Briefly, some updates:
*I've got some bonuses and downloads for Gateless, some really good ones coming. The book is getting great reviews -- if you miss the blog here and haven't gotten your copy yet, go get it.
*My next book, aiming for Q2 or Q3 2015, has "crossed the threshold" where it's definitely going to happen. I'm going to engage in a little less conversation and a little more action as a general rule, but my writing practice goes well and the quality goes up. I'll be more publicly involved sooner or later; for now, I'm enjoying my intense private work cycles and quality improvements.
*Thanks for all the well-wishes, emails, and the positive reviews/feedback of Gateless. Love you guys.
Ok, without further ado --
The High Performer's Performance Distribution Problem
Here's a graph from Wikipedia. We're going to start by looking at what's called a "normal distribution." A lot of the world falls into normal distributions.
A heck of a lot of the world falls into normal distributions.
Without getting into the statistics behind it, just trust me when I say this -- if your performance follows a normal distribution, a little bit more than 68% of your performance comes in within one standard deviation. We could call that "a normal day." Then, a little more than 27% is within two standard deviations, meaning that 13.5% of the time you'd be having a "pretty darn good day" or "really kind of a crummy day." Almost all of the remaining 5% still fits within three standard deviations, meaning 2.5% of the time you'd be having "an absolutely incredible mindblowing day" or, on the other side of the equation, "a day from hell."
Anecdotally, these numbers seem correct-ish for a lot of people. There's 30 days in a month. That means you'd have 20 or 21 normal-ish days, 4 really great days, 4 really bad days, and 1 or 2 "mindblowing" or "utterly awful" days each month.
Of course, any given month can be great or not so great. Having a disastrous week makes it more likely next week is a bad week. Having an amazing week -- if you don't have destructive celebratory habits -- makes next week more likely to be a good week.
Now, let's look at the Wikipedia image above. You'll note that the guy who is green is doing much worse than the guy who is blue. The green guy's best days are only about as good as the blue guy's average days.
Note that down; we'll come back to it.
Next compare the red and blue guys. The blue guy is much more consistent -- even on his bad days, he turns in a respectable performance. He doesn't scrape the bottom of the barrel too often.
Red, meanwhile, is much more varied. To me, this is the difference between the consistent steadfast performer running proven processes (blue) but not really trying to innovate or grab huge gains, as compared to the the person who does more random exploration and, when having a great idea, goes on a caffeine and sugar all-night binge trying to invent new models of the world. There's more days that make breakthroughs and more days that fall absolutely flat.
Red still is respectably consistent though. Yellow? You got no idea what you're getting out of yellow. Yellow has manically amazing days reasonably often, and "depths of hell" days reasonably often.
Performance Relative to Oneself
Here's the problem. Here's why I'm writing this.
No matter how much you level up, you'll still have some distributions of performance, and if underperforming bothers you, it'll always bother you.
It's straight-up better to be Yellow than to be Green... Yellow is like being a hyper-driven, erratically consistent nonconformist inventor in South Korea. Life will feel like hell much of the time, being misunderstood and not having one's science or inventing understood very well.
Meanwhile, green is like being a minor party official in North Korea. Your life sucks almost all the time. The best day you have any given month would be utterly forgettable by English, American, or South Korean standards.
But human emotions are a funny thing. The NK minor party official, ironically, probably goes through less emotional swings and "feels bad" less often; his life is reasonably consistent. Yeah, the electricity isn't working, the water is out, and his cousin that he liked disappeared for watching a South Korean soap opera last year and probably died at a prison camp, but everyone learned the lesson and now no-one watches soap operas... but it is what it is, and it's regular.
A pretty bad day of an eccentric South Korean inventor would be better than a pretty good day for a North Korean petty bureaucrat -- but it won't feel that way.
Oh, but it gets worse
Well, we could figure out some takeaways from these charts if we really wanted to. Whenever you can shift your entire curve to the right -- for instance, by planning each day the night before, and starting every day with a morning routine and brief fitness -- that would be good. It would make sense to do.
You can also reason out what mix of consistency and inconsistency is valuable for you at different times in your life. Okay, that's fine.
Here's the thing though -- that's not the only kind of distribution. Let's go bimodal, this image from this smart article --
That author makes the very smart point that all the standard probablity graphs are wrong. He writes --
Most distributions are not unimodal.
Most distributions are not symmetric.
Most distributions do not have mean = median = mode.
Most distributions are not Gaussian, Poisson, binomial, or anything famous at all.
This, of course, hinges on what "most distributions" are. I won't go as far as the author; simplified math can be useful if not misapplied.
Let's turn our attention back to this graph above. In this case, the person having those days would be having "below average days" almost all of the time -- with a few breakthrough days.
As you get towards high-performance, this becomes the case in many fields. Grind, grind, grind, grind, grind... get miserable... go crazy... breakthrough!
Too bad our minds aren't set up to accommodate that.
In the yellow line from the first graph, our eccentric South Korean inventor had "I'm in the depths of hell" type days precisely equally mixed in with those days where the plea that "O Muses, O high genius, aid me now" is answered.
