Vol. 10: Grab That One First: Which Problem to Solve First, from "On Contradiction"

That evening, I was sitting at my computer with a glazed look in my eyes.

After a whole day of web coding, my brain had stopped turning. Someone with no programming training, stubbornly trying to build a piece of software from scratch: I was making Wenchi, a writing tool that helps people write novels. At the end of the day, staring at the screen, the strongest feeling wasn't tiredness. It was emptiness. That whole day, the project had barely moved.

Deep down I knew why. I wanted to do everything. I wanted the interface to look a bit nicer, I wanted the first-level and second-level pages built, I wanted to tidy up this and that while I was at it. Each one on its own was worth doing. Lumped together, I kept jumping back and forth between them, starting all of them and finishing none. I could spend most of a day fiddling over whether a button belonged on the left or the right, or whether a second-level page was even needed. The more I did, the more anxious I got, and the more anxious I got, the more things I wanted to hold at once, to prove I was making progress. The result was a tangled mess.

That night I went out for a walk. Somewhere along the way, the string that had been pulled tight in my head loosened a little, and a very plain thought surfaced: I don't actually need to rush to hold everything at once. I should build the main axis of this product first, get the most core, most fundamental feature standing, and deal with the rest later.

When I got back, I did just that. It turned out to be right. Build the main axis solid first, then add details layer by layer: that's how a piece of software ought to grow.

It's not a big deal. But it lands right on the question I left you at the end of the last post.

A Pile of Contradictions: Which to Grab First?

The last post was about the particularity of contradiction. It said that once you really analyze something concretely, you find it never has just one contradiction. That evening I was facing a pile of them: the interface, the pages, the core feature. Each was real and each was calling to be solved. But a person only has so much strength, and it can't be spread evenly. So the question became: out of this pile, which one do I grab first?

Mao devotes a whole section of "On Contradiction" to answering this.

I've covered the background in the last three posts: 1937, Yan'an, curing the Party's dogmatism. I won't go into it again. Let's go straight to the core of this section. Earlier, he made clear that contradiction is everywhere and every contradiction is different. In this section, he answers a more practical question: when several contradictions sit in one process at the same time, are they equal? His answer is blunt. No.

There are many contradictions in the process of development of a complex thing, and one of them is necessarily the principal contradiction whose existence and development determine or influence the existence and development of the other contradictions.

The whole weight of this sentence is in "determine or influence the other contradictions." It means this pile of contradictions isn't each minding its own business. There's one in the lead. When the leader moves, the rest move with it. Pin the leader down, and the rest become manageable. So he makes a judgment that's almost a promise:

Once this principal contradiction is grasped, all problems can be readily solved.

He didn't say "readily solved" lightly. Right after, he says that people who study things or do things, if they don't know the method of grasping the principal contradiction, will be "lost in a fog, unable to find the center and so unable to find the way to resolve contradictions." Lost in a fog: that was pretty much me at my computer that evening.

He uses the example he knew best. China at the time had a whole heap of internal contradictions, between landlords and peasants, between the various factions, all sitting there as usual. But once Japan invaded, the situation changed instantly. The national contradiction shot to the front and became the principal contradiction, and the domestic contradictions that had mattered so much retreated to a secondary place for the time being. Everyone faced the outside enemy together first. When the outside pressure changed, the order of contradictions would be reshuffled again.

There's one point to note here, because we'll need it later: the principal contradiction isn't fixed in place. As a process reaches different stages, the lead role changes hands.

He puts this method in a very high place. In his view, the first thing in studying any complex process is to work hard at finding its principal contradiction. Find it, and you've lifted the headrope of the net, and the meshes open on their own. Fail to find it, and all you can do is rush around blindly among a pile of problems, never landing your effort where it counts, however hard you push.

Not the Most Urgent, but the One That Moves Everything

Here's this section twisted into one sentence to take away:

The pile of problems in front of you is never equal. There's always one principal contradiction that, once solved, loosens the rest. Finding it and putting your strength on it is the first step in getting anything done.

This sentence overturns an instinct buried deep in us. When there are too many problems, we usually react in one of two ways, both very natural and both traps.

The first is spreading it evenly. You try to cover everything, a little for everyone, and your strength turns into a thin mist that soaks nothing.

The second is grabbing the loudest one. Whichever looks most urgent, most glaring, most unbearable, you pounce on first. But honestly, the loudest one often isn't the most important. It's just closest to your emotions. That doesn't make it closest to the root of the whole.

