Per-Hour Learning Potential / Utility: ★★★★★★ (6/7)
Readability: ★★★★★★ (6/7)
Challenge Level: 1/5 (None) | ~230 pages ex-notes (320 official)
Summary: This book is a perfect introduction to topics like luck, sample size, social proof, path-dependency, and hindsight bias for readers who haven’t encountered the topics before. While it lacks the depth and nuance of books like How Not To Be Wrong, Superforecasting, or The Signal and the Noise, it makes up for that by being extremely compact and at least touching briefly on a number of concepts like arms races and local vs. global optimization that readers will encounter elsewhere.
Highlights: Mauboussin’s “two jars” metaphor, as well as the MusicLab experiment, are extremely memorable and important concepts. His three pages on MusicLab are some of the most valuable in any book anywhere.
While I do feel some topics should have been covered in more depth, Mauboussin generally does a very good job of making things understandable and accessible without “dumbing down” the content.
Lowlights: There’s not much to dislike about the book; parts of the end do get a bit repetitive, but that’s no big deal.
The only major detractor, in my view, is the nauseatingly frequent references to Nassim Taleb, a vastly overcited author who is A) often wrong, B) when right, not particularly insightful or interesting, and also C) kind of an arrogant jerk who just isn’t the kind of person we should all admire and respect. Mauboussin is a thoughtful intellectual with value to contribute; Taleb is not.
I go into it in more details in the notes. If you don’t get there, see Philip Tetlock’s Superforecasting (SF review + notes), specifically pages 237 – 245 thereof, for a more balanced but still critical discussion on Taleb’s nihilistic worldview that falls somewhere between my take (perhaps contrast-biased derision and contempt) and Mauboussin’s (inappropriate fawning).
Mental Model / ART Thinking Points: social proof, probabilistic thinking, base rate, inside / outside view, process vs. outcome, fundamental attribution error, hindsight bias, luck, overconfidence,storytelling, sample size, cognition vs. intuition, conditional probabilities, zero sum games, absolute vs. relative skill, disaggregation, feedback, local vs. global optimization, status quo bias, nonlinearity,social proof, path-dependency, utility, correlation vs. causation, structural problem solving, margin of safety, contrast bias, loss aversion, a/b testing,
You should buy a copy of The Success Equation if: you’re a novice reader looking for a compact introduction to a wide variety of important mental models, or you’re an advanced reader looking for a light, easy weekend read to get you thinking about some concepts you maybe need to dust off.
Reading Tips: none in particular; feel free to skim the last few chapters if they start to seem repetitive.
Pairs Well With:
“Superforecasting” by Philip Tetlock (SF review + notes). Superforecasting goes more in-depth on some of the concepts like cognitive biases and process covered by The Success Equation, while also touching on a number of other important topics.
“The Halo Effect” by Phil Rosenzweig (Halo review + notes). This book essentially takes Mauboussin’s brief reference to Jim Collins and turns it into circa-200 pages of detailed, thoughtful, engaging analysis about the risks of business case studies.
Reread Value: 2/5 (Low)
More Detailed Notes + Analysis (SPOILERS BELOW):
IMPORTANT: the below commentary DOES NOT SUBSTITUTE for READING THE BOOK. Full stop. This commentary is NOT a comprehensive summary of the lessons of the book, or intended to be comprehensive. It was primarily created for my own personal reference.
Much of the below will be utterly incomprehensible if you have not read the book, or if you do not have the book on hand to reference. Even if it was comprehensive, you would be depriving yourself of the vast majority of the learning opportunity by only reading the “Cliff Notes.” Do so at your own peril.
I provide these notes and analysis for five use cases. First, they may help you decide which books you should put on your shelf, based on a quick review of some of the ideas discussed.
Second, as I discuss in the memory mental model, time-delayed re-encoding strengthens memory, and notes can also serve as a “cue” to enhance recall. However, taking notes is a time consuming process that many busy students and professionals opt out of, so hopefully these notes can serve as a starting point to which you can append your own thoughts, marginalia, insights, etc.
