “We wanted to do a 100 percent sale, because the process takes a shitload of time,” Rahal told It’s almost like he’s read Brent Beshore’s Messy Marketplace.

When asked about his success Rahal said:

“I’m very lucky to be born in the United States, to have my father and mother raise me like they did. I’m very lucky to have met Jared to have a humble partner to do this with. I’m very lucky this is the golden age for entrepreneurship in food. Talent has to be a variable but it doesn’t happen without good timing and luck.”

The sale to Kellog was a finish line for one part of the company’s life but the starting line for another. McBride told his audience that in the early days “We never sat down and wrote out our core values. They just sort of are what they are and everyone knows them.” That’s easy with four people. It’s harder with hundreds and McBride noted that they had to be more thoughtful around hiring as the company grew.

One part of the RXBar culture is clarity. “Super direct, totally transparent, and authentic,” McBride said. To emphasize the point McBride had his annual review at an all-hands meeting. “It was really effective conveying that being vulnerable is okay and that everybody’s got pretty decent-sized-weaknesses.”

RXBar is transitioning from selling and growing to strategizing and gathering. McBride explained:

“I think of it as a continuation from execution to strategy. In the early days, you’re like 95% execution and 5% strategy. You know where you’re going but all you’re doing is executing. But as the business grows you have to spend much more time on strategy because you’re driving a much larger organization towards a goal.”

Kellogs helps with this. Chris Dixon noted that there’s often a race between companies with good ideas and companies with users. Can the company with the ideas generate more users faster than the company with the users can generate good ideas?  McBride said that Kellog, “brought a global perspective that would have been hard to build as fast as we needed to build it.”

RXBar is another example of Scott Adam’s slot machine approach. The company was a longshot and in an alternate history was not the category winner. But it this one they were and that happened because Rahal attempted it. That’s a good lesson early into 2019, to make the attempt.

Thanks for reading.

Michael Ovitz


Supported by Greenhaven Road Capital, finding value off the beaten path.

Michael Ovitz says that “past is prologue” and with that idea in mind, we’ll look at his book, Who is Michael Ovitz?. We laud Ovitz for his persona of success but that’s not the character in the book. He’s contrite. He’s reflective. He longs for lost relationships.

He made a lot of money. He did loads of charity work. He empowered artists. Ben Horowitz said, “CAA was really an impossible thing.” But was it worth it?

Maybe the customary CAA red is a warning. This is what it takes and this is what you get.

That’s not to say there’s nothing to learn. One lesson is to be different. “In any multiplayer contest, you want to be the outlier.” CAA succeeded they were different and different was good. Ovitz said that client’s ideas were like clay and CAA was like the studio where it got turned into something great. “That worked really well for the creative clients, but it did not make us popular.”

CAA helped clients with everything and it was that culture of service Ben Horowitz brought to a16z. But this isn’t ‘copy and paste’. “There’s no one thing where you can say, ‘My god, you guys called everybody a partner that’s really a game changer.’ It’s a composite,” explained Ovitz.

Horowitz added, “Culture supports the strategy. You can’t come in from the outside with the culture of somebody else’s company.” Amazon can’t have Apple’s culture and Apple can’t have Amazon’s culture Horowitz said. Culture also has to fit into the time and place a group is in.

Ovitz et al. succeeded because timing matters, “We were lucky to work in a golden age of commercial film. People went to the local multiplex three times a month, piracy had yet to explode, and cable was in its infancy.”

CAA could package actors with directors and writers for studios because of the era they were in. Gary Vaynerchuk talked to Ovitz too and pointed out that creators can go straight to consumers via technology now and Ovitz agreed on how things have changed. He told Horowitz, “When we started there was a very large barrier to entry in the media business. Anything you could read, see, or hear was controlled by twenty-five companies.”

That’s less so today and why we have Modern Monopolies, Machine, Platforms, Crowds, and Streaming, Sharing, Stealing. It’s why Here Comes Everybody and the rise of the Attention Merchants.

Boring Businesses. When Ovitz fled William Morris, CAA was buoyed thanks to daytime TV. It, “wasn’t glamorous, but it was lucrative.” Another revenue stream was authors. A third was syndication. CAA’s foothold was in the perceptive mud but the profitable pay dirt.

Scott Galloway recalled an investment summit where a few attendees “own(ed) media properties and the national airline, but most killed it in iron/ore smelting, insurance, and the like.” Ali Hamed told Patrick O’Shaughnessy about a Stanford Ph.D. who built an app for heavy construction equipment operators. “That’s a person who’s solving a problem that no one in Silicon Valley gives a rats ass about.” Others like Brent Beshore and Josh Wolfe find their own ‘boring’ businesses.

Ovitz became ritz, but the foundation of CAA was poured with hard, unglamorous work and reinforced with honesty. Ovitz told Horowitz “We were less concerned with telling people things they wanted to hear. We were more concerned with telling them things they didn’t want to hear.”

And wrote in the book, “When you tell someone the truth, all they can do is get upset – they can’t call you an idiot.”

Roz Hewsenian told Ted Seides something Ovitzian, to never say ‘No’ to a client. “You figure out how to give them some of what they want so that you show them you heard them, which is ninety-percent what the issue is.”

Aggressive Education. “Patton was a threatening motherfucker, and so was I.” Ovitz may have wanted a role in a buddy cop comedy but instead, he played the role of bad cop. He told Horowitz “I don’t think we could have done it differently because we were in a cutthroat business.”

He was always working. He was always learning. Assisting with M&A work he wrote, “Collaborating with Pete (Peterson) and Steve (Schwarzman) and Herb Allen beat going to Harvard Business School; they taught me how to be an investment banker.”

Then years later, “As Marc (Andreessen) and Ben (Horowitz) led me into their world, I felt like a privileged student in a graduate school of one.”

Ovitz’s aggressive growth mindset is reflected in others who have less cutthroat images. Alice Waters’s education was in the fields of France. Rich Snyder (In-N-Out) was over kitchen friers and in leadership seminars. Stewart Butterflied got his continuing education at Yahoo (and saw what not to do). Seth Klarman got his at Mutual Shares.

A great what-if that we can only ask now is ‘What-if Michael Ovitz had social media?’ Paul Newman said Ovitz was “a cross between a barracuda and Mother Teresa.” Somewhere between the pre-2018 perception of Ovitz and the post-book-tour perception is the real Michael Ovitz.

When he and CAA were humming, Rowland Perkins advised Ovitz that “you can only start fires” if you talk to journalists. So with nothing to talk about the media wrote what they wanted to be talked about.

In Ovitz’s era, he should have followed Buffett’s lead and wrote letters. Today, everyone has their own outlet for storytelling. As Seth Godin wrote, “making is insufficient.” What really matters is change and change takes making and communicating.


That’s what Ovitz did, that’s what Ovitz got.

Thanks for reading.