Let a thousand connections bloom

The first phase of IT applications featured computers doing useful calculations. The next phase, which really gathered momentum worldwide from the late 1980s onwards, consisted of connecting up computers, making useful networking possible. This next phase has multiplied the usefulness of computers manifold.

Consider a bank branch. It is obviously possible to make the bank branch more efficient using computers. The next step consists of having minimal links to the head office, whereby the status at 5 PM every day is sent to the head office. Going further, 24 hour connectivity makes it possible for customers to operate their account from any branch, regardless of where the account was opened. This is particularly useful in the context of using any of a network of ATMs. Along the way, connectivity is likely to go along with email within the bank, enabling tighter integration and control of activities at each branch. As applications start exploiting connectivity, the network capacity would have to grow.

This example illustrates a common theme of network growth -- networks tend to grow in a step by step way, responding to needs. There need not be a single grand jump from "no networking" to "complete networking".

The origins of the Internet are shrouded in early attempts by myriad individual sites to link up to their neighbours. The objectives at the time were certainly not about building a gigantic, planet--spanning network which would prevail over the most powerful corporations. The modest objective of each site was to hook up to neighbouring sites and exchange email and news.

In doing this, it was found extremely useful to route traffic: in a network A -- B -- C, it is a trivial cost for B to volunteer to take messages from A and send them on to C, and vice versa. This can be done instantly if all three sites have 24 hour connectivity, and with delays if this is not the case. Such flexible, efficient connectivity was rooted on two vendor--neutral technologies: a computer operating system called Unix and a set of networking standards called TCP/IP.

Networking typically started out with extremely low--cost links -- such as local calls or late--night STD calls. Once it became useful and ordinary telephone lines proved to be inadequate, the networking moved on to more powerful links such as leased lines. But it is all too common to understate the potential of ordinary voice telephone lines.

Through the 1980s, a host of grassroots efforts sprang up worldwide, of sites linking up to each other using whatever means were convenient and affordable. This led to an ever--increasing scope of connectivity, and newer applications kept coming up, which justified further expenditures on connectivity.

This was obviously a windfall for telecom companies. A common joke in the hacker community imputes a deep conspiracy to the phenomenon of AT&T developing this marvellous operating system, Unix, and giving it away to the world for free. Once users started exploiting the remarkable networking capabilities of Unix, they would be forced to rush back to AT&T to buy telecom capacities!

Whether this was the intention or not, it certainly worked. Sites continually upgraded their telecom links, in response to use. Powerful telecom made new kinds of connectivity possible, spanning from plain email to the world wide web. Unix evolved and grew along the way, becoming the foundation of the modern networked enterprise. Unix networking got a new name, the Internet, which grew at rates like 5% per month, and became a phenomenon which hit the front page of newspapers and prime time television.

Where do we stand in India in terms of this growth? Can an A -- B -- C network spring up at the grassroots, out of the clever responses of individuals at these sites to their needs, using whatever communications links are at hand? No.

This entire process has been killed off in India by one single government policy intervention: the "Closed User Group" (CUG) policy. The CUG policy says that links that go outside one office can be done within one CUG (e.g. the branches of a bank) but two CUGs cannot connect to each other. At one stroke, this policy eliminates the potential for creative, grassroots networking in India.

This has led to tremendous waste and inefficiency. A classic example is the delivery of depository services. Supose a bank has built a VSAT network connecting up its branches. The depository has built a VSAT network connecting it up to depository participants (DPs). The link from the depository to the bank extends till the head office of the bank. From here, the two CUGs cannot be connected together! For each branch where the bank wishes to offer depository services, it would need to put in an additional VSAT terminal!

The potential of networking is: once a bank has built a VSAT network, by merely obtaining a link from one location to NSDL, all branches can immediately offer depository services to customers. The CUG policy eliminates this potential.

A similar example is the host of email and Internet initiatives which have sprung up in the country. By the CUG policy, they are not allowed to connect up to each other. This leads to situations where email from the 4th floor of a building in Delhi reaching an office on the 5th floor of the same building via a computer in the US.

These are direct situations where the CUG policy generates waste. But the most fundamental cost is hidden. The CUG policy blocks the creative efforts at every site in the country to think about how useful hookups can be obtained with a few entities around them. The CUG policy blocks the growth of IT in the country in the direction of understanding and exploiting computer networks.

From an economic policy perspective, the CUG policy is remarkable: it amounts to the owner of a road discriminating against certain kinds of traffic even if it pays full fare. Telecom companies have a right to charge fees for telecom links. Beyond this, telecom providers should have no role in determining how these links are used!

From a political perspective, free interactions between the citizens of India in this age are critically linked to unrestricted information exchanges that citizens (and not government) choose. Restricting citizens in using links in certain ways is exactly like restricting citizens from using roads in certain ways. This should not be allowed to happen in a free society.

The IT task force has done many important things, but few will perhaps have an impact as large as the few lines on the subject of interconnectivity of networks. Now we find debates about various kinds of partial notions of connectivity which could replace the CUG policy. A partial removal of the CUG policy would help, but it still involves a world where government can dictate conditions about how individuals in India communicate with each other. There is a principle here which should be brought to the fore: "Telecom providers should have no powers in restricting the nature of traffic that runs on a telecom link". This freedom flows from sensible notions of democracy and from sensible notions of economic policy. Moreover, the uncompromising application of this principle will yield a revolution in IT in India.

Back up to Ajay Shah's media page