Part I: Socrates - the wisest man in Greece
One day Zeus, the Greek god of the sky, took two eagles and flew them in opposite directions. He hoped that their meeting point would help determine the center of the world. The birds crossed paths at Delphi, a hill at the base of a mountain in central Greece. Over time a small sanctuary developed at this site. The location thus became the worshipping ground for Gaea, goddess of the earth. Gaea placed her son, Python, as a guard for the oracle (priestess) at Delphi.
Apollo, who was a much more powerful god than Gaea, desired the site for himself. He thus killed Python and took the site for himself. However, he let the oracle stay to serve him directly. Since the oracle spoke directly with a god, the legend of the Oracle of Delphi was born. Soon thereafter, politicians and nobles from all over Greece began visiting the site to ask the oracle for advice.
Around 440BC, Chaerephon - a friend of Socrates - paid the oracle one such visit. During his conversation, he asked her who she believed was the wisest man in Greece. Her response:
"Of all men living Socrates most wise"
Chaerephon later recalled this statement to his friend, a puzzled Socrates. Socrates knew that the oracle could not lie since gods did not lie. But he personally knew that he possessed very little actual knowledge. Determined to prove the oracle wrong, he set about interviewing intellectuals (poets, politicians and craftsmen) in Greece. He expected to find that they knew a lot more than him, repudiating the oracle’s claims.
Very soon, he understood the wisdom behind the priestess’ words. Socrates realized that the intellectuals definitely knew their respective fields. But they knew very little about matters outside their domains. However, they mistook their intelligence in one field to believe that they could speak wisely about any topic. Socrates, on the other hand, lived under no such illusion. He knew that he knew nothing. As such, by simply understanding the limits of his knowledge, he was wiser than the rest.
Socrates was the wisest man simply because he knew that he knew nothing
Part II: Five Levels of Knowledge
I do not plan to declare myself as some wise old fish who has ‘always known how little he knows’. If anything, being an MBA-type, I am more likely than your average Joe to feign knowledge of topics of which I have a cursory understanding. I have recently been reminded of this point after I started watching a Wired series/playlist on YouTube called ‘5 Levels’. In these videos, an expert explains any particular topic at Five Different Levels of difficulty. See the video below for an example (watch the entire playlist if you are really curious):
Going through the videos, I was shocked by how little I actually understood. For many topics where I was sure I’d be a semi-proficient expert, I was left red-faced by my elementary understanding of the topic. Anyways, now that I am done with my show of false humility, let me return to offering opinions on things that I have a cursory knowledge of.
Part III: What Feynman knew
Richard Feynman was a great scientist. Probably one of the greatest of the twentieth century. But he was definitely the very best teacher of science during the last century. Key to his success was his ability to prune bullsh*t and to tear any concept down to its very first principles. Also central to his teaching philosophy was this idea that names don’t constitute knowledge.
Watch this short clip to get an idea of what that really means:
If you can’t watch the video, below is the relevant bit in which Feynman recalls a conversation with his father:
See that bird? It’s a brown-throated thrush, but in Germany it’s called a halzenfugel, and in Chinese they call it a chung ling and even if you know all those names for it, you still know nothing about the bird. You only know something about people; what they call the bird. Now that thrush sings, and teaches its young to fly, and flies so many miles away during the summer across the country, and nobody knows how it finds its way.
This idea that people try and reduce a concept down to some noun or verb was so antithetical to Feynman’s values, that he spent a great portion of his lectures explaining phenomena without employing ‘scientific’ terms. This philosophy extended to his vision for how kids should be taught. As his own son approached school-going age, Feynman worked hard to lobby the California board to ban a certain textbook which he felt didn’t properly explain concepts. Here is the relevant bit, lifted from James Gleick’s biography of Feynman:
It began with pictures of a mechanical wind-up dog, a real dog, and a motorcycle, and for each the same question: “What makes it move?” The proposed answer—“ Energy makes it move”— enraged him.
That was tautology, he argued—empty definition. Feynman, having made a career of understanding the deep abstractions of energy, said it would be better to begin a science course by taking apart a toy dog, revealing the cleverness of the gears and ratchets. To tell a first-grader that “energy makes it move” would be no more helpful, he said, than saying “God makes it move” or “moveability makes it move.”
Part IV: The FizzBuzz Test
This reflection around what constitutes knowledge reminds me of an old post by Jeff Atwood. In it, Jeff laments the state of computer science education as he reflects on his and his colleagues’ experience interviewing programmers. Here is the juicy bit (mostly passages lifted from blogs that he quotes):
“Like me, the author is having trouble with the fact that 199 out of 200 applicants for every programming job can't write code at all. I repeat: they can't write any code whatsoever…
After a fair bit of trial and error I've discovered that people who struggle to code don't just struggle on big problems, or even smallish problems (i.e. write a implementation of a linked list). They struggle with tiny problems.
So I set out to develop questions that can identify this kind of developer and came up with a class of questions I call "FizzBuzz Questions" named after a game children often play (or are made to play) in schools in the UK. An example of a Fizz-Buzz question is the following:
Write a program that prints the numbers from 1 to 100. But for multiples of three print "Fizz" instead of the number and for the multiples of five print "Buzz". For numbers which are multiples of both three and five print "FizzBuzz".
