Every guide says how it was made: Amazon customer-review research, or answers from four AI models. A human editor checks each one. As an Amazon Associate we earn from qualifying purchases. Learn more›
Updated: October 21, 2026
Your Fix Worked, Then It Made Something Else Worse: Six Second-Order Thinking Examples
Second-order thinking means following a decision past its first effect by asking “and then what?” about the people who react to it. A rule limiting office coffee to two cups works on day one. Then heavy drinkers leave for the café on work time, and the manager becomes the coffee police. To use it today, list who your action touches. Ask “and then what?” two or three rounds deep, with a date on each effect. Circle anything that cancels the first win or is hard to undo.
What is second-order thinking, without the definition fog?
It is one question asked twice: and then what? The first effect of a decision is the one you planned. The second and third are the ones you get, because people adjust around whatever you just changed.
The cleanest sourced version comes from the biologist Garrett Hardin, in a 1963 paper on competition: “We can never do merely one thing.” Change a price and customers react. Then competitors react to the customers. The economist Thomas Sowell later built a book on the same point for public policy, subtitled Thinking Beyond Stage One.
Most explanations stop at “think about the long term.” That advice is too vague to use on a Tuesday. You already think about the long term. What you skip is the step in the middle: picturing how a specific person, who wants something different from you, will respond.
Six examples, each traced two steps past the first effect
Each one is a real case from the research or a scenario from the book. The trouble always came one or two steps later.
Example 1 · Work
The two-cup coffee rule
The first move
A small office runs out of coffee by ten each morning. The manager tapes up a rule: two cups per person per day. Within a week the coffee lasts all day.
And then
Heavy drinkers still want their coffee. Some walk to the café down the street and are gone twenty minutes. One person brings a thermos and shares it with a favored few.
And then
Time away from desks adds up, and the thermos starts a quiet fight about fairness. The manager spends mornings hearing complaints about cups. The budget line went down while the real cost went up.
What a better move looks like
Buying more coffee might cost less than twenty minutes of everyone’s time. Or keep the rule, but test it for two weeks before it becomes policy.
Try it on your own decision: Who wants the thing you are limiting, and where will they go to get it?
Example 2 · Family
The late-pickup fine
The first move
In January 1998, six of ten day-care centers in Haifa, Israel, started fining parents 10 shekels per child for arriving ten or more minutes late. The goal was fewer late pickups.
And then
Late pickups went up, not down. Before the fine, being late meant imposing on a teacher who had to wait. After it, being late meant buying extra time, and buying it cheaply.
And then
In week 17 the fine was dropped. Lateness did not fall back. It climbed again, because the extra time was now simply free.
What a better move looks like
The researchers named the lesson in their title: a fine is a price. Before you add a penalty, ask what it tells people the situation now is.
Try it on your own decision: Could your rule turn a favor or a shared norm into something people can buy?
In the six fined Haifa centers, late pickups rose from about 8.0 to 16.4 per center per week, while the four centers with no fine went from about 10.0 to 8.25 (Gneezy and Rustichini, Journal of Legal Studies, 2000).
Example 3 · Money
The agency paid per lead
The first move
A small business owner compares two marketing agencies. Agency B charges less up front and takes a bonus for each new lead. Its pitch is better and its numbers look stronger.
And then
Agency B will bring in leads, because leads are what it gets paid for. Whether those leads turn into paying customers is not part of its reward.
And then
The owner also finds her own line. She liked B’s founder and had already called it the frontrunner. Her quiet reward for picking B is not looking like she changed her mind.
What a better move looks like
She asks B to tie part of its bonus to signed customers, not leads. She asks Agency A, on a flat fee, what happens in a slow month.
Try it on your own decision: What is being counted here, and what quality could slip because nobody counts it?
Example 4 · Health policy
The checklist required by law
The first move
Earlier studies linked a surgical safety checklist to fewer deaths. So Ontario, Canada, required surgical checklists in its hospitals. A 2014 study covered 101 hospitals and more than 200,000 procedures.
And then
Every team now had a checklist. But the lists were handed down by policy, not built or championed by the teams who had to stop and run them.
And then
Deaths and complications barely moved, and neither change was statistically significant. The follow-up lasted only three months, which is one possible reason. The other is that a ticked box is not the same as a team that stops to check.
