Ecologists understand the exploit/explore distinction. When an animal looks for food, it can either exploit (use previous knowledge of where food is) or explore (try to learn more about where food is). With ants, the difference is visible. Trail of ants to a food source: exploit. Solitary wandering ant: explore. With other animals, the difference is more subtle. You might think that when a rat presses a bar for food, that is pure exploitation. However, my colleagues and I found that when expectation of food was lower, there was more variation — more exploration — in how the rat pressed the bar. In a wide range of domains (genetics, business), less expectation of reward leads to more exploration. In business, this is a common observation. For example, yesterday I read an article about the Washington Post that said its leaders failed to explore enough because they had a false sense of security provide by their Kaplan branch. “Thanks to Kaplan, the Post Company felt less pressure to make hard strategic choices—and less pressure to venture in new directions,” wrote Sarah Ellison.
Striking the right balance between exploitation and exploration is crucial. If an animal exploits too much, it will starve when its supply of food runs out. If it explores too much, it will starve right away. Every instance of collapse in Jared Diamond’s Collapse: How Socieities Choose to Fail or Succeed was plausibly due to too much exploitation, too little exploration (which Diamond, even though he is a biologist, fails to say). I’ve posted several times about my discovery that treadmill walking made studying Chinese more pleasant. I believe walking creates a thirst for dry knowledge. My evolutionary explanation is that this pushed prehistoric humans to explore more.
I have never heard an economist make this point: the need for proper balance between exploit and explore. It is relevant in a large fraction of discussions about how to spend money. For example, yesterday I listened to the latest EconTalk podcast, a debate between Bob Frank and Russ Roberts about whether it would be a good idea for the American government to spend $2 trillion on infrastructure projects (fix bridges, etc.). Frank said it would create jobs, and so on — the usual argument. Roberts said if fixing bridges was such a good idea, why hadn’t this choice already been made? Roberts could have said, but didn’t, that massive government shovel-ready expenditures, such as $2 trillion spent on infrastructure repair, inevitably push the exploit/explore balance toward exploit, which is dangerous. This is an argument against all Keynesian stimulus-type spending. I have heard countless arguments about such spending. I have never heard it made. If you want examples of how the American economy suffers from a profound lack of useful new ideas, look at health care. As far as I know, there are no recorded instances of a society dying because of too much exploration. The problem is always too much exploitation. People at the top — with a tiny number of exceptions, such as the Basques — overestimate the stability of their position. At the end of The Economy of Cities, Jane Jacobs says that if a spaceship landed on Earth, she would want to know how their civilization avoided overexploitation. When societies exploit too much and explore too little, said Jacobs, problems (in our society, problems such as obesity, autism, autoimmune disease, etc.) stack up unsolved. Today is China’s birthday. Due to overexploitation, I believe China is in even worse economic shape than America. Ron Unz, whom I respect, misses this.
My broad point is that a lot of economic thinking, especially about growth and development, is one-dimensional (measuring primarily growth of previously existing goods and services — exploitation) when it should be two-dimensional (measuring both (a) growth of existing stuff and (b) creation of new goods and services). Exploration (successful exploration) is inevitably tiny compared to exploitation, but it is crucial there be enough of it. If there is a textbook that makes this point, I haven’t seen it. An example of getting it right is Hugh Sinclair’s excellent new book Confessions of a Microfinance Heretic (copy sent me by publisher) that debunks microcredit. Leaving aside the very high interest rates, the use of microcredit loans to buy TVs, and so on, microcredit is still a bad idea because the money is, at best, used for a business that copies an existing business. (The higher the interest rate, the less risk a loan recipient dares take.) When a new business copies an already-existing business, you are taking an existing pie (e.g., demand for milk, if the loan has been used to buy a cow and sell its milk) and dividing it into one more piece. The pie does not get bigger. As Sinclair says, the notion that dividing existing pies into more pieces will “create a poverty-free world” is, uh, not worthy of a Nobel Prize.
Sure, it’s hard to measure growth of useful knowledge. (It is perfectly possible for a company to waste its entire R&D budget.) However, I am quite sure that realism does better than make-believe — and the notion that growth of GDP is a satisfactory metric of economic growth is make-believe. If you’ve ever been sick, or gone to college, and have a sense of history, you will have noticed the profound stagnation in two unavoidable sectors (health care and education) of our economy. That are growing really fast.
To the extent that microcredit increases individual security, it increases the likelihood of exploration. Also, even a conventional business has room for some exploration, even if most don’t use it.
