Moore's Law of Internet Latency

(first postulated in 1994 - Still true!)

Moore's Law of Internet Latency states that:

As long as Internet users do not pay for the absolute (integrated over time) amount of data bandwidth which they consume (bytes per month), Internet service quality (latency) will continue to be variable and often poor.

Analysis - Information User Behavior

(1) Users are ultimately limited by the total bandwidth of their individual connection, a technology/cost factor. The speed of this connection is rising very rapidly due to technological innovation and rapid capital investment.
(2) The only economic cost that a user experiences for marginal increases in bandwidth usage(below the fixed limit) is in latency - a slowing of the response time for that user. Thus in this economic system, the unit of cost is latency.
(3) A user's marginal increase in bandwidth usage inflicts costs on others, but the user does not experience this cost directly ("externalities").
(4) A user exhibits a highly elastic demand for information because:

bulletThere is an almost infinite amount of desirable information available. For example, a user might choose to use full motion sound and video for a screen background, if it were possible. The same user might have high speed data backups occurring between his workstation and a network service. The user may have an "agent" actively shopping for some product by generating hundreds to thousands of inquiries into a marketplace such as travel. This demand could easily exceed hundreds of megabits per second with currently foreseeable services.
bulletMuch of the information is of marginal utility or can be obtained, at lower quality, at significantly lower bandwidth.

(5) Thus, a user increase his bandwidth consumption until his cost (latency) and demand (highly elastic) balance.

Analysis of Information Provider Behavior
(1) Information providers are limited in the bandwidth which they provide by the costs of their internet connection and their server systems.
(2)Competition and opportunity will compel information providers will strive to increase the amount and quality of information which they provide to each user. For example:

Imagine a service which provides beach front ambiance to one's computer or (in the future) a flat screen "picture window." This would include full motion, extremely high resolution video, and multi-channel audio. This service might be paid for by unobtrusive advertising, or by monthly charges. Such a service would consume tens of megabits per second per window. If bandwidth is inexpensive enough, this service will be affordable to many users.

Or, consider an advertiser supported service that chooses to put full motion high resolution video advertisements on it's web site. This is driven by the proven advertiser desire to provide the most impact with an advertisement "impression."

(3) An information provider's marginal increase in bandwidth usage inflicts costs on others, but *may* not inflict this cost on the provider.
(4) An information provider may be less elastic in his demand - a given enterprise may not be practical without a certain level of bandwidth usage. However, the set of all potential information providers exhibits highly elastic demand.


(1) This thesis indicates that George Gilder's widely published prediction that bandwidth will become very large in the near future does not guarantee that latency will be any better than it is today!

(2) [added 2021] In the late 1990's, George Gilder claimed that bandwidth to the home would rapidly increase very soon as fiber was laid. I told him that I doubted it - that the "last mile" cost of fiber would prevent fiber-to-the-home from happening any time soon. Gilder was wrong. We now (2021) see fiber-to-the-home happening but it is still far from universal. I don't have it, for example.