Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
The Ants of Entrepreneurialism
As election season heats up, you’ll hear a lot about “investing” in technology, “encouraging” innovation through tax incentives, and praise for the government’s ability to improve research, development, and invention through some form of top-down direction or encouragement. You’ll hear it from Republicans and Democrats alike. Bad ideas. Why?
Consider the nature of invention. Where do breakthroughs come from? How do we maximize growth in research and development of new goods and services? The answer is illustrated by looking at the humble ant.
Like humans, ants live in social groups. Like humans in an ideal free market economy, ants are governed by a very few simple rules, but otherwise free to do as they please within those constraints.
Ant colonies have no central authorities at all, yet exhibit what looks like very intelligent cooperative behaviors that benefit the entire community. For example, if ants are blocked by a chasm, they will form a bridge with their own bodies, allowing other ants to crawl across. Ants wage war against other colonies, sometimes taking prisoners that they bring back to their own and enslave. Two colonies at war will establish fronts, engage in complex maneuvers, and you would swear that some smart ant general is guiding it all. Yet as far as we know, there is no guidance whatsoever.
If an anthill is flooded, ants have been known to swarm out and link together to form rafts that the colony can climb onto and float to safety. Ants use rotting vegetation and their own bodies to maintain precise temperatures in their nurseries. An alien coming to earth might well believe that an ant colony is a highly intelligent creature. But there’s no intelligence there at all! How does this happen? Ants don’t think. Their brains are not complex enough for any kind of real cognition. An individual ant is useless. But together they do amazing things with no central control whatsoever.
To make this happen, ants need a way of communicating with each other, and they need a simple set of rules to guide their behavior. They communicate through pheromones – chemical signals that ants emit and other ants detect. Ants emit many different types of pheromone, each triggering behaviors in other ants. Ant brains contain groupings of rules organized into “states,” with environmental triggers switching the ant from one state to another. We call this type of organization a state machine – each ant has multiple states it can be in at different times, and each state has its own set of rules and triggers for switching to other states.
The way ants find food is remarkably similar to the process of innovation in human society. There is a search space containing unknown amounts of food in unknown locations. What’s the most efficient way of finding things when there are unknown unknowns like that? Since you’ve got nothing to go on, the best thing to do is to spread out and search randomly: a stochastic search process.
Imagine you are an ant. You are in the state called “forager.” The rules are simple – walk away from the anthill in a random direction. If you run into another ant from your ant hill, move away from it. Keep doing this until you find food. Once you do, your state changes to “food carrier.” In that state, one rule is, “Pick up the food and head straight for the ant hill.” Another is, “If you are walking to the ant hill with food, emit a trail of pheromones.”
When other ants pick up the pheromone scent, they switch from the “forager” state to a state in which they head for the trail, then follow it away from the ant hill until they too find food. Then they follow the trail back to the ant hill, leaving their own pheromones behind. This strengthens the pheromone path, attracting more ants. So long as the food source is not exhausted, the pheromone trail grows more potent, attracting the distant ants. The trail grows in proportion to the amount of food discovered. Eventually, you’ve got a huge assembly line of ants shuffling food from the source to the ant hill. Meanwhile, other ants are discovering food as well, and their pheremone trails begin to compete. The end result is an anthill with organized lines of ants running out to each food source, with the number of ants in the line proportional to the amount of food available.
When the food supplies run out, there’s another rule. “If you get to the end of the trail and don’t find food, change state back to ‘forager’ and mill about in that area.” The ants begin to fan out in the region were the food was originally found; after all, it is the place most likely to have more food. But if that’s the last food in the area, the trail eventually dries up, and the ants return to general foraging, or join other trails to food after they run across the scent.
It’s all pretty cool – and highly efficient. But the individual ants don’t even know they are doing it. This entire ballet is an emergent property of a complex system governed by simple rules. This is what it looks like, simulated:
The blue dot is the food source. The purple line is the pheremone trail. Notice how efficient those ants are at finding and exploiting the unknown food source. If you don’t trust the simulation, there are plenty of videos on YouTube of real ants doing the same thing. I just chose this video because it’s easier to see what’s going on.
