(This post was collaboratively written together with Duncan Sabien.)


Startup founders stereotypically experience some pretty serious mood swings.  One day, their product seems destined to be bigger than Google, and the next, it’s a mess of incoherent, unrealistic nonsense that no one in their right mind would ever pay a dime for.  Many of them spend half of their time full of drive and enthusiasm, and the other half crippled by self-doubt, despair, and guilt.  Often this rollercoaster ride goes on for years before the company either finds its feet or goes under.






Well, sure, you might say.  Running a startup is stressful.  Stress comes with mood swings.  


But that’s not really an explanation—it’s like saying stuff falls when you let it go.  There’s something about the “launching a startup” situation that induces these kinds of mood swings in many people, including plenty who would otherwise be entirely stable.


Moving parts


There are plenty of stressful situations which don’t cause this sort of high-magnitude cycling.  Waiting on blood test results from a cancer scare, or applying to prestigious colleges or companies, or working as a subordinate startup employee[1]—these are all demanding, emotionally draining circumstances, but they don’t induce the same sort of cycling.

Contrast those situations with these:

  • People in the early stages of their first serious romantic relationship (especially those who really really really want it to work, but lack a clear model of how)
  • People who are deeply serious about personally making a difference in global poverty, space exploration, existential risk, or other world-scale issues
  • People who are struggling to write their first novel, make their first movie, paint their first masterpiece, or win their first gold medal
  • ... all of which have been known to push people into the same kind of oscillation we see in startup founders.  As far as we can tell, there are two factors common to all of these cases: they’re situations in which people must work really, really hard to have any hope of success, and they’re also situations in which people aren’t at all sure what kinds of work will get them there.  It’s not a response to high pressure or uncertainty, it’s a response to high pressure and uncertainty, where uncertainty means not just being unsure about which path to take, but also about whether the paths (and the destination!) are even real.

    Subconscious intentionality


    It’s easy to point to the value in euphoria and optimism.  You get lots of code written, recruit lots of funding and talent, write a perfect draft—it’s the part of the cycle where you’re drawn to working seventy hour weeks, checking off each and every item from your to-do list.  But the “down” parts often feel like they’re pointless at best, and dangerous or counterproductive at worst.  Most of the online commentary on this phenomenon treats them as a negative; there’s lots of talk about how to break the pattern, or how to “deal with” the extremes.  In our own pasts, we found ourselves wondering why our brains couldn’t just hang on to the momentum—why they insisted on taking us through stupid detours of despair or shame before returning us back to apparent “forward motion”.


    Interesting things happened when we began treating that question as non-rhetorical.  What, we wondered, might those down cycles be aimed at?  If we were to assume that they were, on some level, useful/intentional/constructive, then what sorts of instrumental behavioral patterns were they nudging us toward?


    Imagine breaking the two parts of the cycle into two separate people, each of whom is advising you as you move a startup forward.  One of them is the voice of confidence—the classic "inside view"—with a clear sense of the possibilities and lots of enthusiasm for next actions.  The other is the voice of pessimism, with a range of doubts and concerns from the object level all the way up to the fundamental assumptions of your whole project.


    A naive manager might assume that one of these advisors is better than the other.  Certainly one of them is more “pleasant” to have around.  But it’s almost trivially obvious that you’ll build better models and more robust plans if you keep both of them involved.  It’s not just about balancing positive and negative outlooks, or constructive and destructive impulses (although that’s a part of it).  It’s about the process of building an accurate map in unexplored territory, which often requires a kind of leapfrogging of provisional model-building and critical model-tearing-down. You start with a bad but workable model, collide it with reality, and then refine the model in response before colliding it again.  Both halves of the process are critical—a good entrepreneur needs both the ability to follow a line of thinking to its conclusion and the ability to throw the line away and start from scratch, and would be equally crippled by a lack of either.



    Why despair?


