Constraints & Slackness as a Worldview Generator

Last post spec­u­lated that cap­i­tal was much more scarce (rel­a­tive to la­bor) in me­dieval/​re­nais­sance China than Europe. We in­tro­duced the hy­poth­e­sis that this was a ma­jor cause (though not a root cause) of the failure of China to au­to­mate and, by ex­ten­sion, the failure of China to kick-start the in­dus­trial rev­olu­tion. More im­por­tantly, we in­tro­duced some sim­ple first-pass ways to test this hy­poth­e­sis: go look at mar­ket prices of la­bor and cap­i­tal in me­dieval/​re­nais­sance China and Europe.

We did not go into depth ac­tu­ally test­ing that hy­poth­e­sis, nor will we. In­stead, we’re go­ing to go up a meta-level, and gen­er­al­ize this whole tech­nique. We have a very gen­eral hy­poth­e­sis about the large-scale his­tor­i­cal evolu­tion of China com­pared to Europe. It’s part of a whole wor­ld­view, in which the cen­tral gear—the cen­tral me­di­at­ing fac­tor—is the scarcity of cap­i­tal rel­a­tive to la­bor. What’s the pro­cess which gen­er­ates this sort of hy­poth­e­sis? Can we gen­er­al­ize it, to gen­er­ate other broad wor­ld­views?

Here’s the idea, in 4.5 steps.

Step 1: Pick a Gen­eral Re­source/​Constraint

Pick some very gen­eral re­source, which serves as an in­put to pro­duc­tion in a wide va­ri­ety of ar­eas. Our China/​Europe ex­am­ple mainly con­sid­ers cap­i­tal as the re­source (though we also use la­bor as a baseline for com­par­i­son in this case). Other im­por­tant ex­am­ples in eco­nomic his­tory in­clude trans­porta­tion, en­ergy, and fer­tile land. More mod­ern and ab­stract ex­am­ples might in­clude re­search/​in­no­va­tion, com­mu­ni­ca­tion/​co­or­di­na­tion, sig­nals of sta­tus/​virtue/​in­tel­li­gence, or ma­te­rial goods as a whole.

Step 2: Qual­i­ta­tive Analysis

Rea­son qual­i­ta­tively about what would hap­pen as the con­straint be­comes more taut/​slack. What tech­nolo­gies (ma­chines, con­tracts, or­ga­ni­za­tional struc­tures, etc) would be adopted? What would hap­pen to mar­ket prices? What other con­straints would be­come more/​less slack as a re­sult?

For in­stance, as cap­i­tal con­straints be­come more taut, we ex­pect to see more peo­ple perform­ing tasks which could be performed by ma­chines (i.e. not adopt­ing cap­i­tal-in­ten­sive tech­nol­ogy), and we ex­pect re­turns on cap­i­tal in­vest­ments to be higher (i.e. higher mar­ket price of cap­i­tal). When cap­i­tal con­straints be­come more slack, we ex­pect the op­po­site.

Step 3: Com­pare to the Real World

Com­pare real-world ob­ser­va­tions (mar­ket prices, tech­nol­ogy ac­tu­ally used) to our pre­dic­tions to figure out (qual­i­ta­tively) how taut/​slack the con­straint ac­tu­ally is, across many differ­ent in­dus­tries.

In the China/​Europe ex­am­ple, we can do this by look­ing at adop­tion of cap­i­tal-in­ten­sive tech­nolo­gies (i.e. tex­tile ma­chin­ery, wa­ter mills, au­toma­tion in gen­eral) or by look­ing at mar­ket prices of cap­i­tal (i.e. rate of re­turn on in­vest­ments).

Step 3.5: San­ity Check

If our hy­poth­e­sized con­straint is ac­tu­ally the right gear for this model, then we should see similar taut­ness/​slack­ness in differ­ent in­dus­tries—e.g. we shouldn’t see a rele­vant tech­nol­ogy adopted in one in­dus­try but not an­other, or differ­ent in­dus­tries pay­ing rad­i­cally differ­ent mar­ket prices to re­lax the con­straint.

In the China/​Europe ex­am­ple, we should check that there ac­tu­ally is a broad pat­tern of China us­ing la­bor on tasks where Europe used ma­chines. This doesn’t mean China didn’t use any ma­chines—our com­par­i­sons are rel­a­tive, we’re think­ing about more/​less at the mar­gin—but we should at least see the differ­ence across many differ­ent in­dus­tries. Similarly, we should check that cap­i­tal in­vest­ments had higher eco­nomic re­turns in China across what­ever in­dus­tries the Chi­nese in­vested in (though it may be that in­vestors had a tough time cap­tur­ing those re­turns… an­other po­ten­tial gear in our model).

