← All insights

A · Manage · 2026-10-13

First Understand How Each Person Thinks, Then Work Out How the Whole Moves

Illustration · not a service record: Ask: a deep conversation at the whiteboardIllustration · not a service record

The rules on campus incident meetings (校事會議) have been amended three times in three years. The problem may not lie in the provisions, but in whether anyone asked at the start: how will each person react? The Nobel Prize in economics announced this Monday happens to be about exactly this.

<The Groundedness of Major Education Policy, Seen Through the Nobel Prize in Economics>

Why would a well-intentioned education policy need its rules amended three times in three years?

The Nobel Prize in economics, announced this Monday (October 12), happens to offer an answer: people are not cells in an Excel sheet. People rewrite their own formulas.

【Economics once thought it had found a law】

In 1958, the economist A. W. Phillips found in nearly a century of British data that in years when wages rose quickly, unemployment was usually lower. Later people extended this into “when inflation goes up, unemployment goes down, and vice versa,” and called it the Phillips curve.

Policymakers took one look and thought: wonderful, tolerate a little more inflation and we get more jobs in exchange.

Then came the 1970s: prices soared and unemployment did not fall either. Textbooks call it “stagflation.” The strategy guide had been written perfectly well, and then the game was updated.

The economist Robert Lucas, who later won the Nobel Prize, put his finger on the problem: people form expectations, learn, and change their behavior. Once everyone knows the government intends to trade inflation for employment, workers ask for raises first and companies raise prices first, and the curve shifts away. A statistical relationship from the past may stop holding the moment it is used as a policy lever. This is called the “Lucas critique.”

【What this year’s laureates did】

This year’s prize went to Robert E. Hall of Stanford and Michael Woodford of Columbia University, “for microeconomic foundations of macroeconomic stabilization.”

In plain words: they stopped looking only at the statistical relationships among big numbers such as inflation and output. Instead they built models starting from “how each household spends and how each company sets prices,” and then added these up into the whole economy. Hall studied household consumption; Woodford studied how firms set prices, including why prices do not adjust right away.

The chair of the prize committee said these models are now used by central banks and governments around the world, leaving us better able to respond to shocks such as an energy crisis or a sudden change in household consumption.

In one sentence: first understand how each person thinks, then work out how the whole moves.

【Education policy needs “micro foundations” too】

Anyone who has studied medicine knows that however effective a drug is in the lab, it does no good if the patient does not take it on schedule.

Education policy is the same. Aggregate indicators and past statistics matter, but the moment a policy is rolled out, teachers, students, parents and principals all redo their own sums: what risk does this pose to me? How much more time will I have to spend?

The final effect of a policy is what comes out when the decisions each of these people makes are “added up.”

【Example 1: Campus incident meetings】

The campus incident meeting is a procedure for handling “teachers suspected of being unfit,” introduced in 2020. The original intent was entirely legitimate: teachers who truly harm children should not be able to hide behind the system.

The aggregate logic is also reasonable: accept more complaints → investigate more cases → children are safer.

But go back to each person’s decision:

・For some parents, the cost of filing a complaint is very low (in the past it could even be anonymous), so it can turn from a “last resort” into the first option when communication gets stuck.

・For principals and administrators, not investigating may get them accused of “burying the case,” so the safest choice is to “investigate first and see.” Teachers’ groups have also pointed out that this is one reason minor cases get handled as major ones.

・For teachers: one classroom teacher has said that the more a teacher cares about the children, the more likely that teacher is to be the subject of a complaint. So what is the safest strategy? Step in a little less.

・For the child who has truly been hurt: if a complaint cannot be anonymous, will the child be afraid of being identified and simply say nothing? Student groups have raised this concern too.

No one is in the wrong; each is making a rational decision from where they stand. But added up, the result may be teachers who dare not step in and children who dare not speak, with neither side protected.

The numbers reflect this complexity too. A report by the National Human Rights Commission shows that in the 113 academic year (2024–25) there were 2,486 complaints nationwide involving teachers alleged to be unfit, and about 78% led to a campus incident meeting investigation. Those that ended in dismissal, suspension or non-renewal came to 53 in total. Teachers’ groups read “abuse of the complaint system” in these figures; the National Human Rights Commission points out instead that in nearly 70% of the substantiated cases of improper discipline, the penalty was a reprimand or lighter. The same set of aggregate numbers can be read to reach opposite conclusions, because it does not tell us why each person acted as they did.

