Two Kinds of Technology Change

Many of the ex­am­ples in this post are drawn from Vol­ume I of Fer­nand Braudel’s “Civ­i­liza­tion and Cap­i­tal­ism: 15th-18th Cen­tury”, which I strongly recom­mend for any­one in­ter­ested in a quan­ti­ta­tive ap­proach to his­tory.

You may have heard that Guten­berg rev­olu­tionized the in­tel­lec­tual world with his in­ven­tion of mov­ing type in the mid-1400’s.

Here’s the thing, though: Guten­berg was not the first to try mov­able type. The Chi­nese were us­ing ba­sic print­ing presses in the ninth cen­tury; Pi Cheng in­tro­duced mov­able char­ac­ters be­tween 1040 and 1050. So why didn’t it catch on then? And even set­ting that aside, surely some tired monk must have thought of the idea sooner.

Turns out, prior to the 14th cen­tury, books were pri­mar­ily printed on parch­ment — cre­ated from sheep skins. A sin­gle 150-page book re­quired the skins of 12 sheep to make the parch­ment. That much parch­ment wasn’t cheap — the parch­ment on which a book was writ­ten cost far more than the ac­tual writ­ing. With that much cost sunk in the ma­te­ri­als, it’s no won­der that book-buy­ers wanted beau­tiful, hand­writ­ten script — it added rel­a­tively lit­tle to the cost.

It was pa­per which changed all that. Euro­pean pa­per pro­duc­tion didn’t get prop­erly un­der­way un­til the 1300’s. Once it did, book prices plum­meted, writ­ing be­came the pri­mary ex­pense of book pro­duc­tion, and print­ing presses with mov­able type fol­lowed a cen­tury later.

The print­ing press offers a clear ex­am­ple of a tech­nol­ogy change whose ar­rival was limited, not by the ge­nius of the in­ven­tor, but by eco­nomic vi­a­bil­ity. The limit­ing fac­tor wasn’t in­sight, it was prices.

Once you go look­ing for it, there’s a lot of tech­nol­ogy shifts like this. New­comen’s steam en­g­ine (and Heron’s, long be­fore). Sch­wen­teer’s tele­graph. Bush­nell’s sub­marine. Bab­bage and Lovelace had all the key ideas for the mod­ern com­puter in the 1820’s, but it wasn’t un­til the 1890 cen­sus that some­body wanted to pay for such a thing. And of course, Moore’s Law led to all sorts of ideas go­ing from un­prof­itable to ubiquitous in the span of a decade or two.

In all these cases, the pat­tern is the same: the idea for an in­ven­tion long pre­dates the price shifts which make it mar­ketable.

On the other hand, this isn’t the case for all tech­nolog­i­cal progress. There are some tech­nolo­gies for which de­mand pre­ceded ca­pa­bil­ity. After some in­sight or break­through made the tech­nol­ogy pos­si­ble, adop­tion fol­lowed rapidly. Con­sider the Wright broth­ers’ flyer, or Edi­son’s light­bulb. Both had badly in­fe­rior pre­de­ces­sors, which didn’t re­ally solve the prob­lem: gliders and hot-air bal­loons for the Wright broth­ers, arc lights for Edi­son. Both built fast iter­a­tion plat­forms, tested a large pos­si­bil­ity space, and even­tu­ally found a de­sign which worked. And both saw rapid adop­tion once the in­ven­tion was made.

One no­table fea­ture of these break­through-type tech­nolo­gies: the eco­nomic in­cen­tive for flight or the light­bulb was in place long be­fore the in­ven­tion, so of course many peo­ple tried to solve the prob­lems. Both Edi­son and the Wright broth­ers were pre­ceded by many oth­ers who tried and failed.

Here’s a sim­ple model: tech­nol­ogy de­ter­mines the limits of what’s pos­si­ble, the con­straints on eco­nomic ac­tivity. We can think of these con­straints as planes in some high-di­men­sional space of eco­nomic pro­duc­tion. Eco­nomic in­cen­tives push us as far as we can go along some di­rec­tion, un­til we run in to one of these con­straints — and the tech­nol­ogy we use de­pends on what con­straint we hit.

Fol­low­ing the in­cen­tive gra­di­ent in the di­a­gram above, we end up at the smiley face — us­ing a mix of tech­nolo­gies A and B. This point is in­sen­si­tive to small changes in the in­cen­tive gra­di­ent — the prices can shift a bit one way or the other, shift­ing the in­cen­tive gra­di­ent slightly, and the smiley-face point will still be op­ti­mal.

How­ever, if prices shift enough, then we can see a sud­den change.

Once the in­cen­tive gra­di­ent moves “down” suffi­ciently, we sud­denly jump from the A-B in­ter­sec­tion be­ing op­ti­mal to the B-C in­ter­sec­tion be­ing op­ti­mal. A new set of con­straints kicks in; we switch from tech­nol­ogy A to tech­nol­ogy C. That’s the print­ing press: in­vent­ing C doesn’t mat­ter un­til the prices shift.

On the other hand, we can also change tech­nolo­gies by re­lax­ing a con­straint. Sup­pose some new-and-im­proved ver­sion of tech­nol­ogy A comes along:

Tech­nol­ogy A’ al­lows us to ig­nore the old A con­straint, and move fur­ther along that di­rec­tion. If we were us­ing A be­fore, then we’ll definitely want to switch to A’ right away. That’s Edi­son’s light­bulb.

In or­der for a tech­nol­ogy to go from not-used to used, one of these two situ­a­tions must hold: ei­ther the tech­nol­ogy was un­prof­itable be­fore and a price shift makes it prof­itable, or else it was prof­itable be­fore, and many peo­ple tried to figure it out but couldn’t. If you your­self want to mar­ket some kind of tech­nol­ogy, then you should con­sider which of these two situ­a­tions ap­plies. Has a re­cent price shift made it prof­itable? Have you made some sort of break­through which oth­ers have tried and failed to find? If the an­swer to both of those ques­tions is no, then the tech­nol­ogy will prob­a­bly re­main un­used. If the an­swer to at least one of those ques­tions is yes, then you may be on to some­thing.