In the world of high angle rope rescue, we have a concept called the "whistle test". The idea is that if someone were to randomly blow a whistle at _any_ point, and everyone let go of whatever they were holding, that no one would be dropped.
It takes a lot of thought and planning to make sure you're 'default alive' in all circumstances. It slows you down, and it requires you to think through the implications of every decision, big and small.
Here's an interesting thought in response: (Keep in mind I know jack about rope rescue, so I'm making assumption as to what it is)
Imagine you had 100 people do a "rope rescue race". Half of them used all the safety mechanisms (they would pass the "whistle test"). The other half didn't use any safety/redundancy (they would fail the "whistle test"), but instead relied on their skill. In the end, it's likely that a few of the high-risk ones would die. But it's also likely that many of the high-risk ones would survive, and also that they would be the winners of the race. The safer, lower risk ones would eventually get there, but far behind the high risk ones.
I don't rope rescue, but I sure do rock climb. You can rock climb with ropes or without. Ropes = a large safety margin, and the majority of people who climb, utilize them, as part of their safety system. Unless you're pushing your own grade level, the actual safety system isn't actuated - you're just using it "just in case" (like, "just in case you fall from a cliff, 500 feet up - gee it's nice I'm tied to something).
There are also those who climb without ropes. Perhaps you've heard of some of them [0], and most likely, because they actually climb without a rope. High Risk. Your idea of startups, let's say.
But that's just what the public sees. The free soloist has to be absolutely comfortable to go without the safety margin of a rope. Because of that, the majority of their climbing is done, still with a rope. When they go ropeless, the climbs they do are much, much easier. Yes, the risk is there, but also there is the understanding, clearly, what the risk is really about.
The death rate of climbers that are unwilling to ever use a rope, because they find that high risk = high reward would be close to 100%. But those who do survive are not going to be doing too well - they're never going to progress.
So, perhaps startups have this illusion of high risk, but maybe also they're also kidding themselves. They're burning through their lives, and offloading the real risk onto their safety systems - whoever is putting up the money. And they can do that (the investors), as they've diversified their portfolio enough that risk is spread around.
The analogy breaks right at the start, start-ups are not 'fatal' in the same sense that rock-climbing is fatal. So the GP is right, the race really would be won by those willing to cut corners and sacrifice safety and procedure even though statistically quite a few of those would end up 'dead' for those exact same reasons with possible fall-out for their end-users as well (hacked, data loss and so on).
The strategy widens the bell curve and lowers it, whereas the rock-climbers try very hard to narrow it to where there are no outliers and everybody makes it to safety.
Imagine a strategy where every start-up would survive in the long term, there would be very few outliers in such a situation and that's why start-ups that are comfortable with some risks at the expense of safety but a huge increase in productivity will be more likely to be amongst the winners than those that play it safe all the time.
I don't think this is a huge problem anyway because very few people doing start-ups are risk averse.
You raise a good point. It's not a matter of caution and being double careful. The point of the essay is the drive towards self sufficiency, and being realistic about how investor expectations change in successive rounds.
Perhaps: don't think there's a rope backing you up if it isn't there.
> Perhaps: don't think there's a rope backing you up if it isn't there.
If you use the metaphor of rock climbing, you'd actually rehearse your route with ropes/safety system, before you would ever try the route without. When the stakes are so high that it's you're life you're gambling with, is it really worth it for most to go so blind out there?
It's either that, or you go without a rope, when the difficulty is far below your limit (in other words, you have amassed a lot of experience)
For software, the rope is the engineering. Whenever I need to move fast in a production environment I make damm sure my software is as good as I can make it because I know something will come and bite me in the ass otherwise. I know patience and carefull design makes me fast in the long run. I've done projects in short span and they were successfull precisely because of the quality I put in the code, and aggressive application of YAGNI.
Note: good here means understandable and simple, not some god awful super-elegant object abstraction scheme.
It might seem people with cowboy coding skills do well but it's only an illusion - they have faced as many bugs but they just code around them or do a quick fix.
The result from the latter is unmaintainable spaghetti.
