Do you think that the distinction between growth at the frontier and catch up growth is always so well defined? So if you think about China, they seem to have innovations: how quickly they can build things; they have an autocratic government, but they’ve managed to keep reasonable stability and the public on board. Isn’t that kind of innovation like a technological innovation and their growth in a way is at some other frontier rather than being just catch up?
That was a question from Tyler Cowen to Paul Romer. I think this point needs to be better appreciated, particularly in defense policy where we do not have the benefit of the market-test to really know whether and how far in advance we are of Russia and China. That can only be known in war. If they have a culture of rapid experimentation and failure, then I can imagine China is pushing ahead of the US on many defense technologies where such freedom is highly restricted.
This relates to the influence of culture and institutions on technology. Here is Romer:
To have a rich enough theory of development, to explain the variety of outcomes we see in different countries, and to offer useful advice in different contexts, we need to understand a lot more about culture.
If I may simplify without too much harm, what Romer goes on to say is that you cannot just force proper institutions like law, property rights, and contract onto non-compatible cultures. To me, it seems like a chicken-or-the-egg problem with culture and institutions as to preconditions for success. But when trying to reform complex institutions, Romer seems to say that we need to select the individuals who are most compatible with the culture that supports the reformed, liberal, institutions. Then, over time, they can absorb more and more of the remaining population without destroying the culture that underpins the institutions.
I think this is an important point if you wanted to reform defense acquisition. I often say that the most important step in reform is to move from a program to an organizational budget. But that may pre-suppose something about culture that doesn’t really exist. Devolving more authority requires trust, and the public might not have financial trust in the DOD.
Perhaps the best way to go about reform is that you give select tech labs, PEOs and Combatant Commands their own procurement budgets. After all, SOCOM and then CyberCOM got their own procurement budgets. We can slowly expand the practice and learn lessons on the way. These organizations should have a strong and unique cultures and are prized to work at, but only select those with compatible attitudes. Slowly, the culture and institutions can grow together.
After all, when the program budget was installed in the DOD in 1949, agencies had to start preparing a double set of books for many years until they completely switched over to the program budget. Today in procurement, we have avoided many of the contradictions in program budgeting because we have perfectly aligned programs with organizations (i.e., the program office and PEO).
So perhaps we swung too far in the top-down direction since Robert McNamara in 1961, but that doesn’t mean we need to swing all the way back to bottom-up. Paul Romer reminds us that:
You need in a city both some design, which you can think of like top-down, and some market-like or decentralized forces.
Yes, striking that balance I think is the challenge of the post-industrial world, and has been a particularly salient issue in defense acquisition for 70 years. Romer elaborates on some of the ideas that will be important in the internet age:
If you think about the theory of teams, we don’t have an analysis where we have these agents with more complicated motivations and you think about teamwork…
How would you make progress. I wouldn’t sit down and make assumptions — like here are the axioms of management — what I would do is go out and look at the evidence. I think some of the most interesting evidence about management comes out of Toyota who applied principles from Edward Demming, but there’s one mechanism called kanbon at Toyota that used to manage things on the production line, that has now been taken up by people in software development, and it has I think some very interesting principles.
One is that transparency is very helpful in teamwork, because when people don’t know what others are doing, they tend to have a bias toward the negative. They think something bad is happening. So transparency is actually a good helpful way to build trust. So what kanbon does is trying to make clear what everyone else is doing.
That’s interesting. When we have small “two pizza” teams, you basically know what everyone else is up to and can hold them personally accountable. But for larger endeavors, you need some way to maintain that trust.
The transparency point reminded me of stories from East Germany. The Stasi, a state security organization, had to be asked permission for many privileges, such as travelling. People would send requests into this huge bureaucratic process and decisions were somehow reached. Eastern Germans had no idea how decisions were reached, and why certain people were or were not given permission. It created paranoia of all manner about who did what to get approval, when in reality, it was largely random and poorly administered. This is the kind of opacity that not only destroys trust, but breeds mistrust.
Here is Tyler asking whether Paul believes in general purpose technologies:
This brings up a deep philosophical concept. Do I believe in abstractions of any form? I think the general purpose technology has been useful to distinguish a kind of innovation that you can think of as being aligned along a continuum. Some have long paths of exploration that induce a lot of complementary innovation. At the other end of the spectrum could be one-offs, like the paper clip. We’re done with that.
The abstraction takes a continuum, or very high dimensional set of possible innovations, and then defines a category as a discrete entity, and says these have a particular character worth studying. I think this has been a useful abstraction for thinking about things like electrification and so i imagine it will be helpful contemplate things as we go forward. But the question there is if it is a general purpose technology, will be know it as it’s unfolding or will we only see that when we look back.
This is similar to the argument I was trying to make when I talked about uncertainty of objectives and hindsight bias in trying to measure technological change. It is, in part, why I am skeptical of claims that technology has stagnated even if I am sympathetic to them. Perhaps we have just built in our minds all these categories of technologies, and we just have much less information about the frontier, which has been pushed out so far in so many directions, but provides many options for combinatorial innovation.
Here’s Romer on being random:
I’ve always been more inclined to take risks, or to sample a lot more ideas. I make fun of myself and say “I’m just a random idea generator” and then what’s needed is others to filter out the bad ones and then on average I can be helpful.
A random idea generator is actually pretty important because there is no definite procedure for creating randomness. The first random number generator was based on the radio-active decay. But just as important is not so much the randomness, but that by absorbing a lot of ideas, the “random” conjectures are actually quite likely to be on an innovative path so long as you are willing to learn along the way. I think we are leaving the time when “smart” third-parties can simply say what conjectures are likely to yield success or not. They need to be empirically tested.
Romer also talks about finding alternatives to “positivist machine” ideas about technological progress by reading history, an approach I highly approve of:
One of the joys of reading, and to me a slightly frightening thing, is that there is so much out there. A hundred years later you can discover somebody that has so many things to say that can be useful to someone like me.
Romer won the 2018 Nobel memorial prize in economics for endogenous technological growth.
Leave a Reply