The amendment in January this year removed anonymous reporting and added a triage step. The Ministry of Education says that by the end of May the number of cases had fallen from about 448 in the same period last year to 202, and the share that went on to investigation fell from about 82% to about 22%. The ministry also says it will not treat the drop in cases as a policy achievement, and I very much agree. Nearly half of the cases have been diverted to the teacher evaluation process, so has the pressure really eased, or has it just moved somewhere else? We will only know by watching what people on the ground actually do.

The new amendment of October 7 further provides that where something is clearly reasonable discipline, the school may close the case directly. Parents’ groups have also endorsed this direction. But whether it succeeds may hinge not on the provisions, but on whether a principal, at the moment of signing off on “close the case directly,” feels sure of their footing.

【Example 2: The AI Talent Ark Program】

In May this year, the Ministry of Education launched the “AI Talent Ark Program” (AI 人才方舟計畫), about NT$11.755 billion over four years. It includes achieving “a tablet for every student” within four years at the latest, as well as AI teaching assistants, AI tutors, education big data and more.

The aggregate logic: devices in place + content in place + training in place → children’s academic ability and AI skills improve.

But go back to a single classroom:

・A junior high school teacher has 45 minutes per class. Handing out tablets, logging in, dealing with the ones that are out of battery and the ones with a stalled connection: a chunk of the time is gone. The sum in her head is: does AI save me time, or is it one more job? If something goes wrong when a student uses AI and a parent files a complaint, who carries it? You see, in a teacher’s mind, campus incident meetings and AI policy are actually entries in the same ledger.

・When a student gets an AI tutor, the first thought may not be “let me ask a good question” but “give me the answer.”

・When parents see their child bring a tablet home, the first thing they worry about may be screen time.

・If schools are required to report a “usage rate,” the most rational thing to do is to push up the number of logins. Once an indicator becomes a target, it stops being a good indicator. This is the education version of the Lucas critique.

To be fair, this program puts a substantial budget into “digital teaching support and guidance,” about NT$1.6 billion a year, and it also has a teacher education alliance. The direction is toward “people,” and that is right.

The real test is this: what do we look at to judge results? People who work in digital (myself included) easily stare at the numbers on a dashboard and feel pleased, and forget to ask why teachers use it and why they don’t. I would suggest that beyond the usage rate, we look more at behavior and stay curious about the reasons behind it: have teachers saved time as a result, and spent more of it with the child who is falling behind? Have children become better at asking questions? At Junyi Academy we build AI tutors for math, science, Chinese and English, and what the third-party study saw was not a surge in scores either. It was that students’ results did not drop when they were not using AI, and their ability to ask questions actually got better.

I wrote earlier that I hope we can move from “a tablet for every student” to “Tokens for every teacher.” Today I want to add one line: before that, we first have to understand the ledger in teachers’ minds.

【Crouch down first, then move forward】

It took economics decades to put “people” back into its models. Education has always been about people, so all the more reason it cannot look only at curves.

Before any major education policy is rolled out, perhaps we could first ask three questions:

1. For each kind of role on the front line, what is the most rational response to this policy?

2. Added together, will these responses run opposite to the direction we want?

3. Will the indicators we watch become distorted the moment the policy is pushed out?

This is of course not easy. In fact, the challenges decision-makers face and the variables they need to balance are often far more numerous than most people take into account. But one fundamental does not change: before a policy formally takes effect, find a few principals, homeroom teachers, parents and students, take a real case, and ask one question: “Under the new rules, if it were you, what would you do?” That is the micro foundation of education policy.

Of course, sometimes when the question comes directly from the government, people may not dare to tell the truth. So frontline opinions compiled by NPOs that take no government funding but work in the relevant field may have some reference value.

After winning the prize, Hall said that California taxpayers gave him a very good education in public schools and at Berkeley, and that the federal government paid for him to attend MIT. “I hope they feel the return on this investment has been reasonably satisfactory.”

Public education is an investment whose returns can only be seen decades later. Precisely because the payback period is so long, every major education policy deserves that we first crouch down and truly hear each person.

First understand how each person thinks, then work out how the whole moves.

(References are in the comments.)