Been there. It was a death spiral. The investors (which where customers, too) always wanted more features (of which not half of them made sense) so the directive given was just throw something at them that remotely resembles that requested feature to get the money. In the end the product was very unappealing and way too complex to sell to anyone else than this particular investor/customer. And as I heard recently they hired two more sales people in order to fix that. So what I'm trying to say is that not only you have to make sure your code base has a certain level of quality (fsloth's point) but this goes also for the product.
I disagree that the problem is cowboy coding. I'd identify the rope as carefully understanding the actual goals of the software and the way it functions in the world. Good engineering teams going "ropeless" rarely fuck up in a technical sense. I've been writing software long enough that my engineering judgement is usually pretty solid. Where they fuck up is building tons of extraneous functionality with the extra capacity, losing touch with what their software actually does in the world.
It very much depends on the situation. In an engineering environment (industrial control, regular applications, aerospace and so on) slower really is faster. But in a SAAS prototyping race with a whole bunch of like-minded characters those who cut corners may very well win the race. This is frustrating to those of us (including me!) who would rather do a proper job but the 'cowboys' seem to be both getting the funding and seem to be statistically doing better at success in the longer term. Of course plenty of the cowboys die with self inflicted bullet holes but the press does not seem to care and neither do investors. It's there on every level from 'go big or go home' to 'don't be scared to break stuff' to 'oh, we went bust, sorry customers, we'll try better next time'.
In a nutshell this is the conflict of interest between founders and investors, even though quite a few founders would rather have a 1:10 chance of a huge success than a 1:2 chance of a medium one.
eh, the thing about margins of safety is that when done properly, they let you recover from mistakes, which lets you take risks more often. I see a break-even business as a 'save point' almost - you can go and put effort then into this or that or the other that might or might not work out, but as long as you've got the break even business paying everyone's rent, you fall back to a place where you still have all your infrastructure ready to go for your next try.
If your company is 'default dead' as it were, a schedule slip (a slip past the end of your runway, that is) means, at best, bankruptcy and starting over.
And really, you want some safety built into that too- if you screw up your taxes, or screw up your corporation badly enough, failure can mean personal bankruptcy, or if it is a tax matter, something much worse than personal bankruptcy.
This is a good analogy in one sense, but I think the probabilities in your example are reversed from the startup case: for high-risk startups, only a few survive, but they're the winners.
And of course, that's what makes them better on average for investors (who spread risk across their portfolio) than for founders, who have all their eggs in one basket.
As pg says here, as a founder, you want to survive.
I'm not sure if that's an apt metaphor because the consequence of failure is much, much lower on the startup side.
If failing at a startup means you'll never have another chance to succeed, I bet the norms for risk tolerance would be entirely different than what's in the landscape today.
Your premise is based on the assumption that those who win in the marketplace are those who move fastest, not those who have the better product, or make something people love.
I think that relatively few startups would get started if they all had to be 'default alive' - founding a startup would be limited to folks who were already independently wealthy. The point of the article, I think, is that if you're 'default dead', you should know that you're default dead, and your attention should be focused on getting to a state where you're 'default alive'.
On the other hand, Facebook and Google were both founded as "default alive" - the startup part didn't happen until the founders were certain they had something so good they should drop out of school to do it.
They also didn't have much in the way of worthy competition. That really helped them out a lot. It's different now. So much saturation in many areas things have moved from the era of bold ideas to the era of carving out niches.
Facebook (launch Feb 2004) was founded after Myspace (launch Aug 2003), which was founded after Friendster (2002). So there was certainly competition.
Was it 'worthy competition'? Hard to say without hindsight bias (only the winners 10 years later were 'worthy'), but I doubt Mark Zuckerberg or the Google founders felt there was no worthy competition. It seems fairly clear to me that Facebook (and Google) succeeded because they made better decisions (on average) at a gazillion small decision points, not because they had some great idea that no-one else had thought of.
I remember seeing a talk by Zuckerberg (or another early Facebook person) where they talked about clones of Facebook, and how they were racing to win over different colleges / universities before other social networks did.
I remember one particular college-targeted social media site (besides Facebook) at the time called CollegeClub. Met a couple women through that. It didn't have anywhere near the reach that Facebook did a couple years later, though.