Third, perhaps most importantly of all, I contextualize authors’ points with points from other books that either serve to strengthen, or weaken, the arguments made. I also point out how specific examples tie in to specific mental models, which you are encouraged to read, thereby enriching your understanding and accelerating your learning. Combining two and three, I recommend that you read these notes while the book’s still fresh in your mind – after a few days, perhaps.
Fourth, they will hopefully serve as a “discovery mechanism” for further related reading.
Fifth and finally, they will hopefully serve as an index for you to return to at a future point in time, to identify sections of the book worth rereading to help you better address current challenges and opportunities in your life – or to reinterpret and reimagine elements of the book in a light you didn’t see previously because you weren’t familiar with all the other models or books discussed in the third use case.
Page 2: “Mauboussin’s trash can” is totally a fun punchline, and also it sort of rhymes (if you pronounce his last name in a French way.)
Pages 6B – 7T: One of the core concepts in this book is the base rate, closely related to the inside /outside view. Mauboussin notes that the “ inside view” can predominate when luck plays a small or no role; when luck plays a larger role, the “ outside view” should predominate.
Page 10: Mauboussin notes that:Statisticians who are serious about their craft are acutely aware of the limitations of analysis. Knowing what you can know and knowing what you can’t know are both essential ingredients of deciding well. - Michael Mauboussin Click To Tweet
This worldview, on probabilistic thinking and overconfidence, is largely consistent between Mauboussin, Nate Silver, Jordan Ellenberg, Philip Tetlock, and pretty much anyone else who is numerically literate. As Silver puts it in “The Signal and the Noise” (SigN review + notes):
[…] It is when we deny our role in the process that the odds of failure rise. Before we demand more of our data, we need to demand more of ourselves.”
And, Mauboussin again:Not everything that matters can be measured, and not everything that can be measured matters. - Michael Mauboussin Click To Tweet
Page 12: Mauboussin sets this bit up so that you fall into the availability heuristic trap… Bill Gates, right? Nope. Also a good introduction to the idea of counterfactual thinking (and, of course, luck and path-dependency).
Page 13: Mauboussin, like Tetlock, has little use for pointless philosophical debates. Yay Mauboussin.
Pages 14 – 15: Mauboussin notes that small sample sizes increase the potential variability of outcomes and impact on randomness; I can’t remember if he cites the cancer-extremes-in-the-Dakotas data that Jordan Ellenberg does in How Not To Be Wrong, but it’s the same principle.
Page 16: One of the hardest lessons to learn regarding luck is that while it evens out over the grand scheme of things:
“the observation doesn’t hold for any individual, and the timing of luck can have a large cumulative effect.”
This shows up all over the place, mostly for the path-dependency reasons explored by Mauboussin later. Good examples can be found in many entrepreneur books: Brad Stone’s “The Upstarts” (TUS review + notes), for example, explores how many Uber competitors with reasonable products fell by the wayside for one reason or another.
He’s an anybody, an ordinary guy like you and me… but one who was almost a somebody. While Zuckerman’s book (naturally) focuses on energy magnates like Harold Hamm and Aubrey McClendon who rode the fracking revolution to riches, Zuckerman also does a good job of exploring the counterfactual: the people who had similar ideas – sometimes even better ones – but, for various reasons, didn’t reap the rewards.
Sanford Dvorin, for example, at one point before he was forced out of the business:
‘had leased five thousand acres in the Barnett at an average of $50/acre. Less than a decade later, the same acreage would sell for $22,000 an acre, or $110 million.”
Anyway, luck can, of course, have different impacts in different areas of life: for example, I feel like I’ve been extremely lucky professionally in many ways, and extremely unlucky personally.
Page 18: maybe not the best joke ever (you can read it here), but I’ll retell it sometime!
Page 19: An activity involves skill if you can lose on purpose.
