If this is your first time reading, please check out the overview for Poor Ash’s Almanack, a free, vertically-integrated resource including a latticework of mental models, reviews/notes/analysis on books, guided learning journeys, and more.
N-Order Impacts / Unintended Consequences Mental Model: Executive Summary
N-order impacts in one sentence: when we evaluate the potential impact of our actions on a given system, we have a tendency to ignore the impacts of the system’s first-order responses to our actions, and the second-order responses to those responses – collectively, what I call “n-order impacts.”There is so much stuff that has yet to be invented. There’s so much new that’s going to happen. People don’t have any idea yet how impactful the Internet is going to be and that this is still Day 1 in such a big way. - Jeff Bezos Click To Tweet
Key takeaways/applications: considering n-order impacts – the “and then what” – can allow us to make more effective decisions, and to have intellectual humility about the decisions we’re capable of making effectively.
Three brief examples of n-order impacts:
Putting “greats” in perspective. The path to success usually doesn’t escape hard work – but when everyone works harder and harder, the level ofabsolute skill may rise, while the level of relative skill– i.e., what allows us to prevail over opponents – falls. Mediocre NFL players today would likely run circles around all-time greats in their prime decades ago. Michael Mauboussin explores this quantitatively in “ The Success Equation” ( TSE review + notes); this arms race phenomenon, where each side invests more and more for no return, is explored in more depth in the zero-sum vs. win-win games mental model.
Anddddddd it’s gone. While it’s comforting to believe that if we could only separate ourselves from our schema, we’d be able to identify objective truths, that’s not always true. Sometimes reality depends on what we believe. Consider, for example: is a bank that’s made some dodgy loans solvent? The answer involves n-order impacts and path-dependency: it might be solvent if depositors believe it’s solvent andsleep well enough at night to let the bank deal with the challenge; it’s almost certainly not solvent if depositors decide en masse the bank’s not solvent and, consequently, pull their money. This is an example of reflexivity, which we’ll explore in more depth.
Health decisions aren’t always easy. Parents face challenging n-order impacts when raising kids. Even the best and most laudable of intentions – for example, maintaining a clean / sanitary environment – can have unintended consequences.
David Oshinsky’s Pulitzer Prize winning Polio: An American Story ( PaaS review + notes) explores how, paradoxically, the early-1900s cleanup of America led to tremendously better health outcomes on the whole, but also unintentionally created the breeding ground for the polio epidemic.
More recently, oversanitized environments during early childhood have been implicated in increasing prevalence of allergies and asthma, which farm kids (exposed to more microbes) may suffer less from. Obviously no reasonable person would suggest doctors stop washing their hands, but maybe cleanliness is dose-dependent and it’s okay for kids to follow the chocolate exemption to the five-second rule every so often. And we won’t even go into antibiotic-resistance (an intersection of n-order impacts and trait adaptivity).
If this sounds interesting/applicable in your life, keep reading for unexpected applications and a deeper understanding of how this interacts with other mental models in the latticework.
However, if this doesn’t sound like something you need to learn right now, no worries! There’s plenty of other content on Poor Ash’s Almanack that might suit your needs. Instead, consider checking out our discussion of the sleep, stress/humor, orvividness + salience mental models, or our reviews of great books like “ Getting to Yes” ( GTY review + notes), “ How To Win Friends and Influence People” ( HWFIP review + notes), or “ Rust: The Longest War” ( Rust review + notes)
N-Order Impacts / Unintended Consequences Mental Model: A Deeper Look
“I’m all for fixing social problems. I’m all for being generous to the less fortunate. And I’m all for doing things where, based on a slight preponderance of the evidence, you guess that it’s likely to do more good than harm…
What I’m against is being very confident and feeling that you know for sure that your particular intervention will do more good than harm given that you’re dealing with highly complex systems wherein everything is interacting with everything else.”
N-order impacts pop up all over the place in life and business. For example, safety innovations like ABS (antilock brakes) certainly make cars safer, but don’t necessarily reduce the number or severity of accidents, as drivers compensate for the shorter stopping distance of the new brakes by driving more aggressively.