Now we get back to humans. We have a very similar problem – there is a search space for innovation, discovery, and invention. Capitalism works so well for innovation because we have developed the same process. Humans do not have instinctive rules and states in the sense that the ants do, but we have rationality and free will. In a free market, money acts as both the incentive and the pheromone trail, which gives us the information and incentives we need to use our brains in a coordinated way. We’re searching intellectual spaces stochastically, like ants.
People have an incentive to make money. This causes them to search for new goods and services. In a free market, no one directs that search. Everyone may intellectually mill about. When someone hits paydirt – by discovering a new way to fill an economic need – money starts to flow in that direction. Other entrepreneurs see the money trail, and try to find more value using the information uncovered by the first innovation. New markets are created.
If the new market or food source is large (say, semiconductors), huge amounts of capital will eventually flow in that direction, maximizing our use of the new resource in direct proportion to its value compared to everything else in which we could be investing our money. This is like the multiple pheromone trails around an ant colony. This is how progress happens.
Some people don’t follow that path, or the path gets saturated and new food becomes harder to find. People aren’t ants, so everyone has a different perspective, different knowledge sets, different abilities and experiences. That’s what you want when you are exploring a unknown space – economic and mental diversity. You want lots of different people exploring different avenues. You want parallel searches.
Eventually, the money flow to one area stabilizes when the market matures or shrinks. The utility of the discovery diminishes, and people start looking elsewhere in the search space of unknown unknowns. This approximates an optimal strategy for discovery when you have no information telling you where the good stuff is. Trying to direct this research does nothing but add noise and decrease efficiency. Throwing down false pheremone trails of government money crowds out other smaller trails that may actually lead somewhere. Government subsidies of R&D tend to cause consolidation by a few large companies, reducing diversity, crowding out other entrepreneurs, and leaving more of the search space unexplored.
Capitalism and the free market provide the basic rules for economic interaction. Prices provide the mechanism for transmitting information across society and allow us to coordinate our activities without coercion or control. They allow our individual intelligences to share information and magnify their power. Money is the information bus of society’s massively parallel computing system. This is how people work together best in a complex system, and how capitalism and prices ensure that as we seek our own benefit, we must benefit others. Adam Smith called this ‘The Invisible Hand.”
Ants have an invisible hand guiding them. So do we. Nature hasn’t found anything better after billions of years of evolution; nor have human societies found anything better after millennia of experiments. If central control were more efficient, you’d see a lot of special “controlling” animals with bigger brains and better senses guiding others, but you don’t. Nature has discovered the wisdom of the crowd – it’s about time our politicians did too.
Published in General
And here I thought I was the only person who had ever read Gödel Escher Bach….
Ants are marvellous machines. And then you consider that the ant colony itself becomes an input into an even more complex ecosystem around it…
When people today talk about AI taking over, they seem to think it’s about building faster, smarter processors. But the complexity comes from the interconnections between all the little computing units. Google’s Face Recognition AI Neural Network is astounding.
So while we’re worrying about CPU power, I’m wondering what’s about to happen to our society, our economy, and our security as the number of interconnections between us grows exponentially. Not just us in terms of social graphs and digital footprints and such, but all the machines, the routers, the financial systems, The ‘internet of things’ stuff, etc. Each node in our giant complex system is getting smarter, and the interconnections are growing exponentially.
At some point are we going to lose control of that system? Are emergent patterns going to start playing havoc with our attempts to use the system?
I think we’re already seeing some of that in the heavily connected and high speed financial system. I still don’t think we know why the last ‘flash crash’ happened.
At some point, does it become complex enough to be considered an AI? If so, it’s so alien to what we thought of as ‘intelligence’ that we may never really understand it.
But surely an anthill is more than the sum of its parts.
Wow, this is fascinating.
So government intervention for ants would be like taking some of the food that has been brought back to the colony and dropping it off in another place where the ant planners think there is the best chance for more food. The ant’s food finding planners would then pheromone the new officially approved path so that other ants would eagerly follow it and if that path doesn’t have a good ROI then the planners would just bless a different path.
At some point this colony would start to decline and a neighboring colony would swoop in and take over the famished ants’ colony making them their slaves.
This could make for an interesting story. I’d call it “Ant Farm”.
The Bible tells us to go to the ant and consider her ways. Now more than ever!