I prefer the analogy of driving a car: cognitive is when you still have to actively think “check left, check right, check left.” Associative is when you have the hang of it, sort of, but only under well-controlled conditions. Autonomous is when you’re used to it.
Page 21: On process vs. outcome, Mauboussin cites Jeffrey Ma, involved with the famous MIT blackjack team. Ma notes that it takes a large sample size for the process to demonstrate itself, since there’s a lot of luck involved.
Investing is similar: luck plays a huge role over short time frames, and a lesser though still significant role over medium timeframes, and medium-term samples often don’t contain all the scenarios that could have been (or will be) faced by a particular strategy.
Page 23: Mauboussin notes that expertise doesn’t necessarily translate to models with superior predictivity. See also Tetlock’s Superforecasting (SF review + notes) for a really great discussion of the whole idea.
Page 25: oh hey look here’s the Dakota example. I think Ellenberg’s explanation is a bit better
Page 29: Mauboussin brings up the important necessity of differentiating between independent and path dependent type events. A player’s performance in any individual professional sporting game is independent (unless his performance last time was so bad that he gets benched, or you consider psychological impacts), whereas a phenomenon like fame is path dependent.
Page 34: this is truly a phenomenal page about storytelling that, for once, does something other than just tell the gosh-darn chicken story.
Page 35: again, here’s the idea of counterfactual thinking: just because things did play out a certain way doesn’t mean they had to. Which is why the aphorism “all’s well that ends well” is complete nonsense.
This can be a hard concept to wrap your head around if you’ve never heard of it before – what could be more obvious than results? Again, Mauboussin does a good job throughout the book of explaining this.
Page 36: hey look it’s the gosh-darn chicken story. (Okay so the thing is that it’s actually a great and memorable story, but if you read as much as I do, you will come across it so many times that you will end up wishing that, for once, someone would use a story that is NOT the chicken story.)
Page 37: Boo, Taleb, boo. We’ll get to this later when Mauboussin discusses MusicLab.
Page 38: What Mauboussin calls creeping determinism, I usually call hindsight bias: after the fact, it’s easy to rationalize that what happened was obvious. That’s why decision journaling is important: it’s often not whatsoever obvious ex ante, and it’s dangerous to think that it was after the fact.
A lot of this is driven by the way our memory and cognitive processes work. See books including Schacter’s The Seven Sins of Memory (7SOM review + notes) and Tavris/Aronson’s Mistakes were Made (but not by me) (MwM review + notes).
Richard Thaler also hits this in a business context in “Misbehaving” ( M review + notes) – one of my favorite parts of the book is pages 188 – 190, where Thaler explores how loss aversion and hindsight bias interact to create what are viewed as principal-agent problems ( local vs. global optimization problems) but may, in reality, be “dumb principal” problems – because executives view results as deterministic with the benefit of hindsight bias, thereby punishing people who made good decisions but were unlucky…
The point here is basically survivorship bias: case studies that focus on winners don’t see all the companies that followed the same strategy and ended up with nothing.
Ellenberg’s How Not To Be Wrong ( HNW review + notes) hits this one in some depth, returning to the “Baltimore Stockbroker Problem” throughout the book: a classic investment scam is to send out recommendations to buy or sell a stock to a large audience, then keep sending recommendations only to the ones you got right last time; by the end, you have a small group of people convinced you’re a magic stock whisperer.
Page 42: Mauboussin references the famous Ioannidis paper about “all research being wrong.” I spent a bit of time analyzing it in the notes to Ellenberg’s How Not To Be Wrong, specifically the notes to pages 147 – 155. Worth reading if you have a chance.
Mauboussin also brings up conditional probabilities – i.e. just because you’ve flipped four heads in a row doesn’t mean you’re “due” for a tails. The probability looking forward is 50-50, just as it always is.
