| Complexity
and Organizational Structure by Emily F. Breuner |
|
|
Chapter
4 |
|
History of the InternetThe Internet has been the subject of much media hype in the last few months, but still there are millions of people who do not understand its structure, its nature, or even its purpose. It connects computers and people all over the world and enables two way communication between them, yet is run by no one and everyone. Its decentralized nature is what people find so incomprehensible. The first question people ask is "who runs the Internet?" The answer they get is "No one." For the same reasons that people assume that there is a leader in a flock of birds, they think there must be someone in charge of the Internet. |
|
| The Internet started out as a conceptual problem faced by a group of scientists at the Rand Corporation in the early sixties. How could U.S. forces successfully communicate after a nuclear attack? No matter where a central communication center was located, its switches and wires would be vulnerable to nuclear attack, and it would become the primary target.(1) They needed to design a network whose purpose was to enable the exchange of information under any circumstances, and the only way to do that was to build a network with no central authority, "designed from the beginning to operate while in tatters."(2) |  |
| The scientists developed a concept whose principles were simple: the network itself would be assumed to be unreliable at all times. It would be designed from the outset to transcend its own unreliability. All the nodes in the network would be equal in status to all other nodes, each node with its own authority to originate, pass, and receive messages. The messages themselves would be divided into packets; each packet would begin at some specified source node, and end at some other specified destination node, and contain its own separate address. It would be rather like mailing a ten page letter in ten separate envelopes. The post office would not have to keep such a letter together for its journey from the sender to the receiver so long as the letter could be reassembled when it arrived. | |
| Each node on the network would have a list of other nodes that it "knew" about. If the sending node was unable to connect to a particular node, it would try the next one on the list until it made a connection. This process would be repeated on multiple nodes until finally the packet would reach its destination. This rather haphazard delivery system might be "inefficient" in the sense that a packet would not travel the shortest distance to its destination---but it would be tremendously robust and effective. | |
| Based on some research on this idea, in 1968 the Pentagon's Advanced Research Projects Agency (ARPA) funded a project in the U.S. to connect four supercomputers. The goal was to network these computers in order to share valuable computer time for the sake of national research and development projects. Contracts were awarded to several high-powered defense contractors such as Rand, SRI, etc., who needed to coordinate their pieces of the puzzle and set standards in order to deliver on their part of the project and the project as a whole. The legacy of these standards setting procedures and committee decision processes survives today in the form of the Internet Architecture Board, a volunteer group of individuals who to this day is the only "authority" that evaluates and adopts Internet standards. | |
| ARPAnet, as the project was called, was up and running by 1969, connecting computers through dedicated high-speed transmission lines and distributing scarce supercomputing power around the country. By the second year of operation, however, it became clear that the users had "warped" the network into a federally subsidized electronic post office. People had developed tools that allowed them to send personal messages (electronic mail or e-mail) back and forth, and they were using the network not only to share computing power, but to collaborate on projects, share work-related information, and to "gossip and schmooze." They were far more enthusiastic about this part of the network than they were about long distance computing. | |
| Through the 1970's, the ARPAnet grew. Its decentralized structure made expansion easy as did its well defined standard interface. TCP or "Transmission Control Protocol" converts messages into packets at the source, then reassembles them at the destination. IP or "Internet Protocol" handles the addressing, making it possible to send the standard packets across multiple nodes, networks, and standards. As long as a source or destination machine "spoke" TCP/IP, its brand, ownership, and location were irrelevant. Given the proprietary nature of the computer industry at this time, this interoperability was truly remarkable; the Internet transcended what the market could not, namely proprietary standards. | |
| As the UNIX operating system was developing, its open system approach led it to use the public domain TCP/IP standards as the native networking protocol. UNIX workstations found their biggest market in the research and engineering world, and having this built in capability to "talk " to the Internet only fueled the Internet's growth. As more and more people joined the Internet, its value increased beyond that of the research community just as the value of a fax machine increases the more fax machines there are that can receive documents. This positive network externality effect is an important force in the Internet, and an important motivation for maintaining and enhancing interoperability. | |
| Soon whole networks were joining the ARPAnet, while its own traffic was becoming a smaller and smaller percentage of the overall traffic. The ARPAnet then split into the Milnet and the ARPAnet, and the Internet started to take on its current nature of being a network of networks. Soon it also took on its current name as people started to talk about the "interconnected network," and finally the Internet. | |
| Innovation and Standards As the Internet evolved, the clever individuals who used it created new applications for it. File sharing systems like gopher and ftp (file transfer protocol) were developed by Internet "citizens" who wanted to develop ways to make information sharing easier. These applications enabled people to find information on other Internet nodes and transfer them back to their own computers for their own use. The applications were free, as was the information being exchanged. In several cases, a number of redundant solutions emerged from a number of different people. These solutions were usually merely adequate methods of sharing information, but if they were good, they were improved upon by others and then spread rapidly. These types of applications became de facto standards quickly since the Internet itself facilitates the rapid spread of software and information. |