What a better move looks like
The book’s reading: checklists seem to work best when the people using them built them and believe in them. A mandate changes the paperwork first. Behavior is the second effect, and it is the one you need.
Try it on your own decision: Does your rule change what people do, or only what they report?
After Ontario required surgical checklists, mortality went from 0.71 percent to 0.65 percent and complications from 3.86 percent to 3.82 percent, neither statistically significant (Urbach et al., New England Journal of Medicine, 2014).
Example 5 · Work metrics
The support team scored on speed
The first move
A manager scores the support team on how fast tickets get closed. The dashboard turns green within days.
And then
Tickets get closed fast, because closing is what the score rewards. Whether the customer’s problem is solved is a separate question.
And then
Nobody is counting that second question, so it drifts. People follow the number, because the number is what pays.
What a better move looks like
Look hard at what the dashboard does not count. Write down the quality that could slip, here whether the problem was solved, and find a way to check it.
Try it on your own decision: If people chased only this number, what would they stop doing?
Example 6 · Pay design
The night shift paid by the hour
The first move
In Charlie Munger’s 1995 Harvard talk, the Federal Express night shift kept failing to move packages between planes in time. The workers were paid by the hour.
And then
Paid by the hour, a slow night earns more than a quick one. Nobody has to be lazy for the shift to run long. They only have to follow the pay plan.
And then
In Munger’s telling, paying by the shift fixed it. Finishing early now meant going home early, so the second effect worked for the company.
What a better move looks like
Treat this as Munger’s illustration, not company history. The book’s research found no independent FedEx record of it.
Try it on your own decision: What does a slow day earn the people doing the work?
Across 39 studies, financial incentives had a solid link with the quantity of work, a corrected correlation of .34, and no relation to its quality (Jenkins et al., Journal of Applied Psychology, 1998).
Notice the pattern. Nobody in these stories was foolish or dishonest. Parents, agencies, surgical teams and shift workers all did what the new rule paid for. The people who wrote the rule simply stopped thinking one step too early.
The 3-question test you can run on a decision today
Take one change you are about to make that other people can react to. A price, a house rule, a target, a new policy for the team. Give it ten minutes and a notebook, and ask these three questions in order.
Who does this touch, and what do they want? Write the action and the first effect you want, such as “Charge for parking. Fewer people drive in.” Then list everyone it lands on: customers, coworkers, family, competitors, and whoever has to enforce it. Think about what each group wants, not what you want them to do.
And then what, and when? For each group, write their likely response. Then ask “and then what?” about each response. Go two or three rounds deep and stop when the answers turn into pure guesses. Next to each effect, write when it would show up: a week, a month, a year.
Does anything later cancel the win or lock you in? Circle any effect that undoes the first one or would be hard to reverse. If you circle something, you have three exits. Change the action. Test it small first. Or set a date to check what actually happened.
Write one line in your notebook when you finish: the action, the intended first effect, the second and third effects, the check date. When the date comes, add what actually happened. That last line is where you learn which of your predictions were real and which were stories.
Where does second-order thinking mislead you?
It misleads when you trust it more than the evidence allows. The book labels every tool by how well it holds up, and “And Then What?” gets STORY ONLY. Nobody has run a controlled test of prompting people to ask the question before they act.
That does not make it useless. STORY ONLY means the practice is real and well documented, but untested. It is also one of the cheapest tools in the book. Asking twice costs minutes, and missing the second effect can cost much more. Four traps are worth knowing.
Chains of guesses. Each round adds uncertainty. By the fourth “and then what?” you are writing fiction, which is why the tool says to stop when the answers become pure guesses.
Small choices you can undo. The tool matters most for prices, rules and incentives whose effects build on each other. For a one-time choice you can reverse tomorrow, just decide and watch.
False certainty. An honest answer is rarely one outcome. It is “probably this, maybe that, possibly worse.” You end up facing odds, not a forecast.
The wrong question. Howard Marks’ “second-level thinking” asks what everyone else already expects. It is a separate idea about beating the consensus, and many popular books blend the two.
One more caution on quotes. A longer version of Hardin’s line adds that any intrusion into nature has numerous, often unpredictable effects. The book could not confirm that second sentence in his 1963 paper. Only “We can never do merely one thing” has the source behind it.