Pages 52 – 53: Mauboussin’s idea of “drawing from two jars” is a really great visualization/metaphor for how skill and luck works. If your skill level is relatively consistent over time, and the average luck drawing is zero, your results should average to your skill over time, assuming independent events, but results in any given iteration can vary widely thanks to skill.
What Mauboussin doesn’t get into here (but does a bit later) is that this doesn’t hold for path-dependent events, wherein your result the first time influences your starting point the second time.
For example, a startup with the exact same idea, team, market, etc, could be sunk by an untimely recession (or simply pullback in availability of venture capital funding) – and never turn into anything.
Or, if you’re into tabletop games (which I am not, just to be clear!) you could call all this “roll for persuasion.”
Page 54T, 56 – 57: Mauboussin visualizes the pitcher-hitter interaction as an arms race, and notes the difference between absolute and relative skill. To outperform in zero-sum games, you can’t just be good; you have to be better than everyone else.
The obvious takeaway is to find non zero-sum games, or games where modest absolute skill translates to high relative skill. Cross-reference here Peter Thiel on competition in “ Zero to One” ( Z21 review + notes): Thiel notes that::
“competitiveness pushes people toward ruthlessness or death”
And also observes:
Winning is better than losing, but everybody loses when the war isn’t worth fighting.
Pages 61 – 63: The James-Stein estimator is an interesting concept… although like the Kelly Formula, I wonder if it can be wildly misapplied.
Pages 86 – 87: It’s important to differentiate between repetitive tasks and trial-and-error tasks; the former have less luck involved; the latter more. Mauboussin doesn’t mention The Checklist Manifesto (TCM review + notes) here, but it’s relevant.
Also, skill exists. Which is a good rebuttal to Taleb in various ways…
Page 93: I found this page fascinating mostly because of the science.
Page 95: and also this page. I seem to remember investing shows up on this sort of list at like 40 – 50, because crystallized intelligence is more important than fluid intelligence.
Pages 96 – 100: Mauboussin here goes deeper into intuition than Tetlock does in Superforecasting. Dr. Jerome Groopman’s How Doctors Think (HDT review + notes) is also really good here; one of Groopman’s premises is that:
“cogent medical judgments meld first impressions – gestalt – with deliberate analysis.”
Also, oh hey, here’s that table about making investing decisions.
Pages 101 – 102!, 103: here, Mauboussin notes that IQ and rationality aren’t always hand in hand.
Interestingly, the question he poses on page 102 always gives me more trouble than the question posed by Tetlock in Superforecasting ( SF review + notes) about a bat and a ball adding up to $1.10, or other examples about how fast X number of pumps will fill up a pool, which I always find trivially easy… perhaps because I’m very used to thinking algebraically, but not in formal logical / decision tree terms. ( Habit / conditioning.) This is also kind of an example of the MECEdisaggregation framework I mention in the Superforecasting notes.
I find the Zapper virus problem on page 103 equally trivial/easy because it’s math. Similar false-positive type problems are discussed by Ellenberg in How Not To Be Wrong and (I think, I can’t remember?) The Signal and the Noise by Nate Silver.
Page 106: At the bottom of this page, Mauboussin offers a really brief hypothesis for why superior performance declines over time: that they:
“fall prey to organizational rigidities […] exploiting known markets requires optimizing processes and executive effectively, and leads to reliable, near-term success.”
On the other hand, unknown markets are, well, unknown.
This is the famous Innovator’s Dilemma – a local vs. global optimization problem, as well as a trait adaptivity problem, since organizations select for locally- adaptive traits that create a reinforcing feedback loop.
even when life seems perfect, you have to take risks and jump to the next level, or you’ll start spiraling downhill into complacency without even realizing it.”
Many successful disruptive entrepreneurs are always pushing for change.
Sam Walton, for example, discusses in “ Made in America” ( WMT review + notes) how he “made it his personal mission” to ensure constant change was part of the Wal-Mart culture – sometimes, even just for change’s sake.
