30 June, 2014

Development in pairs

A hot topic these days is about "economic development" and its associated strengths and ills.

Unfortunately, much of these debates on social media or mass media degenerates into mudslinging between opposing camps, and at the end of it, an esoteric entity called people's "attitude" is blamed for all our ills.

From the way I see the debates going, we have almost zero understanding of an important element of any kind of economic or social change -- that of "non-linearity." Non-linearity is used in systems theory to indicate phenomena of positive feedback, where the effect of some cause in turn affects the cause itself.

For instance, a large city is likely to have more job opportunities than a small town, which in turn attracts more migrant population to the large city over the small town. Phenomena like rich getting richer, 80-20 rule and such, are all the outcomes of underlying non-linear processes.

Non-linearity is the reason why many aspects of economic and social phenomena are counter-intuitive. If we think linearly, we tend to approach problems with immediate, symptomatic solutions and often end up making the problem worse.

For instance, suppose the problem we are addressing is that of managing depleting oil supply. One common way of approaching this problem is to build more fuel-efficient vehicles so that they burn less fuel for the same usage. But what likely happens is that, now that vehicles give more mileage, people have a rational incentive to buy and use more such vehicles, thus aggravating the fuel shortage.

I am not saying we should not design fuel efficient cars. Not the point at all. The point is about strategizing in "pairs" which I'll come to in a moment.

Another example is the Cobra Effect story from the colonial days, when the ruling British government, who were afraid of cobras, offered a monetary incentive for people to kill cobras. This incentive, even though initially successful, had the opposite effect overall. Sensing a way of making money, people started breeding cobras instead of hunting them. And when the government sensed this and stopped the incentive, people who were breeding cobras released them into the open, thus making the original problem worse!

The thing with social and economic systems is that they are not inanimate physical systems with static characteristics. They are thinking, scheming, rational entities that responds to your input with a "best response" function that maximizes its own benefit, which need not be what we expected as the outcome.

There are two elements to non-linear systems: growth and saturation. Growth is typically visible, while the dynamics of saturation is much harder to measure and understand.

Growth happens when the system responds positively to our inputs resulting in a "honeymoon" phase. In the cobra effect example, when the monetary incentive was introduced, the people responded to it positively, hunting down cobras and depositing them. The positive response in turn gave an incentive to the government to respond promptly with their reward and to spread more awareness of this program. And hence started the initial "growth" phase of this engagement.

But a positively reinforced growth soon starts depleting resources (in this case, the cobras), that is when the strange effects of saturation sets in. Saturation happens when resources deplete globally and the system is unprepared to handle this depletion. And it is extremely hard to predict how a system will end up responding to a state of saturation. In this case, the system resorted to artificially sustaining the growth, because resources (cobras) could be artificially replenished.

All other "breakdown" phenomena like riots, looting, hoarding, etc. can be seen as a form of saturation dynamics. Something has saturated -- some critical resource has depleted and the system is unprepared to handle this, resulting in large-scale breakdown.


One way to manage saturation dynamics is to approach developmental strategy in "pairs" with two positive feedback loops posing as an alternative to one another.
Consider the above figure where two mobile service providers are competing for market share. Market share dynamics are replete with non-linearity. A service provider with a high market share can afford to spend more on advertising and can capitalize on "network effects" by users attracting other users. This results in a positive feedback loop.

However, at some time, the growth starts saturating. The number of users and the amount of use would have gone up so high that the infrastructure starts creaking.

At such times, nothing is more attractive than having an alternative.

So A and B above are competing over the same resource pool (users). Say A wins the game and gets into a positive feedback loop. It is a matter of time before resources saturate in A. At which time B is rationally attractive to users resulting in a migration exodus to B. Soon B may start saturating and a reverse migration begins. This back and forth eventually settles down to an equilibrium.

In order to tackle saturation this way, two things are necessary. First, A and B should be sufficiently distinct in order to pose as an alternative to one another. And second, the cost of shifting between A and B should not be so high that it would make rational sense to suffer the effects of saturation, than look for alternatives.

Consider another example of growth of cities, like say Bangalore. Recently, Bangalore grew from a mid-sized town of less than 3 million to a burgeoning metropolis of more than 10 million in a matter of 15 years. This growth was largely spurred by the IT revolution that attracted tech talent, which in turn attracted more companies, which in turn attracted more talent, and so on.

We are now seeing several signs of saturation in Bangalore. Not least of which is water supply. Bangalore does not lie on the banks of on any large river or lake. In fact, it is situated on almost 3000 feet of dry granite rock. A large portion of water needs of Bangalore are met from the Cauvery river, which is more than 100 kilometres away and almost 600 feet below in altitude. It is an extremely expensive proposition to pump tons of water up 600 feet to a distance of more than 100 kilometres. And yet, the socio-economic forces that are spurring growth in Bangalore, hardly factor this saturation constraint.

It is very hard to predict how the city will respond to saturation, and I am very scared to speculate. Being a native of this city, I know in whatever way saturation dynamics will get played out, those of us who have long roots in the city will in some sense, bear the brunt of saturation.

Several efforts to decongest the city have met with little or no success. For example, satellite towns like Kengeri and Yelahanka that were once meant to decongest Bangalore are now part of Bangalore.

One of the reasons why decongestion efforts have failed is that there is no alternative attractor for growth. If Bangalore is the A loop above, there is no B loop that can pose a serious alternative to A, and which is easy to reach from A.

One possibility could have been to have a city like Mysore act as the alternative. And facilitate easy movement between the two cities with the international airport somewhere in between Bangalore and Mysore; and with high speed rail and road connections between the two cities. Now that is not possible because the airport is built at the other end serving no other major alternative growth centre that can compete with Bangalore. In fact, the ideal would have been a multi-transfer hub somewhere between Bangalore and Mysore, where people can fly in, and hop into a high speed train or a bus to either Bangalore or Mysore.

We should be thinking in pairs for every major developmental effort. Because, development is a non-linear process and the very success of a developmental effort could be the cause of its eventual failure due to saturation.

You know the folk wisdom that married people are "settled down" to a more stable life than singles? You know the definition of a "couple" in physics? Two forces that are equal in magnitude and opposite in direction? It is the same thing. :-)


So what happens if the A-B pair itself saturates?

This can be addressed at the next level by pairing two A-B pairs together. For instance, let's say we have a well connected Bangalore-Mysore pair each attracting growth to itself, and (say) a well connected Hubli/Dharwad-Belgaum pair up north in the state, contending with each other. We can now pair these two pairs by connecting the B-M hub with the H/D-B hub with sufficient air, train and road links. Other such pairs in the state could be similarly connected to form hubs at different levels. Now anyone from any part of the state will find it easy to travel to any other part of the state by first going to the nearest paired hub, from where they will find several connections to all other hubs, from where they will find connections to their town of interest.

The same pattern can be replicated in other states and the respective paired hubs tightly connected.

This kind of pattern results in a characteristic property of complex networks that are efficient in daily operations and robust against random (routine) failures. (But not necessarily robust against targeted attacks, which is a different matter.) But let me not go into the mathematical details here.