Paul Saffo on Forecasting Innovation in Silicon Valley

Written by Sean Murphy. Posted in Events, skmurphy

Paul Saffo came to the Churchill Club for breakfast and went beyond his HBR article “Six Rules for Effective Forecasting” to address forecasting innovation in the context of Silicon Valley. Currently on sabbatical from his day job at the Institute for the Future and teaching at Stanford, Saffo quipped that a well executed sabbatical required the same commitment as entering the federal witness protection program if you really wanted to get something new done: “you have to change your e-mail, your phone number, and never set foot in the office until you are ready to come back to work full time.”

He took pains to characterize himself as a forecaster or “professional bystander” and not a futurist or “advocate for a particular future.” He had to recover the lost art of speaking-without-PowerPoint since the Churchill Club folks told him it wasn’t allowed. He drew some compelling word pictures but it was ironic that we were in a nice big room at Fenwick where a rear projection screen covered the wall behind him–albeit partially obscured by two portable Churchill Club banners.

The interesting part of the S curve is the twenty years before the inflection point (“take-off point”), the real long tail is the twenty years it takes to be an overnight success. In Silicon valley the civilians are surprised only once. Nothing new ever works, which matches their expectations, but when it finally does they are surprised. Most technologists are surprised twice. They stand at the dawn of a new technology and underestimate how long it will take to succeed, become despondent close the inflection point and move on to something else, at which point they underestimate the pervasiveness of the diffusion.

Most things fail. Silicon Valley is built on the rubble of early failure. The Valley knows how to fail. If you were here for the dotcom crash you understand why we did better than Wall Street in 1929, their buildings were 40 stories tall looking out onto city streets, ours are two stories tall surrounded by grass; the worst we could do was sprain an ankle in the fall and have to hobble home.

If you want short term success look for something that’s been failing for 20 years. I spend my time looking for prodromes, early indicators of significant changes, and listening for the “doppler shift whistle” that things are about to take off.

Saffo brought two examples of early TV remote controls, one that was essentially a flashlight, and another that played tones and made the observation that the question to ask is not “will these succeed” or “why didn’t these succeed” but “what does this mean.” He also brought in one of the $100 laptops built by Nicholas Negroponte’s group: even though it may not succeed it has triggered a cascade of competitors who want to build a $99 laptop with different hardware and software. 80% of the world doesn’t own a PC so the market is there.

Silicon Valley can design products that can be “re-invented by consumers.”

By which I think he means that we design for end user customization as a source of differentiation. He also brought a device made by Faber-Castell that was an electronic calculator with a slide rule on the back and observed

“most people looking at this today misunderstand what was going on with this, the slide rule was there for functions that were too complex for the level of sophistication that early silicon chips could support. More importantly, slides rules continued to be used by the engineers and other professionals that knew how to use them, but they remained very complicated to learn and operate accurately. Calculators replaced pencils and paper: they were used by regular people who far outnumbered the experts, so the market was much bigger. And since calculators were based on silicon chips, the cost advantage went to the semiconductor companies not the existing slide rule firms.”

He had one very provocative observation.

“Analytical forecasting methods are now getting robust enough to effectively complement the primarily intuitive techniques that most forecasters have relied upon to date. The Singapore military establishment has developed “horizon scanning” technology that has allowed them to anticipate and neutralize a number of incipient threats.”

He may have been referring to the SenseMaker software developed by Dave Snowden of Cognitive Edge. Snowden and Valdis Krebs were on a recent call sponsored by Plexus Institute, a summary follows:

SenseMaker – Software supported by methods and an international network for managing uncertain and complex environments.

The ideas behind the software originated in knowledge management, the work of Cognitive Edge in the use of narrative and in understanding the basic patterns through which human beings make decisions….We will always know more than we can tell and we will always tell more than we can write.

The development has been funded by the US and Singapore governments in the context of risk assessment, horizon scanning and the detection of weak signals. A basic assumption of applications of the software is that there is little difference, from an organizational point of view, between a terrorist, an ordinary citizen, an employee or a customer. They all represent the problem of asymmetry, in which an organization has to understand multiple interactions and decisions from large populations which cannot be predicted or controlled by that organization.

The software and linked methods allow a collection of multiple sense making items, which can be anecdotes, web sites, pictures, or blogs. They can be tagged, or labeled, to provide sophisticated metadata. This data is combined with visualization tools linked with methods and models that permit users to sense complex patterns and anomalies that would not be otherwise visible. It is a “pre-hypothesis low-cost research tool, a knowledge repository, and a decision support system” in one package.

This is one of the few software systems built on the basis of natural science rather than management science. It is designed to accompany, not replace, human decision making. “it enables serendipitous un-biased encounters with data in the context of need.”

  • This is designed to enable a symbiosis of machine and human decision making.
  • It enables a quantitative approach to motivations and others issues that have traditionally been handled by qualitative approaches.
  • It provides for “identity marketing” and “ethical and cultural auditing.”
  • It is based on complex adaptive systems theory and an understanding of non-causal, unpredictable systems.

A couple of quick observations. Saffo was careful to point out the he did “real work for real clients who want real answers” and I was left with the impression that they like a nice single narrative. His cone of uncertainly model seems to unfold around a single trend, or one trend giving way to another (television creates broadcast media, time sharing creates e-mail as a media, client-server computing erects the World Wide Web on the wreckage of a billion dollars wasted on interactive television, and sensor networks create smartifacts or blobjects that are the next generation media). His article works within Thomas Kuhn’s “paradigm shift” model where a single dominant paradigm in a field of science gives way to a new one based on accumulating evidence and the death of older scientists who clung to the old theory.

But I think an approach based on “analysis of competing hypotheses” (ACH) might be a stronger platform, especially when it comes to technology markets which may have multiple segments served by multiple competing technologies. In other words, instead of thinking is the present giving way to this new technology, you could look for multiple trends that are reinforcing and competing for resources in different markets. For example, in the “Gunfire at Sea” chapter of “Men, Machines and Modern Times” by Elton Morrison makes the point that the modern naval gunnery was enabled by the intersection of metallurgy advances that could build longer strong gun barrels giving improved range, improved optics so that you could see and range find what you were aiming at since it was a mile or more away, an elevating gear to compensate for the rolling motion of the deck.

Together these three inventions enabled a continuous aim firing method that saw 175x increase in the rate of effective fire in 6 years”

After twenty-five minutes of banging away, two hits had been made on the sails of the elderly vessel. Six years later one naval gunner made fifteen hits in one minute at a target 75 by 25 feet at the same range–1600 yards; half of them hit in a bull’s eye 50 inches square.

It was a great talk and there were a number of excellent questions from the audience. When Churchill Club puts up the audio I will update this with a pointer.

Trackback from your site.

Comments (5)

Leave a comment

Quick Links

Bootstrappers Breakfast Link Startup Stages Clients In the News Upcoming Events Office Hours Button Newsletter SignUp