Pages 110 – 111: Philip Tetlock would be appalled at the lack of forecast accuracy testing here. Anyway, Mauboussin more formally discusses independent and dependent events here…
The point is that the two-jars analogy works for independent events, but many of the most important events are not independent, i.e. those that are path-dependent: one can see how, for example, the outcome of the American Revolution depended on many previous events.
Mauboussin brings up the idea of “preferential attachment,” a fancy name for what’s colloquially known as “the rich get richer.” He talks about “phase transitions” and “tipping points” but really I think the strongest analogy (the one I really love, and always remember) is the “riot threshold” – a small difference in initial circumstances can lead to wildly different outcomes. I would call this a subset of critical thresholds.
Also there’s that story from somewhere about the riots and the food carts. Gosh darn it where is that from?
Pages 123 – 125: Mauboussin extremely briefly mentions economies of scaleand network effects, though not in enough depth to be useful. He also posits the “winner-take-all” line of thinking that is brought up in Cal Newport’s Deep Work(DpWk eview + notes), which I thought was overstated.
Mauboussin takes it more in the direction of public-company CEOs; he notes that it’s difficult to separate skill from luck when it comes to corporate executives, and suggests that in many cases, they’re simply compensated for luck. (He also discusses the feedbackloop that drives compensation higher.)
Usually this compensation for luck is not blatant/explicit, but in some cases it is. The worst executive compensation package I’ve ever seen at a legitimate company, by a long shot, was a hedge-fund-like comp structure at a metals trading company wherein executives were paid an escalating amount based on pre-tax profit above a certain level.
Why is this so shocking? Well, their 10-K literally stated that the primary driver of their revenue, and thus profitability, was market volatility (which I think we can all agree is luck, not skill!) For fun, here is part of my writeup on that company from circa-2016 that covers the compensation package…
That was one egregiously bad. I’ve heard Biglari Holdings is also terrible, but I’ve never done the work there.
Pages 126 – 127: Touching on college rankings, Mauboussin notes that you get what you measure, and some of these measurements have a large degree of social proof. He also tangentially brings up the idea of utility – given that different people have different criteria.
Pages 128 – 131: Hands down, the discussion of MusicLab here is some of the most important and valuable in the book (and, perhaps, my bookshelf). Discussing the interaction of social proof with path-dependency, researchers ran a controlled experiment that determined not only that (unsurprisingly) a lot of songs are popular because they’re popular – cue Yogi Berra quips – but that also there’s a huge degree of luck; a song getting some traction early can lead to massively disproportionate outcomes over other songs that control groups ranked as just as good or even better.
This happens all the time in the real world, too: Snapchat is a dumb and completely undifferentiated app, and Evan Spiegel has never (to my knowledge) displayed any particular talent or insight. So why is Snapchat so popular? Because, unlike tons of other social-media/messaging apps that fizzled out, it was the lucky MusicLab winner.
Of course, thanks to fundamental attribution error, we are sometimes capable of recognizing this when it comes to other people, but not to ourselves: Mauboussin cites an Ellen Langer / Jane Roth paper, whose title, “Heads I Win, Tails It’s Chance,” just about sums it up.
The other real-world example of MusicLab is Nassim Taleb, who I’ve had a number of (unkind) nicknames for over the years. For those who aren’t familiar with the financial literature, Taleb is the go-to reference when you want to sound smart and contrarian because he says a bunch of stuff about how the world is unpredictable and we’re all running around trying to forecast the wrong things and so on. He’s the classic stereotype of the smart-sounding skeptic who has a dedicated following because when you cite him, you get to feel like part of the in-club that’s smarter and cooler than everyone else. (John Hussman probably also goes in this bucket.)
In my view, Taleb is nothing more than a function of the MusicLab experiment that Mauboussin discusses: Taleb happened to be in the right place at the right time to say some provocative, attention-grabbing things, and social proof did the rest. Are some of his lines of thought worthwhile? Sure. Are they so particularly unique or special that he should be put on a pedestal? Of course not.