Megan McArdle highlights a number of such unintended consequences in “ The Up Side of Down” ( UpD review + notes), ranging from normalization of incarceration to how antibiotics are like antilock brakes. Jerome Groopman touches on them frequently as well in “ How Doctors Think” ( HDT review + notes). For example, he cites cardiologist Dr. James Lock, a clearly-brilliant and seemingly thoughtful medical practitioner, who discussed the cause of one of his mistakes:
“My mistake was that I reasoned from first principles when there was no prior experience. I turned out to be wrong because there are variables that you can’t factor in until you actually do it.”
Later in the book, on page 199, Groopman notes as well that while computer-aided detection improved radiologists’ performance in some instances, in others, it had the unintended consequence of:
“shaking the power of a specialist in his initial diagnosis.”
As you can imagine, n-order impacts interact with a bunch of other models. You can find interactions scattered across the Poor Ash’s Almanack latticework, such as in the trait adaptivityand incentives mental models.
To be completely honest, I think those interactions are more interesting than the ones that were left for me to present here (since I’m trying my best not to be repetitive). I would recommend reading those other ones before the ones below.
Nonetheless, here are a couple specific additional phenomena worth naming.
N-Order Impacts x Inversion x Feedback xDisaggregation = “Reflexivity”
Reflexivity is one of those fancy-sounding words (like “emergence” or “complex adaptive systems”) that’s really not that complicated but is useful shorthand for a common phenomenon. This is the phenomenon discussed with regard to banks in the introduction: the idea that solvency is not always a yes-or-no issue, but something more gray.
This often happens in the financial world in a variety of contexts. How much is a company worth? One answer is its net present value, the discounted sum of its future cash flows. Another answer is “however much someone’s willing to pay for it.” In many cases, such as financial bubble, assets’ value is based on people’s expectations of their value. Feedback can drive a virtuous circle… or a vicious cycle.
This is particularly prevalent with regard to companies whose business model requires being able to raise capital – usually from the equity markets – on favorable terms to finance acquisitions or investments.
You can see how this creates “reflexivity” – the company’s future cash flows are dependent on it being able to deploy capital, and its ability to deploy capital is based on, you guessed it, investors’ estimates about its future cash flow (and, likely, even their estimates about the company’s ability to raise capital from other investors!)
Rollup-blowup pharma company Valeant is perhaps the most salient recent example of this sort of situation that is in the history books. CEO J. Michael Pearson was hailed by many value investors as a legendary, revolutionary manager… until, um, one day that wasn’t the case.
The company’s business model, of course, suffered from a number of other problems – such as their strategy of jacking up prices by orders of magnitude on nondiscretionary/critical drugs, which egregiously violated people’s sense of fairness – but one of the contributing factors to their implosion was the n-order impact of the equity market losing confidence in them.
If they’d been able to continue using their shares as currency, they might not have had a debt problem… but all of a sudden they did. Valeant, of course, is just the latest in a long line of such stories.
“McClendon and Ward kept leasing more acreage, regularly spending more than Chesapeake had. It forced the company to borrow big sums, leaving it in a precarious position. “It was near-death on a daily basis,” Ward says.”
You’d be forgiven for thinking that was in the mid-to-late 2000s… nah. It was in, like, 1993. Chesapeake got away with this approach for a long time because of a good deal of luck and the continuing support of the capital markets… and then one day, that support vanished.
The most interesting example of reflexivity in The Frackers actually has nothing to do with Chesapeake per se, but rather with the personal finances of (now-late) CEO Aubrey McClendon. From my notes on pages 284 – 285 and 288 – 289 of The Frackers:
Aubrey McClendon borrowed what he thought to be a “conservative” 33% of his stock value in margin debt (a cool $300MM) to finance purchases of more stock (!) in addition to funding his lifestyle. As long as things keep going up and to the right, it works swimmingly…
… until one day it doesn’t, and the banks more or less liquidate McClendon’s collateral (the fast-falling shares of Chesapeake), further depressing shares and taking a hatchet to his net worth.