Which thinking tools pair well with it?
Three tools from the same book cover the blind spots that “and then what?” leaves open. One looks backward, one checks the rewards, and one checks your feelings across time.
Chesterton’s Fence: why is this here before you remove it?
Second-order thinking looks forward. This tool looks back. Before you scrap a rule you did not write, find out what it was for. In the book, a new manager at a car repair shop wants to drop the phone-number check that slows the front counter.
The longest-serving mechanic explains it. Pickup calls used to go to old numbers, and finished cars sat in the lot for days. So she keeps the check but moves it to the reminder text sent the day before. Chesterton never said to keep every fence. He said find out first. The tool is also STORY ONLY.
The Incentive Check: who gets paid if this happens?
Four of the six examples above were incentive stories. The Incentive Check makes that explicit. List everyone whose behavior matters, write what each is really rewarded for, and predict what they would do if they followed only that reward. Its badge is PROMISING, because the premise is strongly supported but the check itself has not been tested.
10/10/10: how will this feel later?
Some second effects land on you, not on other people. The 10/10/10 tool asks how each option will feel in 10 minutes, 10 months and 10 years. It helps when guilt or fear is loud now and the long-run effect is quiet. It is PROMISING: studies support stepping back in time, not the exact three-frame format.
Mental Models That Actually Work by Ada Fenwick is a 288-page book of 40 thinking tools. Each is traced to its source and given a badge: 6 SOLID, 20 PROMISING and 14 STORY ONLY.
This guide gives you the core of one tool and the cases around it. The book adds the rest of the 40, a Decision Selector that matches common situations to tools, and a 30-Day Clear Thinking Plan.
Second-Order Thinking: the coffee rule, the full “And Then What?” card, the car-shop fence and the Marks mix-up
Incentives: Who Gets Paid If This Happens?: the Haifa fine, the FedEx and Xerox stories, the agency case and the seven-step Incentive Check
Rules and Checklists: the four checklist studies, including the Ontario result, and how to build a list that works
Regret: Choosing Now So You Can Live With It Later: the 10/10/10 card and the research on stepping back in time
Decision Journals: how to log the check date and judge the decision apart from the outcome
Your fixes keep solving today’s problem and creating next month’s
You want to know which famous thinking tools have evidence behind them
You face a rule, price or target that people will react to
Skip it if
You want financial, legal or medical advice; the book gives none
You want the famous legends kept intact, since the myth checks remove several
You want one big theory instead of short tool cards you use one at a time
Fair doubts
“Isn’t this just common sense?”
Mostly, yes. The coffee rule’s later effects were not hard to predict. Nobody asked. The tool is a way to make sure you ask, on paper, from the side of the people who will respond.
“If it is STORY ONLY, why use it?”
Because it costs ten minutes and the downside of skipping it can be large. The badge tells you how much weight it can hold. When a SOLID tool points another way, follow the one with more behind it.
Questions people ask about second-order thinking
What is an example of second-order thinking?
An office limits coffee to two cups a day so it stops running out. That works. Then heavy drinkers walk to a café on work time and the manager spends mornings policing cups. Seeing those later effects before taping up the rule is second-order thinking.
What is the difference between first-order and second-order thinking?
First-order thinking stops at the effect you want: a fine means fewer late parents. Second-order thinking asks how people will respond, and then how others respond to that. In a Haifa day-care study, a small late fine roughly doubled late pickups.
Is second-order thinking the same as Howard Marks’ second-level thinking?
No. Marks’ second-level thinking asks what everyone else already expects, so you can think differently from the crowd. Second-order thinking asks what happens after what happens. Both are useful, but they answer different questions.
Who came up with second-order thinking?
No single person owns it. The cleanest sourced line is the biologist Garrett Hardin’s 1963 sentence, “We can never do merely one thing.” The economist Thomas Sowell made the same point about public policy in his 2003 book Applied Economics: Thinking Beyond Stage One.
How do you practice second-order thinking?
Pick one change you are about to make that affects other people. List who it touches, ask “and then what?” two or three rounds deep, and write when each effect would show up. Set a date to check what actually happened. Ten minutes in a notebook is enough.