Taleb’s fifteen minutes should’ve been over a decade ago and it’s not clear to me why anyone still pays any attention to him; he gets way more airtime than way more insightful and thoughtful authors like Phil Rosenzweig, who makes directionally similar points in The Halo Effect (Halo review + notes) with more nuance and more appreciation for real-world constraints.
Mauboussin certainly isn’t alone in his mindless and mind-numbing idol-worship of Taleb – it’s pretty common, and even extends to God (I mean, Howard Marks) – but The Success Equation would’ve been better if it had never referenced Taleb’s name in the first place.
Nihilists don’t deserve a platform.
He also briefly discusses correlation (which isn’t causation!). Ellenberg goes deeper here, and my favorite pithy quote on correlation vs. causation is from Nate Silver in the thoughtful “The Signal and The Noise” (SigN review + notes):Ice cream sales and forest fires are correlated because both occur more often in the summer heat. But there is no causation; you don’t light a patch of the Montana brush on fire when you buy a pint of Haagen-Dazs. - Nate Silver Click To Tweet
Pages 140B – 141T: I’ve never read Moneyball (I know, I know) but the core concept here is Marks-like second level thinking mixed with utility. The secret to the A’s was that they identified a usefulstatistic that was underpriced by the talent market.
Page 142: Nice Bill James quote here about storytelling.
Of course, the long-term is nothing but a series of short-terms, so at some point results have to matter. Tetlock provides some nice commentary here on page 87 of Superforecasting – life’s not just about a checklist of good-forecasting techniques that don’t ever make contact with reality.
Page 159: Mauboussin sums up deliberate practice, and touches on utility: he alleges that you can’t train System 1 for activities that aren’t stable or linear.
I disagree; I find the truth to be closer to Tetlock’s rebuttal to Kaheman’s quasi-nihilism on Page 236 of Superforecasting. Tetlock:
“My sense is that some superforecasters are so well practiced in System 2 corrections – such as stepping back to take the outside view – that these techniques have become habitual. In effect, they are now part of their System 1.”
Pages 161 – 162: Hey look, checklists (an example of structural problem solving). Also, Mauboussin acknowledges Ritchie’s point in intelligence. He notes that:
“the claim that talent plays no role in how well people do is not supported by the facts.”
This is sort of obvious to most normal people, but there are hardcore “nurture” types who don’t believe it. That’s a problem because of the trait adaptivity mental model: it makes no sense to intentionally compete in games at which you’re at a natural disadvantage.
I’m 5’8 (on my driver’s license) and so, clearly, I probably shouldn’t try to guard Lebron. Lebron, of course, probably shouldn’t try to take the opposite side on my portfolio…
Pages 169 – 170: Mauboussin here discusses, briefly, margin of safety, as well as leverage.
Pages 173 – 174: Mauboussin brings up the conventional/unconventional vs. right/wrong matrix. Marks goes into this in more depth in The Most Important Thing.
Pages 178 – 181: this is sort of a trait adaptivity story, in a limited sense: the idea is that if you have a big advantage, you keep things simple (to minimize the role of luck). If you’re at a disadvantage, you make things complicated (to maximize the role of luck, which is the only way you win).
Page 188: some mention of control groups here. See also Ellenberg.
Pages 197 – 200: Francis Galton sighting! Anyway, Mauboussin discusses reversion to the meanhere, not as comprehensively as Ellenberg.
Page 214: Mauboussin cites recency bias as well as our tendency to overinterpret from small sample sizes.
Page 221: For more on the null hypothesis, see Ellenberg.
Page 225!: Mauboussin briefly references Tetlock and the idea of “ counterfactual thinking,” circling back to MusicLab. One major theme of Tetlock’s Superforecasting is the virtues of being a proverbial many-handed economist.
First Read: 2016
Last Read: 2018
Number of Times Read: 2
Planning to Read Again?: no
Review Date: spring 2018
Notes Date: spring 2018