In our second graph, the Muses don't answer so often. Most days are below average. If you've got standard untrained human emotions, you're going to feel bad most of the time.
And you know what? I know a lot of high performers intimately, and I'll say that -- with notable exceptions -- the majority do feel bad and like they're underperforming far more often than not.
Higher Performance Doesn't Solve This Problem
It just moves the lines around. Almost by definition, if you're a high-achiever, as you improve you start to look for higher achievements. This introduces new variance and randomness. Perfect consistency in output implies no experimentation, risk-taking, or variance.
If you're bothered by your underperforming days, you're always going to be bothered. It doesn't go away. You can't outrace it.
Cordon Sanitaire de son Esprit; Peut-être, L'équanimité
The problem is this:
Many high-performing people have a wider range of outcomes in performance.
The North Korean petty bureaucrat has only bad days and slightly less bad days, but the human mind is remarkably good at adapting to consistency. It's quite possible that he experiences less existential distress, neurosis, and anxiety than the erratic sometimes-brilliant South Korean inventor.
I hope that this post has shown you that you can't outrace this phenomenon just by performing better. All it does is shift the curve. Take the blue curve in the first post and move it to the right -- perform better every single day.
That's well and good, but the blue-line performer will still be upset on his worst days, just like the South Korean inventor is upset on his worst days.
That's the main thing I wanted to show you -- you can't outrun the bad feelings that come from the days you underperform, because as you get better, you establish a new baseline of performance.
You'll of course remember that my ethical system is all about doing things that are worth doing; I'm not advocating chilling out on the beach as the answer. (I've actually tested it out -- it doesn't work. If you're driven to achieve stuff, you'll go crazy if you sit on the beach for more than a few days.)
No, I think the answer is a sort of mental training to reach equanimity -- where you care about the outcomes, but you cordon off your emotional well-being from external events.
This doesn't result in lower performance -- it gives you the resolve to do long-term things that result in days that feel awful, knowing breakthroughs are on the horizon.
How to get there? Well, that's another subject for another time. But I'll say that I've found remarkable similarities in materials that aim to train this attitude up. Marcus Aurelius's Meditations is of course the most currently in vogue, but I've seen lots of similarities across traditions. Boethius's Consolation of Philosophy, Laozi's Tao Te Ching, Kempis's De Imitatione Christi, and really all of the Stoics, most of the Scholastics, and most of the Taoists and Zen Buddhists I've come across follow similar patterns, mental models, actions, and training regimes.
Going more in-depth on the training is another question for another time. The big takeaway I'd like you to take from this -- you can't outrun bad feelings that come when you underperform your potential, because your potential increases every time you improve. Know that and start working from there.
Have you heard about Dan Sullivan at StrategicCoach.com? He evangelizes a concept called Unique Ability for entrepreneurs to better manage their productivity and the impact of their organizations.
I think the point of Unique Ability lights up a point that´s pretty important to consider when you evaluate performance. THE POINT: YOU ARE NOT ALONE AND YOUR PERFORMANCE IS ALWAYS DEPENDANT OF THE PERFORMANCE OF OTHER PEOPLE AROUND YOU.
In a nutshell, the concept of Unique Ability explains that everyone has Unique set of talents that are different from anyone else in the world. And the use of those talents is and will ever be the best use of your time.
Unique Ability = Unique Set Of Talets + Passionate about spending your time on that zone
He stands up for a "BLUE" distribution to be the one that will eventually will have the greatest impact in the world, if you plan, design or get into an organization where your impact is multiplied by other BLUE-UNIQUE-ABILITY prople.
You should check him out. He seems really smart, and have being surrounded by big successful entrepreneurs and leaders for years. He also have some big and really smart clients like Peter Diamandis from XPrize Foundation.
Sorry for my english, I´m from Venezuela. Hope it helps.
Thanks for writing and sharing your work. I´ve read both of your books. Both great. Gateless: Awesome.
I am glad to see you post again. Julien said this is a timely post, and it is for me as well, as this subject has been on my mind recently -- you have a knack for that.
It as if with every three triumphant steps forward, there are two discouraging steps back. Every thing I build is imperfect, marred by significant flaws, and unsatisfactory when I look back in hindsight.
However, as I understand it, life is about the journey, not the destination. And so I have concluded that it is better to focus my energies on the process of building/growth, and perfecting that process, rather than concerning myself about the end product. I am not sure if one can cordon off their emotional well-being from external events, at the same time as caring about the outcome, since caring about the outcome implies emotional investment. However, if it works for you, I cannot dispute Experience.
In my experience, "not caring about outcomes" doesn't to pacification or apathy -- however whether or not it does depends on the individual -- for I am always actively building and I agree with you, that "chilling at the beach" is unpleasant to an active and ambitious mind. It is only that I focus on perfecting the process of building, rather than concerning myself with the end product.
Making mistakes are part of the process of growth, and so when making a mistake, I acknowledge and accept it, whilst recognizing that making them doesn't mean I innately suck and am a hopeless case at ever becoming excellent. Instead, I know that I am excellent because of what I do with the error - I try to remedy it.
