Between the Worlds: Rules and Rebellion

There were three things that meant education in our village: reading, writing, and arithmetic. Reading meant that you could read a newspaper, a letter, or perhaps the movie posters in Guntur. Arithmetic meant that you could manage basic operations and calculate interest. Writing meant that you could write a letter and fill out a form. Much of this eventually came down to one particularly useful skill: reading and writing promissory notes, the IOUs that were an annual fact of a farmer’s life.

Surprisingly, this was enough to get by. What would one do with algebra in a village? Calculus? Forget about it. And Shakespeare? I am not sure what practical problem Hamlet was expected to solve. Yet I learnt all of them, and I learnt the real lesson of school remarkably well: learn what is expected, reproduce it when asked, and you will do well.

By that measure, I was a good student. I did well in exams, remembered what I was supposed to remember, and learnt the rules of the game well enough to play by them. By the time I reached college, however, we had developed considerable contempt for this kind of education. We called it rote learning. We thought memorization killed creativity, exams rewarded obedience, and schools discouraged curiosity. The good student, in our new understanding, was the one who asked questions, explored, challenged assumptions, and discovered things for himself.

Later, America seemed to confirm this. Universities here spoke constantly about inquiry, independent thinking, research, and originality. Research itself seemed to be institutionalized curiosity: take something that everyone knows, ask whether it is really true, and perhaps discover something new. I absorbed the opposite lesson from the one I had learnt in school: education should teach us to question.

For a long time, I did not think of these two versions of education on equal footing. One was the bad education I had escaped; the other was the good education I had discovered. Of late, I am not so sure.

Consider what happens when we actually go to work. A new employee joins a company, and the first few weeks are spent learning how things are done there. How do we review code? Which system holds the customer data? How do we escalate a production problem? What is the release process? Who approves what? Before anybody gets a chance to be creative, there is an enormous amount of convention to learn.

The same thing happens almost everywhere. Factories depend on processes, airlines on checklists, hospitals on protocols, and banks on procedures. My wife is a critical care physician, and I am struck by how much medicine is driven by standards of care and accumulated practice. Judgment matters, particularly when a patient does not fit the standard case, but much of the system works because every physician does not invent treatment from first principles every morning.

Take language itself. There is nothing inherently superior about calling this object a “chair,” but once millions of us agree on the word, I can ask you to bring me one without drawing a picture. Arithmetic, money, and traffic work for similar reasons: we share symbols and conventions. Civilization depends on a large number of such agreements, most of which we barely notice because we learnt them so early. Education is one way we induct each generation into them.

I was reminded of this by the son of a friend whom I interviewed some years ago. I reluctantly concluded that he did not have what I considered a fundamental grasp of computer science. When I asked what would happen if the atomicity property did not hold in a database transaction, he could not answer. To me, this was close to the foundational definition of a transaction, and I did not think he was ready.

He eventually came to the United States, got a job at Amazon, and, to my amusement, is now managing database servers there. I do not think this proves that my interview was wrong, or that fundamentals do not matter. It did make me think about what organizations actually need from most of us. While we say that we want everyone to think outside the box, we also need many people who can work very well inside it: learn the rules, operate systems reliably, provide continuity, and make sure things keep working.

Large groups of strangers cannot cooperate unless enough of them agree about how things are supposed to work. States, religions, guilds, armies, professions, and eventually corporations all had reasons to teach common symbols, measures, procedures, conventions, and stories. One can say cynically that education produces obedient servants for institutions. There is some truth in that. But there is another description: education gives us enough common ground to participate in systems larger than ourselves.

That still leaves the problem that bothered my college-age self. What happens when the system is wrong?

History gives us plenty of cases where continuity was not a virtue. We have discarded hierarchies, laws, and customs that once seemed natural. Technologies and professions change as well. Education therefore cannot merely reproduce the status quo; the same system that teaches us conventions must also make it possible to revise them.

I think part of the confusion comes from treating all questioning as the same thing. A teacher who asks “why?” may simply be helping a student understand an accepted rule. Learning by doing, experimentation, and Socratic questioning are methods of understanding; none necessarily rejects the premises of the system.

There is a second level, where we understand why a rule exists, what problem it solves, and when it applies. A physician who knows a protocol also needs to know when the patient in front of her does not fit it. A programmer who knows the normal database rules should understand what breaks when one is relaxed. At this level, questioning does not destroy the system; it gives us enough context to use it intelligently.

And then there is a third level, much rarer and harder: deciding that the rule itself has outlived its usefulness.

Consider engineering education today. For most of my career, if somebody wanted to become a software engineer, I would have said: learn to program. Learn a language, data structures and algorithms, debugging, databases, operating systems. Write enough programs that programming becomes natural. This seemed almost definitional.

Now AI systems can produce substantial amounts of working code. Do students still need to learn programming languages in the same way? Should they spend months mastering syntax that a machine can produce instantly? How much programming must one know to judge code produced by something else? If writing code becomes a smaller part of software engineering, what should take its place?

I have opinions about these questions, but the more interesting point is that the premise itself has become questionable. “Learn to code,” which for decades was almost a definition of software education, is now something we have to defend, modify, or perhaps partly abandon.

This suggests a three-step view of education that I find more satisfying than either of my earlier beliefs. First, we learn the rules; civilization depends on people who can reliably participate in its systems. Second, we understand their purpose, context, and limits, enough to deal with exceptions and conflicting rules. Third, occasionally, we decide that the old rules no longer serve us and build new ones, carrying forward what we have learnt from the systems they replace.

The third step has a romantic appeal. Fitzgerald’s rendering of Omar Khayyam captures it wonderfully:

Ah Love! could thou and I with Fate conspire
To grasp this sorry Scheme of Things entire,
Would not we shatter it to bits—and then
Re-mould it nearer to the Heart’s Desire!

I sympathize with the sentiment. Yet, I do not prescribe it lightly. Without a deep understanding that comes from practice of the first two steps, the third step sets us up for failure. 

That is why I am now suspicious of both easy conformity and easy rebellion. A society that cannot change eventually becomes brittle. A society that constantly discards inherited systems spends its time rebuilding the arrangements that make ordinary cooperation possible. Education should offer us the balance: it has to preserve enough of the past that we do not begin from zero, while giving us enough understanding to recognize when the past has become a burden.

I encounter this problem now from the other side of the classroom. I teach young programmers Linux, Git, programming languages, testing, and conventions accumulated by generations of software engineers. When they skip the fundamentals and immediately reach for an AI system, part of me reacts exactly like my old teachers did: learn it properly first.

At the same time, I am rewriting those courses because of AI. Things I insisted on teaching a few years ago no longer deserve the same amount of time, while skills that barely appeared in the curriculum have become important. I cannot tell students to question assumptions while treating my own syllabus as sacred.

As a schoolboy, I learnt to follow the rules and was rewarded for it. As a college student, I learnt to distrust that kind of education and celebrate questioning instead. After spending much of my life working inside organizations, and now teaching another generation, I have come around to thinking that both lessons were useful. The difficult part is not choosing between learning the rules and questioning them. It is learning enough to know when to do which.

*