Fred, 50 years is a awful long time. Human knowledge has virtually double every 2-3 years now. In the last 50 years, we've had many, many things that have changed our lives dramatically: the popularity of TV, the construction of interstate freeway, the invention of fabricating intergrated circuit cheaply, the PC evolution, the invention of satellite and the technology to use satellite to transmit data around the globe, the trans-atlantic fiber optics line, the landing on the Moon, the nuclear poliferation during the cold war, the biotechnology break through in which animal cloning is now a reality. Will Internet still be here in 50 years??? Absolutely. Will E-Commerce still be the way people do business in 50 years??? Probably NOT. The reason is simple, 50 years is too long a time, base on what happened during the past 50 years, it's safe to say that there will be other evolving technologies emerge in the next 50 years.
There are a lot of problems facing the internet today, here are my technical analysis:
1. Broadband problem: it's too frigging slow!!! most people still have 28K or 33.6K modem. Without caching, people are still spending minutes to load the home page of a popular site. For people with broadband access via a T1, ISDN, DSL, there are few sites that are tailor-made for fast access. The underlying server application simply couldn't distinguish a fast connection or a slow connection. TCP uses a traditional slow start approach in its sliding window algorithm. It's not good for the WEB and the http protocol since it penalizes a fast link with medium data transfer. It's good for slow link with small data transfer. HTTP building on top of TCP is not a good idea.
2. TCP/IP was originated from the ARPANET project from the department of defense. It served its purpose well up to a certain period. However, the current internet doubles its size every 1.5 years. At this rate, static IP address will be exhausted in 4.5 years. The current IP version 4 scheme only supports 4 billion IP addresses, but not all 4 billion are available because there are special IP addresses reserved for multicasting group purposes. We need IP version 6 to solve this problem. I wrote a paper illustrating this problem and evaluate couple proposed solutions my class:
cs.washington.edu
In my opinion, TCP/IP is a generalize approach to solve everybody's problem. That's just not possible. It will outlive its usefullness in the next 50 years, I am very positive about this.
3. Politics: everybody sees the Internet as a huge opportunity. Everybody wants a big pie of the Internet. As a result, you have a huge divisive groups of company, groups of different opinions on how the Internet should evolve. MSFT tries to wrestle the control of the internet by defining its own set of standard in Java programming, SUNW is doing the same thing, CISCO is trying to sell its ATM technology while Lucent, Ascent are trying to do the same thing. It's very hard to get them sit down together to reach a consensus standard. All you have are these court fights between the companies, or between the company and the DOJ. It's not beneficial to the good of the technological innovation. We don't have an environment where a dictator like the US government says: "OK, all the states, let's build these inter-state highway and connect them together, this will be good to the military mobilization as well as the economy." You have a system where Kansas says I am building my freeway inside my state only, I don't build anything connect to Missouri. In this environment, it will take a long time to weed out the winners and losers, and it's very bad for deploying an outstanding technology to benefit everybody.
Base on the above reasons, I don't believe the TCP/IP based WEB technology will be the way we do business in the next 50 years. 10-20 is probably a good bet, but not beyond that. |