    Startup founders, aspiring filmmakers, and would-be worldsavers “have to” be enthusiastic and positive.  They have to sell their models—to skeptical investors, to potential hires, and oddly often to themselves—and many people struggle to hold both determined confidence and fundamental uncertainty in their minds at the same time.  Instead, their brains use mood to encourage them to alternate.  The euphoric half of the cycle is where assumptions are taken to their conclusion—where you collide your models with reality by building code, attempting to enroll customers, and so on.  And the despair half is where those assumptions are challenged and questioned—where all of the fears and doubts and worrying bits of evidence that you’ve been holding off on looking at are allowed to come forward.


    Interestingly, the real power of the down cycle seems to be less that it allows thoughts like “maybe my startup will never work,” and more that, because it allows such thoughts, it also allows “maybe my understanding of the sales landscape is wrong,” or “maybe the product manager we’ve just spent three months getting up to speed is actually a terrible fit for the team.”  Despair can be a key that unlocks whole swaths of the territory.  When you’re “up,” your current strategy is often weirdly entangled with your overall sense of resolve and commitment—we sometimes have a hard time critically and objectively evaluating parts C, D, and J because flaws in C, D, and J would threaten the whole edifice.  But when you’re “down,” the obviousness of your impending doom means that you can look critically at your past assumptions without having to defend anything.[2]



    Moving forward (or backward, as the case may be)


    Our recommendation: the next time you’re working on something large, important, and very, very uncertain, don’t resist the occasional slide into despair.  In fact, don’t even think of it as a “slide”—in our experience, it stops feeling like a slide as soon as you stop resisting it.  Instead, recognize it for what it is—a sign that important evidence has been building up in your buffer, unacknowledged, and that it’s time now to integrate it into your plans.


    Understanding the subconscious intentionality behind this sort of mood swing doesn’t make it any less painful, but it can make it far more pleasant—easier to endure and easier to mine for value.  Think of it as an efficient and useful process for building accurate models under stressful uncertainty, and trust yourself to hold together in both states, resisting neither.





    There’s a version of you that is good at moving forward—that has the energy to pour seventy hours a week into your current best guesses, stretching the parts of your model that are correct as far as they can go.  And there’s a version of you that is good at dealing with the darker side of reality—that is actually willing to consider the possibility that it’s all garbage, instead of just paying lip service to the idea.  Embrace each in turn, and the result can be periods of high productivity followed by periods of extreme flexibility, a combined strategy whose average utility often seems to outweigh the utility of an average strategy.



    [1] The example of subordinate startup employees seems perhaps particularly revealing, since they share the same uncertainty and long hours as their bosses, but are not expected to come up with the same kinds of strategic solutions.


    [2] Alternating between despair and forward motion may also be a good strategy given human working memory limitations. :) 




    New Comment
    20 comments, sorted by Click to highlight new comments since: Today at 2:13 PM

    [Coauthor's note] If anyone is willing to offer me karma by upvoting this comment, such that I can post my own stuff someday instead of piggybacking on Anna, I'd appreciate it.

    Instead, recognize it for what it is—a sign that important evidence has been building up in your buffer, unacknowledged, and that it’s time now to integrate it into your plans.

    Maybe. On the other hand, I also recently read another article on this, which attributed things more to emotional oversensitivity to a lack of progress, which might also be a reasonable explanation:

    A while back I wrote about how the nonlinear life and our linear emotions aren’t exactly optimally suited to each other. Your brain craves signs of progress, so it could reward you with a burst of feel-good chemicals. Unfortunately, the nonlinear life doesn’t work like that. Often, you can spend days or weeks slaving away at the office/studio/whatever, not really moving forward – or even taking two steps back for each move forward. Despite the hours that you put in, the article/thesis/design never seems to be finished, making you question whether you’re really cut out for this kind of job. Perhaps you’d do the world a favor by setting your sights lower and working as a sales clerk instead. [...]