Oddly enough, Facebook was founded at almost exactly the same time as a fast-growing social networking site backed by a famous company with global reach. That was Google's Orkut.
From Bessemer's anti-portfolio:
Cowan’s college friend rented her garage to Sergey and Larry for their first year. In 1999 and 2000 she tried to introduce Cowan to “these two really smart Stanford students writing a search engine”. Students? A new search engine? In the most important moment ever for Bessemer’s anti-portfolio, Cowan asked her, “How can I get out of this house without going anywhere near your garage?”
Wrong on both counts. Both companies entered markets considered completely saturated.
Friendster and Myspace were the big players for Facebook, Lycos, AltaVista, and HotBot were competitors for Google. A new social network or search engine didn't get noticed back then.
Google was founded 'default dead' (they only changed that when they got into advertising), Youtube was default dead for many years even after being acquired.
Absolutely. My point in making the analogy is that 'default alive' is _very_ expensive, and a lot of work. It's probably not worth it in an early stage startup. The stakes just aren't that high.
I think it heavily depends on your product. If you have something that can be used by a single person (or small group) without a million others to use it, too, than you're IMHO default alive (Slack, Trello, ...). I think as a rule of thumb if you have to first get traction and then be able to monetize than you're default dead. Instagram's social component was default dead but the feature where you can alter your photos was default alive and that pulled them through.
What do you count as independently wealthy? Middle class (1-3 years savings, good education, professional job prospects) to successful startup founder is not uncommon enough to call an outlier - I personally know at least half a dozen people with that background and story. It's not exactly easy, but people can do it.
It is much harder for working-class folks (no savings, no college degree, no unique skills) to make the jump.
Pretty much by definition, yea. If you can actually afford to leave your job (in other words are independent of it) for 1-3 years and go off and start something that might not even make money, you're pretty wealthy.
In a much less serious vein, I had a boss who used to joke about throwing our laptops on the ground at work, saying "if you're not up and running on another machine in 10 minutes, you're doing it wrong"
I can do that at work, because all my code is on an EC2 instance, and I use emacs on the remote box as my editor. I've accidentally logged out of my X session and been back up and running in less than a minute before.
How often is this put to the test? How do you automatically record configuration changes like the occasional "apt-get" to bring in something you're missing?
We use chef to record all our config. Our chef configuration is pretty stable these days, but it gets tested every time we build a new dev box, which happens at least as often as we hire new engineers. I could completely rebuild my dev box at any minute and the only thing I'd lose would be unpushed changes I was working on at that minute.
What's the most common reason to fail a whistle test?
If a real life "whistle test" situation were to occur and everyone let go without being 100% secure, what would happen? how many people would fall?
This idea is really interesting to me, but I'm not sure why.
The general idea is that whatever mechanism is allowing the ropes to move is something that 'fails closed'. The simplest mechanism is to wrap two prusiks[1] around the rope that is moving. Those prusiks are then securely attached to an anchor or some sort. Left alone, the prusiks will 'grab' the rope and prevent it from moving. Someone has to be actively holding them in position to allow the rope to move (otherwise they'll slide with the rope and pull taught).
In other cases, it's a question of how you set up things like descenders[2]. The same basic principle applies though. You need to be actively holding it open in order for the rope to move through it. There are a few exceptions to this (things like figure-8 descenders[3]), but in those cases, they are being used by a single operator to control their own descent.
The most common situations where this goes wrong are during transitions. The system used to lower someone over the edge is different from the system used to haul them back up. This exercise is fairly simple, but it requires a great deal of attention to detail, and the steps have to be done in the proper order to ensure there are no 'gaps'.
The number of people who would fall would depend completely on the operation. Generally speaking you have a single medic over the edge, who may have the victim clipped in to them if they're on the way back up. In some cases, you may have one or more additional rescuers.
In the world of industrial rope-access, it'd be illegal in most jurisdictions to use any rope system that wasn't 100% safe to fall in to at any given time.