|
| As the Internet grew and technology advanced, so did the need for standardization. The committee of defense contractors evolved to become the Internet Architecture Board (IAB), a volunteer committee which evaluates standards proposed by Internet users or technology providers. The process by which the standards are reviewed is explained subsequently in greater detail, but it is important to note that the IAB and its supporting group, the Internet Engineering Task Force (IETF), was not put in place to create standards, but rather to evaluate different standards proposed by users of the Internet and choose ones that were likely to encourage its continued interoperability and growth. | |
| The World Wide Web and Internet Today The world-wide web (WWW) was the invention of some scientists at CERN in Switzerland who wanted to be able to share rich graphical data as well as the text-based data that had been exchanged on the Internet for years. The WWW set up standard formats for storing and displaying such data, and researchers and hackers alike rushed to develop browsers for displaying such data. This new set of Internet applications have taken the Internet a quantum leap from the arcane world of UNIX commands to the point-and-click graphical user interface that has driven the widespread adoption of the personal computer. Now that the WWW has made the Internet more "user-friendly," millions more people are hopping onto the Net to become both content consumers and producers. It is this duality of consumption and production that makes the Internet such a powerful force in today's world. |
|
| Perhaps the greatest irony of the Internet is that it was spawned by the fear of nuclear attack, and yet it served a pivotal role in avoiding such an event. The spread of information around the world via the Internet helped to change the conditions in the Soviet Union and encourage the fall of the Communist government there. Because people can gain access to information all over the world virtually without censorship, and also become information providers themselves, the Internet distributes power and authority among people the same way it does technologically among networks. | |
| For these reasons, those who truly understand the Internet feel passionately that it must stay decentralized. However, there is a new group of people who have come to view the Internet not only as a means of exchanging information, but as a means of capturing value. They see it as the Information Superhighway, a distribution medium for goods and services, and thousands of companies have rushed to set up their Internet store fronts. The issue, however, has become one of accountability. It is well and good that an e-mail message can be split up and sent over many routes to a destination. However, imagine a bank using the Internet to enable a customer do her banking from home. If the packet that says "Please transfer $1000 from savings to checking by 5 p.m." does not arrive, and the customer bounces a check, the customer will want to hold the bank responsible. If the packet truly never arrived, the bank will want to hold someone along the transmission system responsible. The way the Internet is set up today, however, that simply is not possible. Nonetheless, as the stakes go up on the Internet, so it seems must the centralized control. | |
| Up until now, the Internet has been quite successful in achieving its goal: enabling the exchange of value under any circumstances. It has done so without being planned, and without anything more than minute government funding ($20 million per year, which is less than the cost of the toilet on the space shuttle). By studying the Internet's current governance process, I hope to learn the secrets of its success so that these "secrets" may be applied to other large scale organizations that struggle with similar issues of coordination, cooperation, and collaboration. | |
The Internet Standards ProcessThere are many bodies on the Internet which seem to have control over some parts of it. For example, InterNIC, or Internet Network Information Center, is a volunteer organization to whom new Internet nodes must apply for a domain name, or an address on the Internet. InterNIC simply ensures that addresses are unique so that mail can be addressed properly. There is also a group called the Internet Society that is serving the role of an industry association. This group monitors activity on the Internet, puts together facts and figures about its use, and also acts in a public relations capacity. Dave Clark of the MIT Laboratory for Computer Science, and one of the founding scientists of the Internet, said that it also serves as a legal umbrella for the volunteer organization that might be held responsible legally for standards adopted by the Internet. |
|
The truly important "governing" body on the Internet is the Internet
Architecture Board, which proclaims standards for the Internet . One must be
invited to serve on this board, and it is largely made up of the people who
developed enabling technology or have been largely influential in the design
of the Internet. While this group is exclusive, its "working group," the
Internet Engineering Task Force, is not. Anyone can participate in the IETF,
whose meetings serve as a forum to discuss and address the immediate issues
that face the Internet as a whole. Problems that interest a sufficient
number of volunteers result in working groups that report back to the IETF
with a recommendation for solving the problem. The system seems haphazard,
but if the size and growth of the Internet are measures of success, the
system works.(3) As one author said:
You should keep one thing in mind about the Internet and its loose controlling structure: it works, and it works far better than do most other organizations. By bringing control down to the low level by requiring cooperation to exist, the Internet works without the strong central government that most countries use and claim is necessary to avoid lawlessness and anarchy.(4) |
|
| The best way to understand the process is to follow an issue through to standard setting. First, a concerned Internet user brings up an issue on one of the many bulletin boards run on the Net. If there is sufficient interest, the participant(s) will go to the IETF meeting to generate concern and interest there in a "Bird of a Feather" session or BOF. If the group deems the issue to be of significant importance, a working group will be set up to consider various proposed solutions. These solutions take the form of well defined specifications of the standard that are placed out on the InterNIC server in a special information directory known as RFC (for Request for Comments). | |