Seth: By its nature, taking on debt decreases security, if you must pay back the debt. And taking on debt at high interest — microcredit loans tend to have interest rates on the order of 70%/year — decreases security even more. Less security, less risk-taking, as you say.
I like it.
You probably already know this Seth, but the Austrian economists have been stressing the market as discovery procedure for a long time, particularly Hayek’s “Competition as a Discovery Process” and Israel Kirzner’s work on entrepreneurial discovery. Perhaps there is an Austrian textbook that covers exploration/exploitation, though I’ve not encountered it in exactly this form that I can recall.
Just want to note, the exploration/exploitation continuum roughly maps to the The Innovation Spectrum spanning from Sustaining to Disruptive.
Image here: https://vlskvts.co/Ne04U1
>>If you’ve ever been sick, or gone to college, and have a sense of history, you will have noticed the profound stagnation in two unavoidable sectors (health care and education) of our economy. That are growing really fast.
Case in point: https://www.theatlantic.com/entertainment/archive/2012/10/why-there-needs-to-be-a-real-grad-school-of-rock/262802/
Have you read FOFOA? He has a very interesting theory of the current crisis.
He thinks that we will not move into Gold Standard after this Fiat crisis. We will instead move to something like Euro. Where Gold is the major reserve but it does not back the currency directly. Also all Paper claims to gold will disappear. The currency in gold standard is also a paper claim to gold. Which depresses the price of gold.
It makes so much sense.
FOFOA’s dilemma: When a single medium is used as both store of value and medium of exchange it leads to a conflict between debtors and savers. FOFOA’s dilemma holds true for both gold and fiat, the solution being Freegold, which incidentally also resolves Triffin’s dilemma.
https://fofoa.blogspot.in/2011/05/return-to-honest-money.html
Last place I thought I’d find gold-buggery!
My view is we’ve run up against resource limitations, so tilting the balance toward exploration is vital. Even so I reckon we’ve seen the end of GDP growth – peak exploitation and peak-debt are already behind us, the age of making money off money is over.
Perhaps we’re moving into a steady state economy – not a disaster, but certainly far removed from the expectation that it’s just a matter of time before we get back “on course”.
As usual, the hard-science bloggers provide the best insight on areas outside their disciplines because they can apply maths and stats – try this one for the steady-state hypothesis:
https://liminalhack.wordpress.com/
And anyone interested in the Chinese debt disaster has to follow Michael Pettis:
https://www.mpettis.com/
Interesting post. I’m confused about one part, though: why do you think $2 trillion of additional infrastructure spending would effect the explore/exploit balance? I don’t believe any federal research budgets were being cut to fund the $2 trillion, so explore would stay the same; almost all of the resources would either come from re-allocating things already in the “exploit sector” (such as private sector construction workers) or from unused resources (such as unemployed workers).
In computer science, we often think about the explore/exploit distinction in the context of the “multi-armed bandit” model and we try to prove theorems like such-and-such algorithm does this good of a job of managing the trade-off in such-and-such situations. It isn’t just useless theorizing, though: at Google (where I work), we use multi-armed bandits to optimize the ads and websites we show to our users. (See https://support.google.com/analytics/bin/answer.py?hl=en&answer=2677320 for example).
Seth: Why would $2 trillion of infrastructure spending affect the exploit/explore balance? Because it would all go toward exploit, creating a large number of people (everyone who gets the money) whose interests lie in more exploit, less explore. Unemployed workers drift between explore and exploit — they will do what they are hired to do. Once they are hired, they start to become committed. An unemployed worker hired to exploit will want more exploit; an unemployed worker hired to explore (after working in exploit) will begin to see why the country needs more explore.
Seth,
That’s a great explanation – in response to Thomas.
I have seen in my community the results of “government stimulus” – the workers hired to exploit do indeed want more, and see that as the best way forward. Even worse, the community starts to buy into it as it seemed to work better than explore – in the short term anyway.
I think of it as the economic equivalent of sugar/empty calories. It has a short term benefit of increased activity and usually little long term gain.
Explore has an inherent risk component and no one – governments, corporations, people seems to want to take that on. Even universities and research institutions suffer this, as their funding mechanisms often seem to favour exploit.
Can you give an example of someone hired to “explore”?
Seth: Thanks! To answer your question, someone hired by an innovative small business. When something’s new, there’s a lot to learn about how to make it, sell it, improve it.