PSA: Rather than using underscores, _like this_, at HN you can get emphasized text using asterisks, like this. (I see _emphasis_ almost every day on HN, so IMHO it should support both. But until it does, asterisks are definitely the way to go.)
some of us think it should support neither (because _emphasis_ is only slightly less legible than emphasis whereas erroneous use of asterisks is very much less legible than uses of asterisks would be if asterisks were not interpreted -- and because erroneous use of asterisks is fairly common).
But _if_ the present situation continues in which asterisks (or asterisks and underscores) are interpreted, at least the interface should provide a link next to the "ago", "parent", and "flag" links that show the raw, uninterpreted comment the way the user entered it.
> The startling thing is how often the founders themselves don't know. Half the founders I talk to don't know whether they're default alive or default dead.
This accurately reflects my own experience. Most common pitfall: converting VC capital to users at a rate that will not sustain the company once the VC capital runs out. So many companies fall into this particular trap that it should have a name of its own.
Bought growth is only worth it if the users remain long enough to make back the money you pumped into them at the time of acquisition in net profits otherwise you might as well do without them.
I'm not sure if the reference to airbnb helps, whatever they did, they're an outlier and simply doing what they did without carefully evaluating your reasons is going to work about as well as any other cargo-cult strategy to success, it would be (a lot) more useful to see this point expressed in an alternate form, start-ups funded by YC in cohorts of months from when they started hiring besides the founders compared to their survival rate.
My suspicion is most founders actually do have a pretty good idea of the answer on some level of consciousness, but avoid thinking about because it's terrifying.
I'd lay heavy odd the subconscious thought process goes along the lines of "If I'm default dead I have no idea how to fix it and so I feel powerless about it and that terrifies me and so I don't want to think about it and so I don't know the answer."
Not that that's exactly a successful strategy, but people don't always lock on successful strategies.
As a founder, I know all too well how easy it can be to ignore these potential issues. The funny thing is when you actually confront them and analyze your situation, you gain control. It's more relieving knowing than ignoring (because you never really ignore something like this without knowing you're doing it).
That's very strongly tied into personal responsibility, if you identify a problem and you tie it to your own performance you essentially look the problem in the eye and you say 'I can fix this'. If you ignore the problem or make yourself believe it is someone else's job or problem to fix this then you make it impossible for yourself to do something about it which hands control of your future to outsiders. The more problems you're willing to confront and analyze the bigger your chances of succeeding because all of those will be your problems and therefore very likely fixable.
Success is more random than your suggesting. IMO, the goal is to take risks that are in your favor. The less you know the harder this becomes.
If your default alive then you should consider being somewhat conservative. But, if your default dead then you need to risk something because your already losing.
The randomness of luck only comes into play when you've done everything else right, it's one huge 'AND' gate with luck as one of the factors. If you mess up on one of the others you're still dead no matter how lucky you were.
So by improving the rest of the gate functions the importance of the luck element goes up because it may be at some point the only item between you and success and then if it swings your way you're suddenly doing very well.
See also: the myth of the overnight success. All that is is a ton of preparatory work and one opportunity properly seized.
The calculation isn't that simple to calculate, it really requires a spreadsheet or calculus, you can't in general calculate it in your head. A simple heuristic is "is my runway going down or up", it is conservative.
It's amazing to me how deeply ingrained software profit margins are into the start-up world. That calculator...we're an ecommerce company that holds inventory...I spent 30 seconds searching for how to set gross margins on the revenue then realized it assumes all revenue is 100% gross margin. In most businesses (read: anything other than software and maybe pharma), manipulating margin is one of the biggest levers (maybe THE biggest) you have to affect profitability.
Not to mention other big levers like working capital (and potentially running a business with negative WC and generating cash, a la Amazon). It's funny to be running a start-up in SF and still feel a world apart from a lot of the ecosystem.
Is there a better word? Net revenue? It should be the bottom line of everything that scales, while the red line is the bottom line of everything that's O(1), like engineering. GAAP doesn't seem to have a specific word for that.
Gross profits is net revenue minus cost of goods sold. Cost of goods sold is only the direct cost of goods, and does not include any operating expenses.
