The Idea Maze is a Useless Idea

In 2022, I launched an experimental series of cases on the ‘Idea Maze’. The term ‘Idea Maze’ is a Silicon Valley (or tech industry)-ism ; it was originally coined by former Coinbase exec Balaji Srinivasan and then expanded on in a subsequent essay by venture capitalist Chris Dixon.

The basic articulation from Balaji is as follows:

Most of the time, end-users only see the solid path through the maze taken by one company. They don’t see the paths not taken by that company, and certainly don’t think much about all the dead companies that fell into various pits before reaching the customer (emphasis added).

The maze is a reasonably good analogy. Sometimes there are pits you just can’t cross. Sometimes you can get past a particular minotaur/enter a new market, but only after you’ve gained treasure in another area of the maze (Google going after email after it made money in search). Sometimes the maze itself shifts over time, and new doors open as technologies arrive (Pandora on the iPhone). Sometimes there are pits that are uncrossable for you, but are crossable by another (Webvan failed, but Amazon, Walmart, and Safeway have the distribution muscle to succeed).

Dixon then took this analogy and expanded on it:

Good startup ideas are well developed, multi-year plans that contemplate many possible paths according to how the world changes. (...) Imagine, for example, that you were thinking of starting Netflix back when it was founded in 1997. How would content providers, distribution channels, and competitors respond? How soon would technology develop to open a hidden door and let you distribute online instead of by mail? Or consider Dropbox in 2007. Dozens of cloud storage companies had been started before. What mistakes had they made? How would incumbents like Amazon and Google respond? How would new platforms like mobile affect you?

I thought that this might be a useful idea to examine. Everything sounded so … smart. Surely Dixon and Srinivasan know something about traversing the Idea Maze — after all, they have both started companies they sold; surely they rank highly on Commoncog’s hierarchy of practical evidence?

The idea behind my case experiment was simple: what might we learn if we examined the concept of the Idea Maze through a sequence of real world cases? Earlier, I had run an experiment publishing cases on companies with a Scale Economies competitive advantage, which was quite successful. I expected that with this experiment — much like the Scale Economies experiment before it — we would pick up certain common patterns across the cases.

We did not.

You can read the summary essay I wrote at the end of that test and squint; you’ll see that I struggled to find commonalities across all the different cases. You may also read this list of takeaways from Commoncog reader Michael Webb, who noticed certain things that I didn’t.

What was the problem with our conclusions? The problem is that none of what we noticed was very useful. I wrote things like “the idea maze is fundamentally about uncertainty” and “you should expect to make some decisions on the basis of little-to-no evidence when navigating the Idea Maze, and this is what that looks like.”

Which … duh.

Yes, the phrase ‘Idea Maze’ is useful as a shorthand to describe the difficult journey that all successful companies must travel on the way to finding product market fit. But … so what? Let’s say that you’re a founder. Let’s say that you want to find a successful product and win. How does this help you today? How does the concept of the ‘Idea Maze’ make you better at traversing your individual maze?

The short answer is that it doesn’t. The Idea Maze is nice as a post-facto description but is effectively useless when you want to take action. It is a frame that is designed to sound insightful; it is attractive to our brains only because it makes the randomness of reality seem more palatable.

So what do we conclude instead?

The Journey to Product Market Fit is Idiosyncratic

Here’s what I should have said:

The journey to product market fit is idiosyncratic. There is no formula.

In nearly every case of business success, entrepreneurs make ad-hoc decisions to the best of their ability based on information they receive from execution and then look back and go “Huh, I never expected to end up here. But what a journey!”

This is very obvious and easy to say, and I guarantee you that you will nod your head as you read this but you will not believe what I am saying. You will think there are generalisable ‘rules’ that you can follow, certain … ‘good ideas’ that will increase your odds of success, and certain ‘bad ideas’ that will decrease the odds of your finding a successful startup idea.

I think you will believe this — as I did — because most articles about finding product market fit will say that “there is no formula to finding product market fit”, but will then follow that up with “but this is what has worked for me”, or “but here are certain ideas that will increase the odds of your success.”

Notice that I am not saying this second bit. I am saying, quite literally, “product market fit is idiosyncratic and there is no one formula” and I am stopping there. That’s it. I am not offering anything more.

And why should I? The first statement negates the second statement. The only logical conclusion from “… is idiosyncratic” is to not give advice predicated on specific patterns. And yet the majority of ‘how to find PMF’ articles out there will carry on and give you patterns anyway.