"Not that I am already obtained or am already perfected, but I pursue."
Thanks for your thoughtful and thought-provoking writings, Sebastian. It's always a delight.
This is a timely post, as I was just reflecting on the quality of my recent work, and I was finding myself bothered by the fact that I could have done better. Thanks for this excellent post!
This is a great thing to be writing about, Sebastian.
In my experience, distribution-based inference tends to break down when people forget to check the ergodicity assumption—that is, whether or not the time-average is equivalent to the ensemble average. This notion makes irrelevant a single person's general performance distribution because almost all performance measures happen through time and it's seldom that people are not improving over time in those skills that are measured. This means in most cases the performance distribution of an individual is almost never ergodic for relatively large times or whenever the sample rate isn't much higher than mean's rate of movement. Distribution-based inference does make sense from the perspective of managing an organization, on the other hand, because in that you can have truly time-invariant ensemble statistics.
Statistics aside, my guess would be that managing the emotional effects of performance, as you say, is most important for durably improving it. What kind of management is best I think largely depends on the endeavor.
How would you model non-ergodic processes? Ergodic processes are easy to do math on, see, EX, Markov Chains for a closed system.
It's like you'd have to have a mashup of experience curve ( http://en.wikipedia.org/wiki/Experience_curve_effects#The_experience_curve ), individual output, skill accumulation, and then account for the lag time in that higher skill showing up in output, no? Is there any clean/good/useful/simple-enough models that do something like this?
The way to model non-ergodic processes just depends on the specific process. One large class of common problems are time-series, but they are too-varied a class of problem, in my opinion, to make general recommendations about them. The mathematical tools coming out of financial analysis and signal processing would apply in most cases, but they are just the scalpel to perform the autopsy, not a model of the disease.
It might be too general to be really exciting, but I think the most accurate model to durably improve performance—loosely defined—is to, as you often remind me, remain focussed on the process rather than the outcome. In a way, measuring performance is outcome driven. Of course, you need to know your output, this is not the process. The process is learning in the abstract. Learning is the system of cycling through Thesis -> Antithesis -> Synthesis -> Antithesis ...ad infinitum. I like tracking adherence to this method better than only tracking specific performance measures because the learning model supports fundamental changes in the nature of the activities that belong to the desired performance maximization.
For example, I may wish to improve the performance of my writing. This discussion offers two ways to support this, including this last one I just added:
I find 2 better than 1 because 2 doesn't care if you realize somewhere down the road that a more important metric has been discovered. Maybe subject diversity is a better metric than word count? Maybe the homogenization of my audience is better than total reach? In other words, 2 allows for the performance optimization protocol to be included as a hypothesis as well. I see you do this implicitly with your constant note taking of peoples' ideas and their correspondence or contrast to your own models.
I've spent some time thinking through a system to help track adherence to 2. One problem I encountered with my MVP is actually that the overhead in tracking 2 is significantly higher than in 1. Although this is to be expected because it requires constant, thoughtful evaluation of your process, it does make it harder to support in the long term. There is some simple software that would really help if one were so driven to produce it.
I also like 2 better than 1 except for exactly one problem, that's so big as to be almost overwhelming -- it's subjective.
Subjectivity means all data and evaluations more suspect; "general life pessimism" (for instance, during burnout or illness) is more likely to infect subjective assessments/measures... and that's exactly when you want to keep your metrics clean and use them to stay sane and persist.
> One problem I encountered with my MVP is actually that the overhead in tracking 2 is significantly higher than in 1. Although this is to be expected because it requires constant, thoughtful evaluation of your process, it does make it harder to support in the long term.
Yeah, I like the idea of iterations through a learning loop in theory; in practice, my objective trackings have almost always outperformed my subjective trackings. I'm open to finding a way to do it though.
It's great to hear from you, I miss your writing! I've been reading Gateless again.
Have you heard of the Red Queen effect? It's a mental model very similar to this one.
It is ubiquitous: from evolution to antibiotics to market competition to happiness (see hedonic treadmill).
Very interesting stuff.
Thanks Paulo, both on the missing writing and about Gateless. If you enjoyed it, could you leave a review on Amazon by the way? Those help a lot.
Re: Red Queen and hedonic treadmill, yup, I'm familiar with both. Other similar effects including arms races (I'd argue the main cause of WWI is Germany kicking off a naval arms race with England) --
And then, Marx got a heck of a lot very wrong, but the tendency of the marginal rate of profit to fall is both correct and really important --
INTERNAL SCORECARD #8
This is the eighth internal scorecard I've published. By writing these up, you get to see what productivity and production looks like on the inside, and I get to have an extra external accountability mechanism.
This one covers 7 July to 13 July.
RUNNING IN TAR WITH A FEW HUGE SPARKS
That's the best way I can describe this week. It had a very sleepwalking-like quality to it, with not much good work happening and poor compliance with the new habits I set forth the week before.
I wasn't sure if he was talking about me or not. If anything, I figured that he was being sarcastic because I play somewhat deceptively.
"You know what I'm talking about, right?" He asked me.
"No, not really."