Net revenue is Gross revenues minus discounts, returns and allowances.
> In practice there is surprisingly little connection between how much a startup spends and how fast it grows. When a startup grows fast it's usually because the product hits a nerve, in the sense of hitting some big need straight on.
Perhaps the most important underlying point in the article.
It's easy to think that more people will make the company grow faster. Adding people actually makes it harder to tune a product's direction (and thus growth rate). Great to see another dense and on-point post from pg. Every sentence is worth several reads.
I think that having more people also makes it harder to focus. When it is just one or two folks, you have to ruthlessly prioritize. I am working on a side project and we are at the point that I only want to do things that either bring in money or extend our reach (in terms of data). Because it is a side project, it is hard to be ruthless, but it feel it's what is needed. The pressure on a full time founder must be 10x or100x what I feel.
Well, this is mostly only true of the high margin, hoping-for-viral-growth so loved by the internet VCs.
There are a ton of highly profitable, relatively low margin businesses (at least compared to internet services), like consulting or something with high sales effort, like enterprise products. In those cases, you can easily imagine your growth limited by, and accelerated by, your ability to hire and grow the right talent.
VCs aren't merely hoping for hypergrowth, their business model demands it. The results for any given VC fund are dominated by 1 to 3 hypergrowth companies. Therefore, hypergrowth companies should be their sole focus.
You're right that consulting companies can be great businesses. A friend of mine sold a 5 year old consulting company for a good tens of millions last year. That's a stellar result for the founders. But few consulting companies are within a VC's "strike zone".
> Instead you'll be compelled to seek growth in other ways. For example, by doing things that don't scale, or by redesigning the product in the way only founders can. And for many if not most startups, these paths to growth will be the ones that actually work.
Or you could reconsider the size of your total addressable market (hint: it's probably a lot smaller than what's in your pitch deck) and give weight to building a smaller company that's sustainably profitable.
Note that I'm not suggesting growth isn't important. What I am suggesting is that a lot of founders seek "Silicon Valley growth" without considering the possibility that they have an opportunity to build a lasting business that doesn't need hundreds of employees, tens of millions of dollars in funding, hundreds of millions in revenue and billions in enterprise value to succeed.
Most of your examples have raised large amounts of VC funding. It's debatable whether they "need" the funding to success, but they certainly wanted it enough to eat the dilution.
They have raised, but they didn't raise until they were profitable -- that is, WELL past "default alive". They are remarkably different in this respect than essentially every YC startup.
Pretty sure Fog Creek has never raised either, while having 50-100 employees.
The overall point is that you don't have to be a one-person business to grow at a slower but "healthy" rate.
With the above examples: imgur has raised $40 million and can barely keep one website running. GitHub has raised $350 million and can't seem to add new features or fix existing features. 37signals did fake funding just to get connections, but they at least isolated themselves and stopped trying to infect the rest of their world in their hype bubble.
Like most "new" ideas in the Startup world, what's new is old already. This is just people not understanding how their cash flows affect their balance sheet. Does the business turn a profit? Is the profit greater than their payroll? How many founders know their EBIDTA?
And even more, when growing the business, are you growing costs faster than profits? Yes, this employee is necessary. But will they add more profit than they cost, at this stage in the business; or should we hire this person in a few months?
This is because many founders get used to being able to go back and ask for more money if things aren't going well...right up until they get turned down. I'm being unfair, but it's always seemed like investors play a parent, and the founder plays a child, and it's just a child asking their parent for more money. And I honestly believe this level of entitlement is necessary to be a successful founder.
> Does the business turn a profit? Is the profit greater than their payroll? How many founders know their EBIDTA?
The first two questions are considerably more important than the third. That many founders do not know answers to the first two is less excusable than knowing a fairly specific accounting metric.
Calling them old is maybe a bit much but lots of start-ups never make it to the 1 year mark, 8 or 9 months is doing pretty good especially if the company hasn't yet had any outside investment.
Not to quibble, but he didn't refer to them as old. He said from that point on, the first thing he wants to know if whether they are default dead, or default alive.