(Sharp readers will notice the implication: you should give advice that is not predicated on specific patterns. About which, more in a bit).

One quirk of building the Commoncog Case Library is that we selected cases in a somewhat arbitrary manner: we picked business histories from across the full range of time. We did not start out from some hypothesis of the world when we picked businesses; the goal of reading history is concept instantiation, which means that it’s not smart to set artificial constraints at the outset. And when you have commissioned cases on capitalistic pursuits as varied as Teledyne, Estée Lauder, Vanguard, and Munger, Tolles, Olson … any belief that there are general principles for navigating the Idea Maze quickly vanishes. The reality is that there are countless world-changing businesses that have emerged from (take your pick): nepotism, regulatory capture, post-war dreaming, corruption, curiosity, political revolution, loyalty to colleagues or … revenge.

Don’t believe me? Alright, here’s a challenge: as of press time, there are 19 cases in the Idea Maze concept sequence in the case library. Take any framework or assertion about successful business creation and go through the cases in order. You are almost guaranteed to find a counter-example that contradicts the framework or assertion in some way.

For fun, let’s go through a number of illustrative examples.

Remember Balaji Srinivasan? In his original articulation of the Idea Maze, he asserted that “A good founder is capable of anticipating which turns lead to treasure and which lead to certain death. A bad founder is just running to the entrance of (say) the ‘movies/music/filesharing/P2P’ maze or the ‘photosharing’ maze without any sense for the history of the industry, the players in the maze, the casualties of the past, and the technologies that are likely to move walls and change assumptions.”

Ok, so:

Milton Hershey had zero idea what it took to make milk chocolate. He ran into the idea maze with no sense at all for the history of the industry. He bought a farm and some cows and spent three years literally reinventing the wheel — not even bothering to go to Europe like his (later) rival Forrest Mars; Europe had been making milk chocolate for three decades by that point. By pure luck his stubbornness produced a recipe that created a slightly sour chocolate, and his egalitarian beliefs led him to price his chocolate at a nickel, and he pushed for distribution of the chocolate throughout the entire United States. Decades later, when ‘better’ European chocolates sought to invade Hershey’s market, they faced an uphill challenge thanks to the Branding moat Hershey had built — an entire generation of Americans had grown up used to the sour taste of Hershey’s milk chocolate. (Read the full case here).

Milton Hershey wasn’t just a ‘bad founder’ by Balaji’s estimation; he experimented for three years without launching and had zero idea of what he was doing. It was good that he didn’t sell any equity: his company — now over 100 years old, older than any company Balaji has ever touched and primarily owned by an orphanage Hershey founded — succeeded beyond his wildest dreams. It is still headquartered in Hershey, Pennsylvania, a town Milton created for the company.

Hershey’s story is an existence proof that you don’t have to navigate the Idea Maze intelligently. (In reality, if you read all 19 cases in Commoncog’s Case Library — or even stay on the lookout for instantiations of the concept in real life — you will find that intelligent navigation of the maze is the exception, not the norm). Unfortunately the very concept of the Idea Maze itself begins to fray when you start studying the full span of remarkable businesses.

For instance: Henry Singleton started Teledyne, a high technology conglomerate, in 1959. From the outset, Singleton made it known that he ran Teledyne with no business plan, saying, at one annual meeting: “I know a lot of people have very strong and definite plans that they’ve worked out on all kinds of things, but we’re subject to a tremendous number of outside influences and the vast majority of them cannot be predicted. So my idea is to stay flexible.”

At formation, Singleton’s big idea was to buy underpriced quality technology companies by selling overpriced Teledyne stock. He acquired 130 companies between 1961 and 1969, right up to the point where the bull market for tech conglomerates petered out. Then, as methodically as he bought companies on the way up, he started a program of careful share repurchases from 1971 to 1981, eventually buying back 90% of Teledyne’s outstanding float. Concurrently, Singleton ran his subsidiaries for cash generation, incentivising managers to return capital back to headquarters. This stance — coupled with the share repurchases — generated a 42% compound annual return for Teledyne’s shareholders for the 1971-1981 period. Taken together across the three decades of his tenure, Teledyne generated a 20.4% compound annual return, outperforming the S&P 500’s 8% return over the same period.