Dan Hanson’s wonderful piece reminds one of the elegant simplicity and clarity of Leonard Read’s 1958 essay, “I, Pencil.”
And Dan suggests a more fundamental, entrepreneurial approach than we generally hear today to solving perceived ills in the economy, such as “income inequality” and “creating jobs.” Oh, how that latter term has become a snare and a delusion, suggesting that some omniscient politician can turn on a 3-D printer from which pours a torrent of jobs.
We are fortunate to have Ricochet’s platform bringing together disparate people from all over the globe to leave their trails of pheromones and increase the world’s knowledge, its propagation and, yes, its complexity. Thank you, Dan, for doing your part.
Great article. Is there a spreading grasp of the implications of complexity and emergent systems, or is it just that having discovered Hayek late in life, I’m seeing what has been there all along? The thing is it obliterates Keynesian economics, science policy, government global warming amelioration, top down social policy not to mention tax expenditures and centralized regulation as useful tools for advancing human interests.
Wonderfully said.
Great article. Great food for thought (no pun intended).
Thank you! ‘I, Pencil’ was certainly on my mind when I thought about some of this, for sure. But my biggest influence has been the work of Hayek, Claude Shannon, and lately the whole field of complexity science that has been arising over the past couple of decades.
This is a tough subject to popularize, but it is a gold mine of ammunition for the right, because it accurately describes how the world works. Once you get your head around it, it changes the way you look at everything.
That’s a great analogy. Here’s another:
Ant minimum wage discussion: “It’s not fair that some ants only find little tiny bits of food, while others get lots of food for their effort. So we’re going to modify the ants so they can’t detect those tiny pieces of food, for ‘fairness’.”
The problem is that then you’ve got ants who otherwise would have found some of the food finding nothing at all. So then the ant government tries to stop ‘food inequality’ by taking some of the large food sources, breaking them apart and adding bits to the smaller food bits to bring them above the new minimum threshold for ‘good food’. Now there would be pheromone trails all over the place, and pheromone signalling would start to lose its value. Ant GDP declines ‘unexpectedly’. Ant Krugman says the problem is that we just didn’t sprinkle enough food…
Keynesianism: “We noticed that since that one food source ran out, we have way too many ants just milling about. So they aren’t laying pheromone trails, which causes even more ants to mill about. It’s a downward spiral! Here’s what we’ll do – we’ll fly over the region and spray pheromones all over the place. The ants will find them and follow them, and they’ll add their own pheromones and other ants will pick it up, and soon they’ll all be working again! It’s a pheromone multiplier!
The problem with that is that ‘milling about’ was exactly what the ants were supposed to do when they ran out of food, and that ‘milling about’ is what speeds discovery of the next food source. But now the ants can’t tell which pheromone trails are ‘real’, and which ones were ‘stimulus’ pheromones. You get lots of pointless activity, and ants that would have discovered food had they been left to find it are now wasting their time following trails to nowhere.
Analogy is suspect, and people aren’t ants. But complex systems have properties that exist whether the system contains ants, people, or robots.
Hayek was one of the early thinkers in complex systems, and I’ve been surprised by how many times his name has come up in complexity literature having nothing to with economics or even people for that matter.
Complexity theory really has taken off in the computer era, however. Computers have not only given us the tools to simulate and model complex systems to explore their properties (like the ant model above), but I would argue that software was one one of the first fields to really run into massive issues with complexity and this spawned a lot of research in the subject.
One of the sad ironies of economic history is that Hayek was dismissed as being ‘old fashioned’ by the new breed of ‘scientific’ economists. In fact, Hayek’s ideas were the future – it’s just that at the time we didn’t have the computational power or understanding to think about complex systems properly. The other economists were practising cargo-cult economics while sneering at Hayek.
The old equilibrium models of economics were never right. They are an approximation borne out of the scientism of the 19th century when physics and engineering were gathering momentum and social scientists wanted to be ‘scientific’ too. They were also created so that economists could use the tools of that era – basically the same kinds of formulas that physicists were using to describe machines. The first equilibrium models stole material and formulas straight out of a physics textbook of the time. And ever since we’ve been talking about economies as machines that have ‘equilibria’ – forces that push things back to a status quo after a shock happens as if everything were connected with springs.
