Card 1 of 16: Do Things That Don't Scale
Paul Graham's definitive advice for early-stage startups: you cannot wait for users to naturally discover your product. Founders must manually, and sometimes painstakingly, recruit their first users one by one and do things that are structurally unscalable.
A startup isn't a smaller version of a big company — it's an organization searching for a business model that doesn't exist yet, under extreme uncertainty. Because of that, the traditional approach of writing a detailed plan and executing it for years is dangerous: you're just executing a guess. Instead, Ries argues you should build the smallest possible version of your product (a Minimum Viable Product), put it in front of real customers immediately, measure what actually happens, and use that data to decide whether to persevere or change direction ('pivot'). Repeat that loop as fast as possible.
Graham's core claim is that you can't reliably think up good startup ideas by sitting down and brainstorming — the process of deliberately 'coming up with startup ideas' tends to produce bad, generic ones. Instead, the best ideas are noticed, not invented: they show up as a byproduct of living close to some frontier (an emerging technology, an unusual expertise, a niche you're deep in) and personally running into a problem sharp enough that you build a fix for yourself first. Good ideas often sound bad or small to outsiders at first, precisely because the market doesn't exist yet in an obvious way.
You don't need to be a mass-market hit to make a good living as a creator or a small niche business. Kelly's math: if you have 1,000 people who are true fans — meaning they'll buy essentially anything you produce — and each one spends roughly $100 a year with you, that's $100,000 a year, enough to sustain a creator or a small operation directly, with no label, publisher, or platform gatekeeper required in between.
In a perfectly competitive market, economic theory says profits get driven to zero — everyone copies everyone else until nobody makes real money. Thiel's contrarian claim: durable, large value is created by escaping competition entirely and building something close to a monopoly — a business so differentiated that it isn't really competing with anyone. He argues that 'true progress' comes from vertical moves (0 to 1: creating something genuinely new) rather than horizontal moves (1 to n: copying and incrementally improving what already exists in a crowded field).
'Total Addressable Market' (TAM) is often used as a single big number to justify a startup's potential — 'the market is $50 billion, so we could be huge.' Gurley's point is that the size of the market on paper tells you almost nothing on its own; what matters is the structural quality of that specific market — things like whether the product is used frequently or rarely, whether the marketplace naturally consolidates around one winner, whether there's real network effects, and whether new supply/demand can be added cheaply. Two markets can have the identical dollar-figure TAM and be wildly different investments.
Subscription software businesses (SaaS) don't get paid all at once like a normal sale — they get a little bit of revenue every month for as long as the customer stays. That changes what actually matters to track. Skok's essay lays out the core numbers every SaaS founder and investor should watch: how much it costs to acquire a customer (CAC), how much that customer is worth over their whole lifetime (LTV), how fast customers cancel (churn), and how efficiently each new dollar of sales/marketing spend turns into new recurring revenue (the 'magic number'). Get the relationship between these numbers wrong, and a SaaS company can look like it's growing while actually burning cash faster than it can ever earn it back.
Normally, businesses grow carefully: prove a model works efficiently in one place, then expand. Blitzscaling argues that in markets where being first and biggest creates a durable, self-reinforcing advantage (network effects, brand, data), it can be rational to deliberately sacrifice efficiency — burn more cash, tolerate more chaos and waste — in order to grow faster than an efficient competitor could, because in a winner-take-most market, being the fast, slightly wasteful first mover beats being the careful, efficient second mover.
Before the internet, distribution was expensive and controlled by whoever owned the physical channel (a newspaper's printing press, a store's shelf space) — so suppliers had the power. Thompson's theory: online, distribution is nearly free, so power flips to whoever owns the relationship with the end user's attention and preference. Companies like Google, Facebook, and Amazon win not by owning supply (they often don't make the content or products themselves) but by aggregating demand — becoming the place users default to — which lets them dictate terms to suppliers, who become interchangeable ('commoditized') behind the scenes.
An 'asymptote' is a ceiling a curve approaches but a system doesn't realize is there until growth mysteriously stalls. Wei's argument: every business has hidden structural ceilings built into its own model — a limit on how big it can get given its current mechanics — and the ceiling is usually invisible until you're already slamming into it. The best operators try to spot these ceilings in advance (by understanding the mechanics deeply, not just watching the metrics) and re-architect the business before growth actually stalls, rather than being blindsided by a plateau they can't explain.