It’s worth noting that Teledyne was venture backed — pioneering venture capitalist Arthur Rock invested money at formation and sat on Teledyne’s board; Rock was one of the few people Singleton consulted when he started on his share repurchase program. But it’s worth asking if the Idea Maze even applied to Teledyne. Yes, there were some risks — Singleton’s core insight on disciplined acquisition was unproven when he first started, and his follow-up thinking around share repurchases was wildly contrarian (such repurchases were considered a sign of weakness in the 70s). But the Idea Maze seems farcical when applied to Teledyne. This is notable: Teledyne was one of the most remarkable technology companies of the 70s through to the 90s. (Read the full case on Teledyne here).

Legendary startup founder and investor Paul Graham has a famous essay titled ‘Startups = Growth’, which lays out a shape of successful type of business going through the Idea Maze:

How fast does a company have to grow to be considered a startup? There's no precise answer to that. "Startup" is a pole, not a threshold. Starting one is at first no more than a declaration of one's ambitions. You're committing not just to starting a company, but to starting a fast growing one, and you're thus committing to search for one of the rare ideas of that type. But at first you have no more than commitment.

Graham clearly knows what he’s talking about. And yet:

Sony was founded in post-war Japan, got its start with radio adapters, then primarily made tape recorders for 22 years. They launched the world dominating Trinitron colour television system only in 1968 — 22 years after founding, and the Walkman in 1979, 33 years after founding. (Read that case here).

TSMC was founded by the Taiwanese government. Founding CEO Morris Chang looked out at the landscape of possible semiconductor companies and decided to create what is now called a ‘pure-play foundry’ — a semiconductor company that did no design, and would never compete with any of their customers to sell chips of their own. They hired only process engineers, no design engineers, and committed to manufacture chips for other companies. In the 80s, this had never been done before: the vast majority of semiconductor companies owned their own foundries; the industry wisdom was “real men have fabs” — as AMD founder Jerry Sanders so memorably put it. TSMC’s growth was terribly slow in the early years. It had to fight mainstream wisdom, and also it had to demonstrate that it would never compete on chip design. A business model that is built on trust takes a lot of time.

TSMC started in 1985. It grew anaemically — customer by hard-won customer — for the first decade of its life. In the late 90s there was a sudden boom of graphics chip companies: at its peak, there were 70 startups duking it out in a ruthless fight for survival. Of the 70 companies, only two survived: ATI and Nvidia. TSMC rode this wave of growth up, and then — as its customers started dying on the other end of the capital cycle — used the revenue it made from the boom to reinvest in technology and fabs. (This is a pleasant way of saying that its revenue fell off a cliff right when it ramped up its discretionary capital expenditure). Its technology caught up with leading chip manufacturers only in the early 2000s. By the late 2000s, its competitive set dwindled to 14 companies, and then in the mid-2010s, it was down to six. Today, TSMC is one of only two cutting edge semiconductor manufacturers — and it outperforms and outproduces Samsung, the remaining competitor. TSMC has nearly total pricing power. Startups may be built for growth, but it was unclear that TSMC would be this large or this dominant at any point up to the late 2010s — a full 25 years after founding. In 2022, it was Asia’s single most valuable company. (Read the full case here).

Danny Meyer ran a restaurant business for 15 years before starting a hot dog stand as part of a Madison Square Park art project. Four years later, in 2004, the city opened bids for a kiosk-style restaurant in the Park; Meyer improvised on the original stand and submitted a bid for a new concept he called ‘the modern version of a roadside burger stand’. While Meyer’s early dealings with Shake Shack are private, we know that up to 2008, at least, Shake Shack was operated under the holding company that held fine dining restaurant Eleven Madison Park. Sometime after 2008, the burger brand was taken out of its holding company, recapitalised by private investors, and began expanding. It IPOed in 2015 and is now worth around four billion dollars. (Read the full case here).

At which point were any of these companies startups, and at which point did any of these companies commit to search for a fast growing idea? For Meyer, would you say it was when Shake Shack was recapitalised for growth, at least nine years after it started as a cart? For Chang, was it when TSMC caught up to other chip manufacturers in the early 2000s, after one decade of methodical execution and a dramatic capital cycle bust? At which point in Sony’s history were they a startup? Recall that for the first 22 years they were mostly an overlooked player: a tape recording manufacturer and budding consumer electronics company, with some patents to their name, duking it out with the rest. At what point would you have said that it had ‘found product-market fit’ and was ‘destined to grow fast’?