Most good programmers should be able to write out on paper a program which does this in a under a couple of minutes. Want to know something scary?
The majority of comp sci graduates can't. I've also seen self-proclaimed senior programmers take more than 10-15 minutes to write a solution.”
I am neither a computer science graduate nor a programmer. I am not sure what % of developers today enter ‘git clone xxx’ in the terminal without knowing the fundamentals of web development. But it is conceivable that this is a decent minority.
I do know a bit more about PowerPoint and Excel. These tools have done a fair bit to advance modern corporate finance and strategy. However, most folks who use these tools (myself included) primarily operate as slaves to these mediums. Let’s consider Excel for a moment, the more quantitative and ‘rigorous’ of the two tools. Now folks like Drew Dickson who spend their lives doing financial modeling can make any spreadsheet sing’. Most of us don’t possess the facility of Drew Dickson; however, we generally act as if we do. You learn the trade quickly - you project some numbers, run an NPV or IRR function or backsolve growth rate to get your desired graphs. And voila, you are good to go.
Does performing a successful vlookup imply any actual knowledge? How about creating a snazzy slide deck? Or getting XIRR to work on the first try. . What is the fundamental unit of knowledge implied here? Why is this any different from writing a good novel, albeit with different rules and characters. When you get a IRR figure, does anyone really think ‘oh geez, this gives me the discount rate at which the NPV of this project becomes equal to zero’ What is NPV? What even is a discount rate? No, the reality is no one thinks like that. Most people just learn these concepts as tools at some point in their careers. And then ultimately, most of our life is spent talking in useless abstractions.
But every time we engage in such an abstraction, we defer real thinking. This is no different than a programmer cloning a repository and copy-pasting code from Stackoverflow in the hopes that it sticks. Moreover, software development has a short feedback cycle. Typically, your code leads to a binary outcome (i.e. it either works or breaks). Few such comfort exists in the world of spreadsheets and slide decks.
What strikes me as interesting is that for these tools, a vast portion of human effort doesn’t involve any real knowledge. Rather, all these are tools to communicate abstractions of an imagined reality. As a collective, the skill-set on display is taking massive amounts of information and condensing it into bite-size chunks. If you are dealing with PowerPoint, you start with Google, dig up some consulting reports and weave a narrative around a bunch of slides. In Excel, you start with a large CSV file and find some interesting relations in the data. Maybe you do a regression and p-hack your results. Sometimes, you stumble upon a game-changing chart like this:
But generally the day-to-day grind is more mellow stuff. And obviously you write a report and throw in a customary ‘correlation doesn’t imply causation’/’is the sample biased’/’is the training data big enough’ caveats to appear smart.
Someone once lamented the fact that the best minds of our generation were dedicating their lives to optimizing C2S conversions for companies like Google/Facebook. I personally find it more depressing that the second-best minds of our generation (and a larger sample with a greater influence on actual socio-political outcomes) spend their days in the weeds of this Excel-Powerpoint grind.
What is more disheartening is generally the more senior one gets in their career, the more you start to speak in higher level abstractions.
Let’s consider a few examples:
We plan to use AI to deliver benefit for all parts of the business
We are building a digital platform that will be customer centric
I ran a regression and five of my variables are statistically significant
Our cloud transformation is halfway complete
We believe that our unique culture sets us apart from the competition
Blockchain technology promises to unlock some interesting use cases for us
and so on..
Our strategy remains to transform the lives of our consumers
All of these are reasonable sounding statements for a C-suite executive to make in an annual report or results presentation.
Part V: What people get wrong about Elon Musk and SV founders..
I recently watched this Elon Musk interview with Sam Altman. Video below:
Here is a part that jumped out to me (16 mins, 25 seconds onwards).
Sam Altman: So what does your time look like at SpaceX or Tesla?
Elon Musk: A lot of people think I must spent a lot of time with media or business-y things but actually almost all my time, almost 80% of it is spent on engineering and design. So it’s developing next generation product. That’s 80% of it.
Sam Altman: You probably don’t remember this. A very long time ago, many many years ago you took me on a tour of SpaceX. and the most impressive thing was that you knew every detail of the rocket and the piece of engineering that went into. And I don’t think many people get that about you.
Elon Musk: A lot of people think I’m a business person or something. Which is fine, business is fine. For SpaceX, Glynn Shotwell is Chief Operating Officer, she kind of manages legal, finance, sales and kind of general business activity. And then my time is almost entirely with the engineering team working on improving the Falcon 9 and Dragon Space Craft.. and at Tesla it’s working on the Model 3 and some things in the Design Studio, half a day a week dealing with aesthetics/look and feel things. And most of the week is just going through the engineering of the car as well as the engineering of the factory..