We know now that’s not correct. An economy is more like a self-programming computer – every input ripples through the thing, and it mutates and changes its behaviour accordingly. You can have local meta-stability in some areas that might look like equilibrium, until suddenly it doesn’t.
And since the economy and systems like it constantly mutate and change, ‘past performance is not predictive of future value’. The whole practice of predicting what the economy will do by studying what it did in the past is suspect, because you’re always dealing with a new system. The same goes for trying to predict the climate system.
The Santa Fe Institute is the leading hub of research into complexity. If you’re interested in the topic I would highly recommend “Complexity: A Guided Tour” by Melanie Mitchell. It’s a great overview of the subject.
I’d also like to point out that the model shown in the video here is a good example of how models can be useful. We had a theory of how ant colonies behave. So we build a simulation, give it the rules that we think the ant colony uses, and we can watch the output and see if it matches the kinds of behaviour we see with the ant hill. Then we can start tweaking it to see if we can discover other interesting properties of such systems.
But even though we have a near-perfect model of ant foraging behaviour, it is totally useless at predicting the actual future shape and size of the colony! Its only predictive value is in generalities – being able to predict, for example, that when a food source is exhausted we should see a temporary increase of ants milling about aimlessly.
But if you try to predict the size of the colony or the direction it will expand, you will utterly fail. For another property of these systems is that they are incredibly sensitive to tiny changes in initial conditions.
Imagine two large food sources, located in opposite directions from the ant hill. Two ants are heading for those sources, but one of them is slowed down by a twig, or jostled off course by a leaf moving in the breeze. Now the other ant wins the ‘race’, and the colony expands in that direction. Eventually that may lead to more food discoveries over there, and the whole size and shape of the colony winds up completely different, just because the other ant ran into a random minor obstacle – the ant equivalent of the butterfly effect.
Being able to model and understand a complex system does not translate into the ability to predict what it will be or do in the future. That’s the fundamental mistake people who try to model the future of such systems are making.
Ant Krugman is the worst!
Heh.
Oh, another good book that touches on these subjects is George Gilder’s
Knowledge and Power: The Information Theory of Capitalism and How it is Revolutionizing our World
The book has one blind spot in that Gilder seems to deny that such a thing as spontaneous organization or emergence can happen. I believe this is the result of his creationism beliefs. But other than that, the book is full of great ideas and explanations of how things really are.
Dan
Thanks, I’ll take that as yes that it’s now widely understood. Regarding Gilder, I didn’t read the book, just heard his discussion on econ talk or uncommon knowledge. Was he denying spontaneous order or just addressing the circumstances that provided the rules that gave rise to our particular order? Perhaps I projected Hayek on him, i.e. our founders avoided the fatal conceit by accepting received religion as well as insights extracted from history and political philosophers, and that made a great deal of difference to how our order emerged.
I’m pretty sure Gilder denies the concept itself, at least with respect to human society.
This is what he says in the book:
He’s just wrong about this. He seems to think that anything ordered must have been planned. He actually says that Adam Smith was wrong about the invisible hand.
I believe this misconception must arise from his religious beliefs. If you believe that everything in nature is part of God’s plan, that evolution is wrong, and every force in human society is planned by someone, then I guess you don’t need spontaneous order.
So I guess his answer to the myriad complexities of ant colony behaviour is that it’s not spontaneous order, but that it’s part of God’s plan or something.
His book is mostly about information theory, which is also a fantastically powerful concept in defence of free markets. So maybe he’s just not very educated in complex systems theory.
Fantastic article. And thanks for the book recommendations in the comments.
Since ‘I, Pencil’ has been mentioned, this seems like a good place to mention ‘I, Whiskey’. The producers of the movie version of ‘I, Pencil’ have a new project in mind to make a documentary tracing the threads of the whiskey market. They are running a crowdfunding campaign on Indiegogo. Tonight is the funding deadline; there’s still a little bit of time to support them and get a sweet backer perk such as a bottle signed by Carly Fiorina.
(I’m not involved in this crowdfunding campaign, just thought some people here would be interested).
I’m a sucker for a good free-market effort, and I can always use another t-shirt for workouts. Thank you, TD, for the heads-up!
I like how your mind works!
(aside to anonymous)- fourmi-dable!