Until
next week, 
Crystal Balls: Forecasting the
Future
Dateline: February 15, 1998
ARTIFICIAL intelligence is happening now, and what is happening now is exciting enough. But it is nothing compared to what is coming down the road. Readers of my previous features will be well aware of my prediction that a new lifeform, Machina sapiens, is heading our way faster than aliens from Andromeda. How do I know? Because I study the future. How do I do that?
Read on . . .
Back in 1965 Intel co-founder Gordon Moore gave us Moore’s Law, which told us that the number of transistors on computer chips would double every year. Early chips had a few hundred, today's top chips have about 8 million, and billion-transistor chips are expected by about 2010. Moore's Law is an example of a successful attempt to predict the future, and Intel for one has staked and won billions of dollars on its reliability. Similarly, Sun Microcomputer Corporation based its winning strategy of selling network-ready servers and workstations—summed up in the slogan “The Network Is the Computer”—on Metcalfe’s Law, declared by Ethernet inventor Bob Metcalfe, which holds that the value of networks rises by the square of the power of all the computers attached to them.
Such “laws” are not like the laws of physics, nor even like scientific theories. They can be shown, as a matter of historical fact, to have held true in the past, but there is no scientific proof that they will hold good for the future. They are more like working hypotheses or applied common sense; reasonable rules of thumb but neither certain nor precise. So why should we go on believing them, and why should we listen to any prophecy lacking a scientifically solid methodological basis?
The simple answer is: Because they seem to work, they have been right so far, and in the absence of anything better they must still be presumed to point in the direction of truth even if their accuracy is suspect at some arbitrary level of detail.
Future Studies
The field of futures studies is a relatively new professional discipline, arising in the 1950s and 1960s in response to growing complexity and change. Its methodologies, invented in think tanks and activist organizations, include content analysis, scenarios, cross-impact analysis, and adaptations of the old Delphi brainstorming method.
The field is different from traditional strategic forecasting and planning disciplines. It considers a longer time horizon—typically 10 to 50 years out—than economists and market researchers who look out 1 to 3 years. It focuses on systemic, transformational change as opposed to incremental changes from existing trends. It forecasts multiple alternative, possible, and preferable future scenarios rather than single predictions. And it uses qualitative as well as quantitative methodologies (particularly demographic projections from statistical data), where traditional forecasting tends to rely on purely quantitative tools.
Scenarios and Content Analysis
Forecasting by means of scenario building is more like “mapping a cone of uncertainty” than predicting a specific and certain outcome. It is a process of mapping ignorance. A forecast can be designed to reduce uncertainty by allowing its recipient not merely to lie in wait for a predicted outcome and then react to it, but to take action to influence the outcome ahead of time.
One futurist has suggested that while the scenario process is best applied to specific problems— “Should I build a new plant, or buy a company, or get out of a business?”—it is nevertheless important also to identify and forecast the social, political, economic, and technical context surrounding the problem. This latter is what analytical writers like Alvin Toffler (Future Shock) and John Naisbitt (Megatrends) do so well, and they do it using a process called content analysis. Basically, they employ teams of documentary researchers to read through hundreds of newspapers and magazines over a long period of time, culling and collating factual information (not editorial opinion) about social, economic, and other issues.
Successes
Armed with a few years'-worth of such collated information, it is not difficult to write a book like Megatrends. I know, because I did something like it in a former incarnation as a “China watcher” for the British and Hong Kong governments, and also on a smaller scale for Ameritech in 1988. I also used the technique in building my own business plan in 1989-90 for an online service. In all three cases, my 10–20 year predictions turned out to be pretty accurate. The online company I founded thrives as Michigan’s biggest Internet service provider, Ameritech has handled the uncertainties of the past decade better than the other Baby Bells, and Britain and China settled the Hong Kong question amicably. This is not to boast, but only to illustrate that content analysis works.
It worked, for example, in South
Africa, where the Montfleur project, designed to secure the country’s
future in a post-apartheid world, brought together such unlikely bedfellows
as the "one settler, one bullet" Pan-African Congress with rabid white
supremacists, as well as with the moderate African National Congress, the
Incata movement, the Chamber of Commerce, the Government, and unions. Together
they forecast a set of scenarios, published in the South African press,
that convinced everyone that the current path spelled doom for all and
that a different future could be good for all. A key product of this work
was the mandated Truth and Reconciliation Commission, charged with uncovering
the truth about apartheid-era abuses, reconciling the country's divided
races, granting amnesty to those who confessed political crimes in full,
and paying compensation to victims. While South Africa has not been free
of troubles, the feared and anticipated bloodbath was averted.
I believe the topic of AI is deserving
of a full-blown future study, but the best I can hope to achieve through
my scribblings here at The Mining Company is to present to you some of
the results of my content analysis of the trade and popular press for AI
developments and to suggest some plausible impacts and future scenarios—ceteris,
as the economists are careful to say, paribus.
Until
next week,
NEXT WEEK: Crystal Balls 2: Why we need them.