    ...let’s assume that we have a project that has a goal we’re trying to reach. Arbitrarily, let’s say that the completion means we reach a threshold of 100 points. Of course, these numbers are completely make-believe and I pulled them from my magical hat. Now, further, let’s assume that each unit of time – say 1 unit equals 1 day – means we have three possibilities: make progress, stay where we are, or take steps backward. In my personal experience, this is an ok model for work: sometimes you’re actually making progress, and things move smoothly. Sometimes, though, you’re actually hurting your project, for example by programming bugs into the software, which need to be fixed later on (just happened to me two weeks ago). Most often, though, you’re trying your best, but nothing seems to work. Maybe you’re stuck in a dead end with your idea, and need to change tack. Maybe you’re burdened with silly tasks that have nothing to do with the project. Well, I’m sure we all have these kinds of days. So let’s again use my magical hat and pull out some probabilities for these options. Let’s say you have a 5% chance of making a great jump forwards (10 points), 25% chance of making 3 points of progress, 55% chance of getting stuck (0 points), 10% chance of making a mistake (-2 points), and a 5% chance of doing serious damage (-6 points). Now we just simulate these across and get a graph that shows your cumulative progress towards the goal (yes I'm doing this in Excel):

    Even though the numbers are really made up, I feel the above graph is actually a pretty decent example of how the nonlinear work often feels. However, there’s still the additional complication: the emotions.

    Suppose that our emotions work as follows. If you’re making progress, you feel good. And this is mostly irrespective of how much progress you’re making. Suppose the same holds for drawbacks – it hurts, but it hurts almost as much to look for a bug for two hours or the full day. Finally, I’ll assume that if you’re not moving anywhere, you inherit the feeling from the day before. [...] So with these assumptions, we get the following graph portraying emotions:

    [...] I think the above graph is quite a good summary of how the nonlinear life often feels: you love you’re job, but you’re not above hating it when things are not going well.

    Which isn't to say that "don't resist the slide down to despair, instead take it as an opportunity to evaluate the information" wouldn't be a good idea (indeed, that's a reaction that I'd recommend each time one feels a negative emotion), but on the other hand, "are my emotions just ill-calibrated for nonlinear progress" might be a good question to start the evaluation process with.

    If lack of progress is what causes negative emotions, then it seems to me that another possible reason why startup founders might have mood swings is that they usually build one startup at a time. Therefore, if you are not making any progress, you are not making any progress at all. John Conway advises that mathematicians should work at several problems at once in order to avoid such mood swings:

    Work at several problems at a time. If you only work on one problem and get stuck, you might get depressed. It is nice to have an easier back-up problem. The back-up problem will work as an anti-depressant and will allow you to go back to your difficult problem in a better mood. John told me that for him the best approach is to juggle six problems at a time.

    Startup founders may not have such luxury, but perhaps at least in some cases it is possible to structure things in a similarly disjunctive way, where you can note the problems in a part A, but avoid despair by noticing that at least you are making progress in part B (of course, if different people are responsible for parts A and B, the problem might remain for each one of them). Averaging might make your progress look closer to linear in a certain sense.

    John Conway advises that mathematicians should work at several problems at once in order to avoid such mood swings:

    See also: structured procrastination.

    This doesn't really ring true to me (as a model of my personal subjective experience).

    The model in this post says despair is "a sign that important evidence has been building up in your buffer, unacknowledged, and that it’s time now to integrate it into your plans."

    But most of the times that I've cycled intermittently into despair over some project (or relationship), it's been because of facts I already knew, consciously, about the project. I'm just becoming re-focused on them. And I wouldn't be surprised if things like low blood sugar or anxiety spilling over from other areas of my life are major causes of some Fact X seeming far more gloomy on one particular day than it did just the day before.

    And similarly, most of the times I cycle back out of despair, it's not because of some new information I learned or an update I made to my plans. It's because, e.g., I went to sleep and woke up the next morning and things seemed okay again. Or because my best friend reminded me of optimistic Facts Y and Z which I already knew about, but hadn't been thinking about.