This is an example of debt capacity being reflexive.
Misunderstanding of reflexivity can hurt investors another way, via inversion. Many value investors, with completely indefensible logic, like to bemoan (with the benefit of hindsight bias) companies having repurchased their shares in the past at prices higher than the current price, thereby not maximizing shareholder value.
But except to the extent that management’s share repurchase decisions say something about their talent as capital allocators, if the share price is low enough today, it doesn’t really matter, because for management to repeat their error, the stock price would have to go up (benefiting you).
Application/impact: watch out for situations that would create “recursion” loops in Excel – if some assessment depends on a future event which, in some way, depends on your assessment, you’re looking at a reflexive situation that requires careful thought.
N-Order Impacts x Nonlinearity = Network Effects
We’ve mostly been focused on the challenges of n-order impacts, but they can be positive, too. For example, Shawn Achor’s “ The Happiness Advantage” ( THA review + notes) makes a compelling incentives-based case for kindness and empathy, reviewing a lot of the research on how small acts of deliberate kindness – as small as holding open a door or paying for a stranger’s coffee – can make us happier, far outweighing the modest opportunity cost.
It shouldn’t escape anyone’s notice that the n-order impact of everyone (selfishly) being kinder would, of course, be a happier and nicer world for everyone.
Less warm-and-fuzzy: if you’ve studied the business models of companies in the digital era, you’ve probably run into the term “network effects.” In Peter Thiel’s “ Zero to One” ( Z21 review + notes), discussed in more depth in the “arms race” interaction in the zero-sum vs. win-win games, network effects are highlighted as one of the strongest potential competitive advantages a business model can have.
Of course, network effects predate the internet, and are the reason that businesses as varied as Visa/Mastercard and FedEx/UPS have dominant positions. The theory behind the visual to the right is as follows: imagine trying to figure out the number of handshakes at a party.
At a party of one (like the kind of parties I throw, because I’m really boring), there would obviously be zero handshakes. At a party of two, there would be one handshake. At a party of three, there would be three handshakes.
At parties of four and beyond, the mental math gets a little difficult, but it turns out there’s an easy way to think about it quantitatively. A handshake involves two people, like so:
____ <-> ____
At a party of n people, there are n people who could be on the left side of the handshake. Since we can’t shake hands with ourselves (unless we want people to look at us funny), there are n – 1 people everybody could shake hands with: everybody but themselves.
__N__ <-> __(N-1)__
If you multiply those two terms – i.e., n x (n – 1), you count all the possible handshakes. Twice, it turns out: because using that framework, “John-James” would be counted, as would “James-John.” So it turns out the number of handshakes at a party would be n x (n-1) ÷ 2. This is an example of a nonlinearity, where the number of connections in the network increases much faster than the number of participants.
In probability terms, this would be referred to as a combinatorics problem – and we’d be solving for “n choose 2” to find the number of connections in a network with n nodes – but I won’t go into factorials here because, um, I don’t think anyone would care, and it’s not very useful for our purposes.
Bringing this back to n-order impacts, why do we care? The answer is that having more potential “handshakes” makes a network more valuable.
In many businesses, there are no n-order impacts to additional participants: if my friend also starts buying burritos at Chipotle, that doesn’t make my Chipotle experience any better. It might make it worse, if I now have to wait in line longer. The same goes for most companies that make “widgets.” The “and then what” for another customer dining at Chipotle, from my perspective, is “pretty much nothing.”
But in businesses with network dynamics, the addition of another node isn’t just like opening a new store. It has an n-order impact on the value proposition for every other node: if Visa is accepted everywhere rather than just at one store, all of a sudden Visa is more valuable to all the customers who use it… and if more customers use it, more vendors will be pressured to accept it.
Application/impact: while “unintended consequences” can make second, third, and n-order impacts sound like they’re always negative, they don’t have to be. Look for opportunities, like network effects, where n-order impacts can be positive.