Grasping the principal contradiction asks for a judgment different from both. It doesn't ask "which is most urgent?" It asks "which one, once solved, will loosen the others?" The answers to those two questions are often not the same. And emotions love to steal the scene. The more anxious you are, the easier it is to mistake "what's bothering me most" for "what matters most." My mistake that evening wasn't laziness. It was being too diligent: diligently pouncing on every glaring little problem, and never once stopping to ask which of them was the thread that moves everything when you pull it.

My Whole Day Was Spent on Secondary Work

Now let me retell that evening's mess in this language, and it becomes clear.

Whether the interface looks good, how many levels of pages there are: these are all contradictions, but secondary ones. They share one thing: they're all branches and leaves growing on the main axis of the core feature. Pruning the branches before the trunk is standing not only goes badly, it also blocks your view of the trunk. So the truth about my "no progress" day wasn't that I didn't work. It was that I did too much secondary work. All my strength went to the branches, the trunk didn't grow an inch, and of course I felt empty.

Learning these things over the past two years, I've come to feel more and more that this runs in one line from the last two posts. In Vol. 8, "Without Contradiction There Is No World," I wrote about learning AI like this:

These past two years I've thrown myself into learning how to use AI, and looking back, it's been a level that never ends.

In Vol. 9, "One Key for One Lock," I added another layer:

That the levels never end speaks to the universality of contradiction. That each level has a different key speaks to the particularity of contradiction.

This post needs a third sentence: when several levels pop up at once, they also rank as principal and secondary. Not every one should be attacked right now. That's exactly what went wrong in my worst stretch. I treated every level that showed its head as the principal contradiction and charged at each one in turn, until I was battered and dusty. Then I looked back and saw that the one door actually blocking everything, I had never once properly pushed. The day ended with no progress at all, and that emptiness of being busy all day with nothing to show is much harder to bear than plain tiredness.

The thought I had on my walk, "build the main axis first," translated into this section's terms, is: first recognize the principal contradiction, and switch your strength from spread out to concentrated. Once the core feature runs, where the interface goes and how the pages split suddenly have a basis, and many questions I'd agonized over found their own answers. That's what Mao meant by readily solved: the secondary problems don't vanish into thin air. Once the principal contradiction clears, each of them suddenly falls into place.

Figuring it out is one thing. The hard part is going back and actually doing it. The first thing I did wasn't to bury my head and keep working. It was to stop. I paused all the interface and page work and left only the core feature on my desk. A few days later, once the core ran end to end, I looked back at the branch-and-leaf problems that had me scratching my head, and most of them were no longer problems.

This reminds me of Steve Jobs. When he came back to Apple in 1997, the company was close to collapse and its product line was a mess. There was a pile of Mac models alone, so similar that even the salespeople couldn't explain the differences, plus printers, handhelds, and other odds and ends. The conventional thinking would be: with the company in this much trouble, shouldn't you add more product lines and push more stock to survive? Jobs did the opposite. He drew a very simple grid on a whiteboard: desktop and portable across, consumer and professional down. Four boxes, one product in each. Nearly everything else, cut. There was plenty of resistance. In many people's eyes, with the company where it was, every extra product line meant one more chance of survival.

Cutting about 70 percent of the products in one go sounds like throwing away the family savings. But the principal contradiction he had fixed on was never "not enough products." It was "no product good enough that people have to buy it." He pulled people and money out of a pile of mediocrity and poured all of it into those four boxes, and Apple slowly recovered. It's really the same principle as my little piece of software, just a few hundred times bigger: grasping the principal contradiction usually starts not with adding but with subtracting. First pull your strength back from the secondary contradictions, then put all of it on that one.

There's another layer I only came to feel later. Once the main axis is standing, things aren't over. The center of gravity shifts. Once the core feature was stable, the interface and pages, the old secondary contradictions, rose up and became what I needed to focus on in the new stage. That's exactly the point I noted earlier: the principal contradiction isn't fixed in place, and when the stage changes, the lead role changes hands. So grasping the principal contradiction isn't a one-time deal. Every so often you have to ask yourself again: at this stage, is the thread that moves everything still the one it was a moment ago?

Telling Principal from Secondary Is a Skill

Looking back on that little stretch from the evening to the walk, what I learned wasn't some development trick. I learned a skill of ranking: facing a pile of things you want to do, stop first. Don't rush to ask which is most urgent. Ask which one moves the whole. People who can really tell principal from secondary aren't stronger than others. They know where to land the blow.

But the principal contradiction is about who leads and who follows among several contradictions. Look further in, and contradiction has another layer. Even once you've locked onto one contradiction, say you're only watching "the core feature," it's itself twisted together from two opposing sides: usable and unusable, simple and complex, build and cut. Those two sides are squeezed inside one contradiction, and they aren't equals either. One of them has the final say.

So between the two sides inside a contradiction, which one decides what the thing looks like right now? We'll pick that up in the next post.