There are companies that fit Graham’s essay. He has funded many of them. But then there are also many companies that do not fit the essay. In epistemology, if a trait predicts success, you have ‘knowledge’ — you know something about the world because you have isolated an identifiable trait with predictive validity. However, if cases without the trait are also successful, then the trait is not very predictive. You should regard it with some suspicion.

To be fair, people who come up with frameworks usually have some context in which they are working or thinking, and their frameworks make sense for their specific context. (Graham has to invest in companies given a specific context; you, however, do not need to start companies in that specific context). But if folk were serious about making universal pronouncements, they would sound less like framework thinkers and instead start sounding like … well, Charlie Munger, I guess. Munger always drove me nuts with his unwillingness to make first-principles-based pronouncements on business success. He nearly always talked in terms of “this is something that can happen” or “here is a mental model that illuminates this specific scenario”, and then he would always, always, throw an analogous case at you.

But I now know why he does this, of course. Munger has internalised — more than nearly anyone else — that finding a path to a working business is always unique. You have no choice but to resort to ‘general patterns that can happen’ because business is too varied to draw universal conclusions. And the quickest way to internalise this is to go read a bunch of business biographies.

There are a lot more stories where these come from. You should go read some of them.

Context-Free Principles for Navigating The Idea Maze

If you put aside the lie of ‘universal frameworks’, or even ‘recommendations for navigating the Idea Maze’ then a natural question that arises is … what works? What generalisations can you make across all the cases?

What you get boils down to more conservative things like “you must find a secret” or “you must get lucky”.

”Get lucky?” I hear you say, “How is this better advice?!”

Counter-intuitively, “go get lucky” is better advice. It is honest about the shape of reality. It is universal across all cases. Most importantly, it is context-free.

So how do you get lucky?

The answer to this question is one that we’ve already examined on Commoncog, thanks to the work of academic Saras Sarasvathy. It turns out that Sarasvathy’s work is the only thing I’ve found that actually fits with the full range of evidence.

To summarise quickly, Sarasvathy picked 27 ‘expert’ entrepreneurs from a list of founders who were active between 1960-1985 (criterion: “either as individuals or as part of a team, had founded one or more companies, and remained with at least one company they founded for more than ten years and taken it public”). She discovered that all of them thought about entrepreneurship in the exact same way: they do something called ‘effectual thinking’.

Effectual thinking is the opposite of causal thinking:

• If you use causal thinking, you’ll say something like “ok, we’re making carbonara tonight” and then you will work backwards from the end goal (carbonara for, say, five people) to checking for ingredients in your kitchen, to purchasing the ingredients you don’t have, to prepping and cooking carbonara for your dinner party.
• If you use effectual thinking, you’ll say something like “ok, what ingredients and tools do I have right now, and what can I make tonight?” You work forwards from existing resources; the end product is unknown.

To be more specific, entrepreneurs who practice this form of thinking follow three principles, and these are the principles that are actually useful for navigating the Idea Maze:

• While causal reasoning focuses on expected return, effectual reasoning emphasizes affordable loss;
• While causal reasoning depends upon competitive analyses, effectual reasoning is built upon strategic partnerships; and,
• While causal reasoning urges the exploitation of pre-existing knowledge and prediction, effectual reasoning stresses the leveraging of contingencies.

To make these more actionable:

1. Entrepreneurs structure their lives to enable them to take reasonable risks, ensuring that their losses never take them out of the game. They then place many survivable bets over a multi-decade time horizon. Structuring their lives like this makes success more likely: eventually one works out.
2. Instead of doing competitive analysis, in the early stages of a venture they focus on building partnerships with customers or other stakeholders. They do this because market analysis is simply not very useful: a lucrative new business must exploit something that isn’t yet known; if analysis could uncover that information, it is likely not unknown enough to be exploitable. Taking action in a market generates far better, more accurate information of the type that might actually be valuable to the entrepreneur: it teaches you what gaps are actually good.
3. Good entrepreneurs understand that the whole game of entrepreneurship is a game of improvisation. There is no knowledge to be had; no advice one can read. There’s simply no guaranteed answer to anything: you take lots of action to generate information and stay alive, and then roll with whatever comes your way. You are prepared to do this for years.

These principles are the only universal thing across all the cases in the Idea Maze concept sequence. In pretty much every story, you will see that at some point a bet starts to gain traction, and it works because the person has hung around long enough for a gap to emerge; their latest bet exploits that gap when it still isn’t widely recognised. The only thing that uncovers the gap is action — this is the ‘discovering a secret’ observation that Peter Thiel talks about. But how you uncover a secret is idiosyncratic. There is no ex-ante strategy for finding a secret beyond “fuck around and find out”.