There is a common conception in the general public’s mind on the role of the CEO. This stereotype is one of an overpaid fat-cat finance whiz-kid, one who is often disconnected from the wider organization and one who spends his days engaging with media, public analysts and teeing off on the golf course. While this caricature might have something to do with current trends around wealth inequality, I think it does get one point right. Most chief executives are indeed folks who have primarily made it to the top because: a) they are excellent communicators b) they are very numbers driven (ex-consulting/ex-ibanking/ex-private equity) and can geek out over ratios used by public market analysts to ‘evaluate’ company performance.
But the comparisons don’t have to end with Zuckerberg or Musk. An overwhelming majority of tech CEOs are ultimately product-focused CEOs. Period. That means they are often quite close to the engineering of the actual product. Let’s take some iconic SV companies and consider the backgrounds of their founders:
Uber: Travis Kalanick (software engineering)
Airbnb: Brian Chesky (UX + design)
Stripe: Patrick/Jon Collison (software engineering)
Amazon: Jeff Bezos (engineering)
Founders ultimately have an outsized influence on company culture. Facebook’s culture has always been ‘hackey’ and has been about ‘breaking things’. This is because Mark Zuckerberg was a hacker. Google’s founders were computer science PhDs. Therefore, Google has always been focused on solving difficult problems and engineer their pride on solving quaint, difficult math problems (even the interview process tests for that). Brian Chesky is so obsessive over design that he is rumored to have debated every single pixel on Airbnb’s landing page. Stripe is obsessive about documentation and written communication because Patrick Collison is an obsessive reader.
Ultimately, most tech startups today are succeeding with technology because they are lead by product-focused CEOs (Tim Cook is the major exception that comes to mind of an executive of a major tech firm that is ‘ops/finance focused')
Part VI: Whatever happened to the conglomerate discount?
I recently saw this presentation by an Andreessen Horowitz partner (?) speculating on how every company will ultimately have a fintech layer. Here is the video:
I think this is just terrible advice. The notion that just because of open banking, APIs and middleware companies like Plaid, every company should have a financial services component is just plain dumb. It is classic Silicon Valley hubris. It posits that the hardest part of running a banking institution is the technology layer. Sure, technology is the part that banks and insurers struggle the most with. Many of them have their backoffice systems running on COBOL even today. But that is not to say that if you created a financial services application running on serverless software or JAMstack, you have created a superior proposition. No, the biggest pain point in running a financial services institution is rarely the technology bit. It is the regulations. It is the customer service. It is the compliance. And sure, you might come up with some software that improves the workflow for that. But for the most part, it takes people, muscle memory and lots and lots of regulatory filings to run such institutions. It takes capital deposited with central banks. And so on.
More generally, doing anything takes time. Building a successful business in one vertical takes years, often decades. That is why public markets have historically hated companies that do a bunch of unrelated things. This is pretty much what ‘conglomerate discount’ is all about. That is why a company like General Electric often trades at an overall valuation which is well below the sum of its individual parts. Even Amazon would likely see a rise in value if it split up its AWS and e-commerce business. So, yeah public markets might not be perfect, but this is one thing which they get right. Doing one thing = good, doing too many unrelated things = bad. It’s human nature. Do too many things simultaneously and you’re bound to end up doing a half-arsed job at all of them.
And yet, some SV startups today behave as if they’r impervious to this law of nature. Uber is now into food delivery, autonomous driving research, grocery, credit cards, fintech and what not. Slice, a company that sells software to pizza companies, is now launching its own fintech vertical. WeWork was doing a lot of this and is now selling many of these non-core businesses at a hefty discount.
And yet, some of the best startups (think Datadog, Zoom) are often about doing one small thing. Technology and globalization means that sub-verticals today are bigger than ever. Founders have built billion dollar businesses by focusing on one simple idea. Veeva, a SaaS company for life sciences, is worth $22B today. ServiceNow, a SaaS company for customer support tickets, is worth >$60B.
To conclude, here is some advice from Auren Hoffman on building a successful company:
Second thing is to put yourself in a box. Again this is pretty controversial. Most people don’t agree with me here. My core thing is to be as focused as possible, to put yourself in a box, and the box basically says what you are not going to do.
I think you should publicly state to everyone who will listen, that includes your customers, your partners, your employees, your investors, anytime you talk publicly about your company this is what we don’t do. We are not going to do all these things. We are just going to do this tiny little thing.
We are only going to do this thing. Here’s where we are going to go and we are putting ourselves in a very defined box. Almost every smart person in the world will disagree with me. They’ll say, “No, you need to leave your options open. You need to be able to do all these things. You don’t know where the world is going to take you.”
I think you can be smart enough to know where the world is going and put yourself in a box and really, really, really own that particular box. At LiveRamp people would say, “Are you going to do media?” We say, “We are never going to do media.” “Are you going to build applications?” No, we are never going to build applications.
We are just middleware company. We are just going to move data between one application to another. Today that’s still all LiveRamp does all these years later. It doesn’t do anything else. It just basically moves data between applications.
“Do you actually do anything with the data?” No, we just move data between the applications. We only do that. We’ll never do anything else. That’s the only thing we are going to do. If you do that, if you eschew optionality and you stay focused then it’s really easy for all your employees to make decisions.