| Each and every Internet user has the chance to read and comment on the proposed standard(s). Good work is lauded, sloppy work is "flamed," meaning that it is criticized harshly via the bulletin boards concerned with the standard. The working group then evaluates competing standards and decides whether or not each standard addresses the issues raised in the RFC stage. Based on its judgment, the working group will make a recommendation to the IETF who will then ratify the standard; by virtue of having approved the working group in the first place, the IAB has already implicitly given its approval of the standard at this stage. The thing to note here is that the IAB is made up of the "Grand Old Men" of the Internet who rarely use their influence to change or suggest a solution of their own, except when they feel it is absolutely warranted. Of course, they are free to bring up issues through the process as "normal" Internet citizens. Through this process, consensus is reached and standards are set. A process handbook representation of this process appears in Figure 4.1. | |
| The whole process is very clearly documented and the time frames are very tightly set. Academics who have seen the process say it closely resembles the typical academic process used to publish papers in journals or select papers for conferences. Sometimes the participation is overwhelming, but since the rules are clear, well understood and accepted, the process is hardly every questioned. One person who is relatively active in the standards process put it this way: "The amount of participation increases the noise level, but the fresh air in the system acts as an antiseptic to the radical free agents."(5) One wonders if this spirit of cooperation and acceptance of consensus decision will hold when large commercial entities like Microsoft begin using the Internet to conduct business, and have interest in setting proprietary standards in order to gain market share and revenue. | |
Figure 4.1: The Standards Process of the Internet Source: Tor Jacob Rams¯yDollar Flow on the Internet in the United StatesI have described some of the culture of the Internet and how its members cooperate to preserve its interoperability. However, to understand the Internet a little more fully, it is important to understand how the dollars flow through the system that physically makes up the Internet (at least in the United States). |
|
| An individual or small business obtains an Internet connection by purchasing equipment to gain access to the Internet (a computer and a modem) and a subscription from an Internet Service Provider (ISP). Currently, there are about 200 of these in the U.S., but this number grows daily. The ISPs provide a phone number and an account on their computer system through which an individual connects to the Internet, with a charge in the neighborhood of $30 per month for such service. The ISPs can be thought of as the retailers of Internet connections, and they typically have customers numbering in the thousands, each of which is using relatively little bandwidth. | |
| In turn, the ISPs receive their connection from wholesalers, second tier Network Access Providers, of which there are about a dozen. These are large regional networks that have very high speed, high bandwidth connections to the Internet backbone. Typically, these each have customers numbering in the hundreds, usually of a size that requires a larger connection than what an individual might need. The NSPs' customers are generally small to large businesses and ISPs. NSPs generally charge on a flat fee basis, and agree to route other NSPs' traffic over their networks as part of being on the Internet. Usage among second tier NSPs is assumed to be symmetric, and occasionally it is actually measured. If there is an obvious discrepancy, the fees are renegotiated using trust-based bi-lateral bargaining. | |
| Finally, there is the top tier of the structure, the large NSPs. These top tier NSPs generally have capital investments in fiber optic networks and include companies like Sprint, MCI, and AT&T. There are 7 of these, each of which has about 30 customers, some of which are NSPs and the rest of which are large institutions whose volume of traffic puts them on par with the NSPs. For example, MIT buys service directly from MCI rather than going through a second tier NSP. Since the demise of the federally funded National Science Foundation network, these seven providers make up the "backbone" of the Internet, or the extremely high bandwidth, high speed connections that handle most of the interregional traffic. These also charge on a flat fee basis, and make periodic bilateral agreements with each of their customers based on the volume of bits going across their networks. Interestingly, within this circle of top tier NSPs, traffic is assumed to be entirely symmetric and traffic is exchanged for free. | |
| Currently, there is plenty of business to go around for all these service providers and they are cooperating fully. However, as the Internet gains in popularity, these service providers will start to compete in earnest for subscribers. Already they are beginning to differentiate themselves in aspects like customer service, security of subscriber networks, diagnostic network software, etc., and there is concern about how long the temptation to compete will be outweighed by the incentives to cooperate. For example, if MCI decides to make a business out of Internet transactions and grabs 40% of the top tier NSP traffic, the traffic being exchanged between the top tier may no longer be symmetric. These types of issues are just emerging today. | |
| The Internet is a facinating example of how an incredibly large, but relatively homogenous group of people can work together to achieve interoperability. In Chapter Six, I analyze some of the Internet's interesting features as they compare to Visa International. Through this comparison, some interesting shared characteristics come to light, suggesting ways that the Internet can serve as a model for corporations of the future. | |
| Ch. 3 Ch. 5 | |
|
(1) Bruce Sterling, "Internet", The Magazine of Fantasy and Science Fiction, February, 1993. (2) Ibid. (3) Adam Engst, The Internet Starter Kit, Hayden Books, Indianapolis, IN, 1993: 35. (4) Engst, 36. (5) Mitch Kapor, "MAS962: The Politics of the Digital Infrastructure," seminar, MIT, Cambridge, MA, March 4, 1995. |
|