Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : LAST MILE TECHNOLOGIES - Let's Discuss Them Here -- Ignore unavailable to you. Want to Upgrade?

To: lml who wrote (2516)	12/9/1998 9:08:00 PM
From: ftth	Respond to of 12823

Here's a link to some analyses of internet traffic and statistical characteristics: cs.vt.edu. (See especially "Analysis and modeling of WWW traffic" and Characterizing Web Response Time." If you look at the slides of the latter one, you'll see the infamous "internet traffic doubles every three months"). More than one way to skin a cat: MANY different growth measures could be used to come up with "traffic statistics." In fact, there are so many possible permutations and combinations of things that could be measured, how they're measured, when they're measured, where they're measured from, and whether to use average, peak, median, ... of the data collected, that any statement regarding internet traffic growth is all but useless--especially if it's some empty statement with no supporting facts about how, what, when, where it was measured. Besides, the growth by any measure is nonlinear--you can't fit a curve to it (it grows in "fits and steps") and make reliable long term predictions from it. Using a small snapshot of collected data over a few weeks or months, and a little data mining, you could surely fit some data points to everybody's favorite growth curve: the exponential function (like "annualized growth in the average of the peak html page fetches from 3-5 pm EST from the Ameritech NAP over a 1 week period":o). To extrapolate that data forward for multi-year planning purposes with the assumption that the data will continue the same trend is silly. The model is changing while you're making the measurements. Subscriber growth is dynamic; service provider growth is dynamic; infrastructure growth is dynamic; domain growth is dynamic; content growth is dynamic. But, none is dynamic on the same scale as any other. Mixing all these "growths" and trying to aggregate them into a single statistic won't produce anything that is predictive over more than the next day or two at best. Data significance is all but lost by trying to force reams of data into a single high-level figure that the masses can understand and run with (i.e. suitable for a press release). This practice is far too commonplace anymore, don't you think? dh