    I think this is a good point. Despair, which I see as perceived hopelessness, originates in an individual and so it depends on how that individual perceives the situation. Perception is not like receiving a reflection of the world in the mind. It is like meshing together the neural activity from percepts with the existing neural activity ongoing in the brain. The result is that it is context dependent. It is affected by priming and emotions, for example.

    I think the advice in this post, essentially embrace despair, isn't probably that helpful. What do you think about this advice: "Notice despair for it is a signal of hopelessness. It indicates that you may be stuck in a mental rut or that the way that you are viewing a situation may be inducing unnecessary anxiety. In summary, it tells you to rethink how you are trying to solve the problem that you are facing. The first thing you should do is check that it is real. Get advice and talk to others about it. Try to get out of your head. Also, try and find out if it is misattributed. It may be due to low blood sugar or anxiety spilling over from other parts of your life, for example. If you have done this and now know that the despair is real, i.e. resulting from a complex problem that matters to you and that you can't solve, then try to understand the problem you are facing and your plan to solve it. Once you are happy with the plan then you can embrace the incoming depression. Do not view it as anathema, but instead as your body's mechanism to move you into the necessary focused and analytical state that you need to be in to be able to solve the complex problem that you are facing".

    As someone who regularly 'embraces his despair', I've noticed that it's one thing to visibly despair as a startup founder with 18 months of runway, and another thing to visibly despair as a freeloader staring at the possibility of homelessness.

    Despair has social signalling consequences, and whether those signals help or hinder your ability to actually Get Shit Done is highly context-dependent.

    And yes, it has these signalling consequences whether or not you choose to actively talk about your despair - it affects everything you do. For example, one person might be the head of a non-profit FAI research organization, and his tiredness and grimness are seen as evidence that he's obviously super-dedicated and working super-hard - and the fact that he's not talking about it and making jokes instead just show how stoic and resolute he is / the fact that he's talking about it and worrying about what to do just show how sensitive and in-touch with himself he is. Someone else might be a socially awkward quiet girl who doesn't know where she's going to sleep next, and her tiredness and grimness are seen as evidence that she's obviously bad news and going to be a drain on other people / the fact that she's talking about it and worrying about what to do just show how desperate and needy and self-absorbed she is.

    So there's definitely structural social incentives for people who are already on a certain kind of success trajectory to embrace their despair, but the principle of 'equal and opposite advice' strongly holds, and I'd probably advise people on the other side of the impact bell curve to avoid anything that might break the fragile bubble of positivity that's likely shielding them from the howling vacuum of the Horns Effect.

    Yes! This appears true. Not all despair fits into this model (some people fail to get the upswings), but I think a large fraction does. There's one important, non-obvious corollary: not only should you not resist the slide into despair or try to never have it, but you should optimize what you do in that time! If, for example, you respond to the feeling of despair by drinking alcohol, you'll probably never get to have the thoughts that mood had to offer.

    There’s a version of you that is good at moving forward—that has the energy to pour seventy hours a week into your current best guesses, stretching the parts of your model that are correct as far as they can go. And there’s a version of you that is good at dealing with the darker side of reality—that is actually willing to consider the possibility that it’s all garbage, instead of just paying lip service to the idea.

    Interesting idea, but it looks like you are talking solely about the rising and falling sections here. I personally think that the most pertinent parts are instead the peaks and the troughs. What happens in those particular moments and what types of subsequent falling and rising sections do they require or are the best?

    Perhaps, the trough must be preceded by the falling section, not only because old models are discarded there, but also because it is the best place for truly innovative discoveries to occur. It is where arousal is minimal and introspection/inhibition/rumination is maximised. The trough and subsequent rising section, then, is where you make use of the new ideas or paradigms that you have discovered.

    The sense that I've gotten (from my own experience, from anecdotes about startup founders, and from some of Anna's examples) is that there's not much distinction between "falling" and "trough," or between "rising" and "peak." We're looking at a phenomenon of RAPID, LARGE swings, such that you're basically at one place one day, and in a completely different one the next (not that the shifts happen daily—just that they're abrupt).