There is a lot of truth in this essay but readers should know how to interpret it. When PG assumes stable revenue growth for "the last several months", even just revenue at all, for a startup that's been operating for not even a year, that already narrows down the set of startups for which his advice is applicable quite a bit.
What we're essentially talking about then is a specific breed of startup (a popular breed, but still a specific breed):
- Pure software / probably SaaS or a website.
- Margin close to 100%.
- Essentially no R&D component, i.e. just wiring together existing technologies.
- Perfect product distribution infrastructure, i.e. internet or mobile app company.
What readers should know however is that outside of this subset of startups there is a wide variety of flowers in the startup ecosystem that can all bloom in different ways and at different times.
And that early growth isn't the only metric that makes a startup attractive. Growth a powerful one that eventually will be more important than everything else, but there are other things -- such as strong core IP or having early product in a small but growing market -- that make startups interesting.
Then there's startups that require literally tens of millions of dollars of up-front investment and years of R&D before there even is a product. Are those all doomed? I guess not. Given the right execution and strategy and patient investors these companies can also be extraordinarily valuable. They are just a different kind though.
But as a founder your incentives are different. You want above all to survive.
This is a bit like in the early stages of a poker tournament, where you might fold even quite strong starting hands to all-in bets where your expected value is positive - because you're not just betting the number of chips in your stack, you're betting the entire remainder of your tournament.
It's interesting to see the tone change from Paul Graham and Sam Altman in the last couple of months, its almost like somebody finally bought them a calculator.
Doesn't seem that way to me. pg has been saying much the same things for years—e.g. this essay is obviously a continuation of http://paulgraham.com/pinch.html, and that one continued from others I'm too lazy to look up. Sam has been saying the same thing about hiring for years. Trevor's calculator that the essay links to is old as well.
I'd say three of his last four posts are more negative than he has been historically. I've also observed an increasing number of articles about properly calculating financials, unit economics, and hiring "killer sales teams." I don't know that it means anything, but there does seem to more of that kind of chatter - and not from the buzzfeed "Bubble!!!!!!" kind of sources that its been pouring from for years.
Perhaps they are having to dispense this advice more often in their day to day mentoring. I could see how the growth and success of YC could breed magical thinking among its participants.
That's my thinking too. Startups have become more "sexy" since The Social Network and such. So YC is probably seeing a lot more naive/opportunistic gold-rush type folks than before.
For the first dot-com boom, I did Downside's Deathwatch[1], which did exactly that for public companies. (Companies IPOed earlier in that boom, often before profitability.) For a public company, SEC filings give anyone enough info to make that calculation.
For a private company, it's much harder to tell from the outside. Any CEO who doesn't know how many months (days?) of cash they have left is hopeless.
That's because it always takes some kind of "celebrity" to raise awareness :-) It's good for you that you are aware of this but a lot of people (sadly) aren't.
If Twitter would have heeded this advice, it may not be in the bind that it's in right now. Four thousand employees! At Twitter! That's an order of magnitude more than they need.
What about "Default no progress forever."? Let's suppose someone is building a new airplane and has a hundred subscribers paying $10 per year for their newsletter. Nobody is funding this person, and the business is alive forever. Default alive or default dead?
It's alive until the founder realizes he will never raise a round capable of building an airplane, and pivots and does something cheap and easy online instead.
> Say "We're default dead, but we're counting on investors to save us." Maybe as you say that it will set off the same alarms in your head that it does in mine. And if you set off the alarms sufficiently early, you may be able to avoid the fatal pinch.
To make this alarm explicit: if you were that investor, would you save the company? I wouldn't.
Despite this definitely being important for all businesses at some point, where does this come in when evaluating businesses like early Facebook and Google where prior to monetization, wouldn't they appear to be "default dead"?
It feels like this post is somewhat also addressing the very high burn rates which companies have, and that you should have control over your trajectory before you begin burning all your money.
This is effectively saying that these businesses don't have cashflow projections. Business 101 - should be taught by whoever the investors are that are "adding value".
It takes a lot of thought and planning to make sure you're 'default alive' in all circumstances. It slows you down, and it requires you to think through the implications of every decision, big and small.
This sounds like a fairly similar notion...