Which, ok, maybe we can do slightly better than that. The better advice might be: “be curious, develop good taste in businesses and in products by consuming lots of cases, and then put yourself in a situation where you can keep messing around for years and years.”

Aim to get lucky.

Two More Things

The ‘Idea Maze’ is a useless idea for execution, and you’re better off learning to think effectually. What else can we conclude from this essay?

There are at least two other implications that fall out of these ideas, which I think should become obvious if you stop to think for a bit.

First, you should probably realise that much of the bloviating about product market fit is nothing more than content marketing … designed to a) attract attention to a founder, if they’re currently running a startup, or b) attract deal flow to a private markets investor. They are not actually useful to you as an individual, if you are seeking out a working business.

To be fair, I think this has become fairly common knowledge. If you have found product market fit, your problems change and there’s typically expertise and frameworks you can tap to solve the problems you encounter. But before you find product market fit, it is fairly well known — at least in startup circles — that no amount of advice would do anything for you: there is no amount of credentialed people you can find who can help you; hanging out with other founders actually don’t matter; it is irrelevant how many IOI gold-medalists you have on your team; you are well and truly on your own.

This can be quite liberating! Nobody knows anything pre-product-market-fit, so it’s up to you to figure it out. You — and only you — are responsible for your outcomes here. For a certain personality type, this is amazing.

(Yes, this bit of the startup journey can also get lonely, but such is the price of entry. Don’t lose hope. Also, welcome to the club.)

There is a second observation we can make.

It is common to find individuals in the startup world who ‘have taken a year off to find an idea for a company’. Like me, you may have noticed that this tends to be less effective than ‘working on a side project for several months or years before starting a company around it’ or ‘faff around with random projects for an indeterminate period until something catches on.’

I am generally sympathetic to such ‘start a company for a year’ attempts, because I think people should try to start companies if they are interested in entrepreneurship. But I now think that any attempt with a deadline is not a great way to start a new business.

This is not something I would have said even a few months ago. The cases on the Idea Maze have changed my mind — and perhaps they might change yours. The simple truth is that finding a good business takes time. Sometimes it takes a decade. It is trite to say this, but reading 19 cases end-to-end has a way of driving this home, in a way that simply asserting it might not.

Accepting this fact has certain profound implications for the way you might want to conduct your search.

• You don’t want to set a deadline for yourself. Deadlines are fake. The more realistic approach is: “how can I structure my life to conduct a years-long search for a workable business idea?” I do have thoughts on how to structure one’s life, but for now we can conclude that it should be unique to your circumstances; read the case studies for some concrete examples.
• You also want to be maximally open to opportunities. This is not a trivial thing. Deadlines have a curious ability to shut down your curiosity, to put blinders where you should be exploring open vistas, to focus on coming up with a startup idea NOW. This almost guarantees a reduction in luck. It also blocks you from tapping into the power of personal obsession.
• Paradoxically, deadlines also prevent you from fully committing to an idea. You only have one good shot, after all. What if you waste it on a dumb business? Your self-imposed date looms large in your head, and so you fret. And then you hedge. And the simple act of hedging prevents you from making any single idea actually work.

It’s funny, but “I spent two years travelling cheaply through Africa, hacking on interesting ideas and observing the people around me” is more likely to lead to a interesting business life than “I am going to quit my job and iterate on ideas in an attempt to start a startup.” In most places in the world, “I am going to start a company without venture capital first” has more configurations that might work than “I will raise money to find an idea that works.”

This might seem weird to you. But perhaps it shouldn’t. The Idea Maze is such a crapshoot that you’ll need all the luck you can get. Exploring with an open stance increases luck. Keeping your capital structure clean increases luck. Paradoxically, commitment increases luck.

But perhaps the most surprising thing is that there are no rules, except for what you uncover for yourself. Good luck.

Postscript: Prior Art That’s Actually Useful

A couple of links on product market fit in technology that are actually useful:

• Excuse Me? Is There a Problem by Jason Cohen.
• Four Fits for 100M Growth by Brian Balfour

There tends to be good work on how to know you’ve achieved product market fit. There is nothing on how to actually find it. But so it goes.

Originally published 28 August 2024, last updated 31 August 2024.