    In other words, the way we chose to describe it might make it seem like there's a larger distinction, there, but I think what you're labeling "peaks" and "troughs" are actually the same places we were trying to talk about.

    Melancholia, mania and severe depression might be a bit different, but with normal episodes I think the upswing is when you are solving problems without issues. The turning point at the top is when you come across a complex problem which matters to you and that you are unable to solve. The downswing and slide into depression is your body’s way of moving into a kind of analytical mode. Like fever is induced to fight infection, depression is induced to fight despair. I think that despair is separate to the depression. It is the perceived lack of control over negative events, we often call it hopelessness. Depression, on the other hand, is the induced state of mind.

    The reason why such mood swings as you describe might happen with startup founders, for example, is that they face complex problems that they care deeply about and that have high social costs if they fail at them. The idea I am presenting here gets to be quite complicated in practice because the cause of the depression is the perceived complexity of the problem, perceived ability to solve it and perceived social costs with failing. These all depend on the perceiver and we don't see the world as it is, we see they world as we are. It is possible that someone might slide into depression because of misttributed emotions, wrong beliefs etc.

    Depression might also be too strong a word. Perhaps, 'sad' is better. I don't know. I am referring to 'depression' as the down ward sections in your mood swings figure.

    I am basing what I wrote above on my understanding of this paper which posits that depression is an evolved stress response mechanism. This claim makes sense if you believe the broaden and build theory. The paper makes the following claims

    1. Complex Problems Trigger Depressed Affect. The analytical rumination (AR) hypothesis proposes that depressed affect is triggered by problems: (1) that are complex (analytically difficult); and (2) that affected fitness in evolutionary environments. [...] One effect of sad or depressed mood is to promote an analytical reasoning style in which greater attention is paid to detail and information is processed more slowly, methodically, thoroughly, and in smaller chunks [...] Complex social problems may be the primary evolutionarily relevant trigger of depression in human beings
    2. Depression coordinates a suite of changes in body systems that promote rumination, the evolved function of which is to analyze the triggering problem. [...]Analysis is time consuming and requires sustained processing, so it is susceptible to disruption, which interferes with problem-solving. Depression induces changes in body systems, producing effects that facilitate analytical rumination by reducing disruption [...]Specifically, depressed affect: (1) activates neurological mechanisms that promote attentional control, which gives problem-related information prioritized access to limited processing resources and makes depressive rumination intrusive, persistent, resistant to distraction, and difficult to suppress; (2) induces anhedonia, which reduces the desire to think about and engage in hedonic activities that could disrupt problem-related processing; and (3) promotes psychomotor changes that reduce exposure to stimuli that could disrupt processing (e.g., desire for social isolation, loss of appetite).
    3. Over evolutionary time, depressive rumination often helped people solve the problems that triggered their episodes [...] Like fever, then, the impairments associated with depression are usually the outcome of adaptive tradeoffs rather than disorder. For instance, because processing resources are limited, a decreased ability to concentrate on other things is a necessary tradeoff that has to be made in order to sustain analysis of a complex, depressogenic problem The fourth claim is that depression reduces accuracy on laboratory tasks because depressive rumination takes up limited processing resources [...]In summary, studies of clinical, subclinical, and experimentally induced depression all show that when given a laboratory task, depressed people ruminate about other things, which takes up limited cognitive resources and interferes with their ability to perform well on the task. It is therefore illegitimate to conclude that depression generally impairs problem-solving from studies showing reduced performance on laboratory tasks. They have nothing to say about how successful depressed people are in solving the problems that they are ruminating about.

      In summary, we hypothesize that depression is a stress response mechanism: (1) that is triggered by analytically difficult problems that influence important fitness-related goals; (2) that coordinates changes in body systems to promote sustained analysis of the triggering problem, otherwise known as depressive rumination; (3) that helps people generate and evaluate potential solutions to the triggering problem; and (4) that makes tradeoffs with other goals in order to promote analysis of the triggering problem, including reduced accuracy on laboratory tasks. Collectively, we refer to this suite of claims as the analytical rumination (AR) hypothesis.