This should give us some idea of how gen­er­ally-rele­vant our con­straint is, in prac­tice.

Step 4: Generalize

What other qual­i­ta­tive pre­dic­tions can we make, be­yond what we’ve already ob­served?

If we’re think­ing about new tech­nolo­gies or new busi­ness ideas, to what ex­tent do we ex­pect them to be en­abled/​blocked by slack­ness/​taut­ness of the con­straint? What other con­straints do we ex­pect to be­come more taut/​slack in re­sponse to this one? Is the con­straint be­com­ing more taut/​slack over time? If so, what changes do we ex­pect that trend to in­duce?

If you want a quick ex­er­cise in this, con­sider com­pu­ta­tional ca­pac­ity as the con­straint. You’ve prob­a­bly already heard plenty about how the taut­ness of that con­straint has changed over time, what new tech­nolo­gies have been adopted as a re­sult, and what other con­straints have be­come taut/​slack as a re­sult. Can you trans­late that in­for­ma­tion into the lan­guage of con­straints/​slack­ness/​prices? Does the con­straints/​slack­ness/​prices frame­work sug­gest other ques­tions to ask, or new pre­dic­tions to make?


We’ve already talked a bit about two re­sources to which this frame­work ap­plies: la­bor and cap­i­tal. One can build a whole wor­ld­view this way, and much of macroe­co­nomics does ex­actly that—see MRU’s videos on the Solow model for a friendly start­ing point to the macro mod­els.

There are many other very generic in­puts we could think about. The next two posts dis­cuss two other gen­eral con­straints:

  • Man­u­fac­tured goods. Across the board, real prices of most man­u­fac­tured goods have fallen dra­mat­i­cally over the past two cen­turies—sug­gest­ing those con­straints are rel­a­tively slack to­day. Yet that does not mean that we live in a world of to­tal ma­te­rial abun­dance. What would we ex­pect a world of slack ma­te­rial con­straints to look like? What other con­straints would be­come taut?

  • Co­or­di­na­tion. Pro­duc­ing and sel­l­ing goods or ser­vices re­quires co­or­di­nat­ing sales­peo­ple, en­g­ineers, de­sign­ers, mar­keters, in­vestors, cus­tomers, reg­u­la­tors, sup­pli­ers, ship­pers, etc, etc. Jobs which have the po­ten­tial for mak­ing very large amounts of money—e.g. en­trepreneur­ship, man­age­ment, in­vest­ment bank­ing, merg­ers & ac­qui­si­tions, etc—are largely char­ac­ter­ized by be­ing pri­mar­ily about co­or­di­na­tion. Similarly, if we look at the list of highly suc­cess­ful tech com­pa­nies over the past 25 years—Google, Face­book, Ama­zon, Uber/​Lyft… - the pri­mary busi­ness of most (though not all) of them is to solve some par­tic­u­lar co­or­di­na­tion prob­lem. Both of these sug­gest a very taut con­straint.

Some other ex­am­ples which I might ex­plore in fu­ture posts:

  • Trans­porta­tion. Be­fore the mod­ern era, pack an­i­mals’ food & wa­ter re­quire­ments sharply limited the range of over­land trans­port, es­pe­cially in arid re­gions. This con­straint shaped the paths of armies and lo­ca­tions of cities, sug­gest­ing a very taut con­straint. How taut was the trans­porta­tion con­straint his­tor­i­cally, and how taut is it to­day? What would the world look like with a to­tally slack trans­porta­tion con­straint?

  • In­ter­faces. If Alice wants to pro­duce some­thing for Bob, then first they must ac­cu­rately com­mu­ni­cate to Alice what Bob wants. Some in­ter­est­ing cases:

    • Alice is a com­pany, Bob is a customer

    • Alice is a com­puter, Bob is a pro­gram­mer/​product de­signer/​user

    • Alice is an em­ployee, Bob is an employer

    • Bob is Don Nor­man, Alice is a fridge. Yes, a fridge. We’ll get to that.

How taut/​slack are com­mu­ni­ca­tion/​trans­la­tion con­straints, in gen­eral?

  • The­ory and data, es­pe­cially in the sci­ences. In the past few decades, a num­ber of fields (e.g. biol­ogy, eco­nomics, and many in­dus­tries) have built up mas­sive piles of data. Yet the abil­ity to turn that data into use­ful knowl­edge and in­sights—i.e. the­ory—has lagged be­hind.

Also, there might be an­other post go­ing into more depth on cap­i­tal, es­pe­cially on the main places where cap­i­tal is de­ployed to­day.