It's hard for an analyst not to feel just a tiny bit envious of those responsible for Cisco Systems Inc.'s Visual Networking Index (VNI) -- the index that launched a thousand (no, make that ten thousand…) PowerPoint presentations and has become the go-to reference point for anyone looking for a fix on likely network bandwidth requirements over the next five years. (See Cisco Report Predicts Mobile Domination in 2017.)

Envy begets skepticism, and most analysts are apt to cast a cynical eye over other people's forecasts, knowing how difficult it is to forecast anything with real accuracy except the sunrise. But putting cynicism to one side for a moment, there is a serious issue here: Why do people so often present Cisco's forecast as though it was somehow established fact?

Talking to the director of network planning at a major European mobile operator, we were interested to hear that he thought the VNI forecast for mobile data traffic was somewhat overblown. Yet, at the same time, he acknowledged that any such forecast is, as much as anything else, a matter of opinion, not fact -- he just happened not to have the same opinion as the Cisco analysts. This operator said that it creates a very detailed (and generally pretty accurate) internal traffic forecast for the coming 12 months, but said its own five-year forecast is seen as much less important -- necessary for long-term planning, but with the caveat that any forecast over a year or two out is apt to be disrupted by unexpected events and so must be continually recast.

And indeed, any large-scale forecast does raise some pretty big questions. For instance, what if we reach a point in the near future when enough bandwidth, so to speak, is enough? That isn't a fashionable view, but nor is it a ridiculous one. Human beings only have so much time and sensory apparatus to process the flood of data coming in. Perhaps the endless bandwidth escalator we've been riding will slow, or even stop in the next five years?

And yet, on the other hand ... perhaps there is an application germinating right now in the mind of some unsung genius, about to burst into life and fill our pipes with some hitherto unknown application? These are the events that Nassim Nicholas Taleb, in the course of his withering attack on almost all forecasting exercises, called "black swans": events that no-one can predict.

So does this mean that forecasts like VNI are not worth doing? Not at all. The real purpose of forecasts is, or should be, to create the basis for a debate. Does the forecast seem credible? If not, why not? What factors were taken into consideration? Were they the right ones? What might disrupt this forecast? These are qualitative matters of opinion, not quantitative matters of fact.

Figure 1: Questions, questions, questions...

Putting my cynicism aside, I have to acknowledge that Cisco really has done the legwork and created a set of very defensible scenarios. The forecast takes into account a wide range of secondary inputs from analysts, government agencies and others, as well as using data collected by Cisco itself. VNI is a great resource, and it's free. But it's just that -- a resource to be deployed and debated, not a set of stone tablets to be worshipped. And I’m pretty sure its creators would agree.

– Graham Finnie, Chief Analyst, Heavy Reading