    Certainly one of them is more “pleasant” to have around. ... Both halves of the process are critical

    future post request - strategies for:

    • making the other more pleasant
    • dealing better with the other being unpleasant
    • doing things with the unpleasantness that make them more productive (or making good use of the sway)

    I've found that simply asking the question asked near the beginning, "what is this for?" to be fairly helpful with negative emotions in general.

    Yeah, that's a new heuristic I've adopted recently, that I think other more effective humans around me were already running.

    Compose while manic, edit while depressed. This strategy has a long and successful history.

    In general I'd think bipolar people would know the best coping strategies. If your mood-regulator is permanently broken you probably have worse moods than someone who can return to equilibrium just by letting go of the problem.

    If these extreme emotions are indeed useful, then maybe we should look to explicitly seek them out (depending on whether we think we're below the optimal level of the emotion in question).

    • To generate the ups, maybe it'd be a good idea to hang out with friends, get drunk, and dream about ambitious ideas. Only focusing on the happy paths. Well, people already do this but I guess my point is that maybe it should be done more often and more explicitly.
    • I'm not sure how the downs would be generated.

    Thoughts on the usefulness of these emotions:

    • I agree that the ups are useful (motivation, communicate your excitement...). There are downsides like overconfidence, but I think these downsides are usually outweighed by the benefits.
    • As for the downs, my impression is that a lot of the time it just makes people lethargic and doesn't really motivate change. But then there are times when they do motivate change. It seems that people experience more than the optimal level of "downs". You could make the point that this is because people don't handle the downs well, but even if they were better handled I'm still bearish on the value of downs.
    • (I spent over a year working on a startup that failed and went through these emotions.)

    To anyone who hasn't seen it, Inside Out is a movie that is relevant to this talk about the usefulness of "ups" and "downs".

    When you’re “up,” your current strategy is often weirdly entangled with your overall sense of resolve and commitment—we sometimes have a hard time critically and objectively evaluating parts C, D, and J because flaws in C, D, and J would threaten the whole edifice.

    Aside 1: I run into many developers who aren't able to separate their idea from their identity. It tends to make them worse at customer and product oriented thinking. In a high bandwidth collaborative environment, it leads to an assortment of problems. They might not suggest an idea, because they think the group will shoot it down and they will be perceived as a generator of poor ideas. Or they might not relinquish an idea that the group wants to modify, or work on an alternative to, because they feel that, too, is failure. Or they might not critically evaluate their own idea to the standard they would evaluate any other idea that didn't come from their mind. Over time it can lead to selective sidelining of that person in a way that needs a deliberate effort to undo.

    The most effective collaborators are able to generate many ideas with varying degrees of initial quality and then work with the group to refine those ideas or reject the ones that are problematic. They are able to do this without taking collateral damage to their egos. These collaborators see the ideas they generate as products separate from themselves, products meant to be improved by iteration by the group.

    I've seen many cases where this entanglement of ego with idea generation gets fixed (through involvement of someone who identifies the problem and works with that person) and some cases where it doesn't get fixed (after several attempts, with bad outcomes).

    I know this isn't directly related to the post, but it occurred to me when I read the quoted part above.

    Aside 2: I have similar mood swings when I think about the rationalist community. "Less Wrong seems dead, there is no one to talk to." then "Oh look, Anna has a new post, the world is great for rationalists." I think it's different from the work related swings, but also brought to mind by the post.

    I realized this about my own mood swings.

    I tend to do a lot of "writing therapy" in workflowy. Thus, if I'm ever at a high, or a low, I record a lot of my thoughts into workflowy so that I can refactor/combine them later.

    I like model this as "emotional retrograde amnesia". I like the movie "Memento" by Nolan because the protagonist writes lots of notes to himself because he has actual retrograde amnesia; similarly, I have emotional retrograde amnesia, so I write lots of notes to myself, so that I can piece together the "puzzle" later.

    stereotypically ... have been known ...

    So just you are just musing on stereotypes?