Chapter 9. The Do Loop – The Eight Skills of Adaptive Foresight

Skill 2. Anticipation (Convergent thinking).

In the Evo Devo model, the only foresight thinking and action skill that is as fundamental as possibility exploration and generation is probability exploration and generation. Even though we may only be able to usefully predict 5% or so of our future at any time in many environments, as the 5/95 Rule proposes, finding that special set of predictable elements, including understanding what things are presently accelerating, converging, and emerging, and placing uncertainty boundaries around the less-predictable elements, gives us a framework of constraints on the future, and a critical advantage in strategy and action. For just one excellent example of anticipation, read Thomas Hainlin’s From Headlines to Trendlines: Long-Term Investing for Wealth Expansion (PDF), 2013. The more high quality info is accessible via the web, and the more evidence-based our models, the more key aspects of our near-term future become predictable and profitable to understand.

Great Anticipation is All Around Us Now

Great Anticipation is All Around Us Now

As Bill Clinton of the Clinton Global Initiative says, anticpators must learn to look beyond the headlines, which are often inflammatory and emotion-oriented, to find the trendlines, which are often going the opposite direction to what the media is portraying, for their own practical, self-serving reasons. The reporter’s cynical expression is “If it bleeds, it leads.” Meanwhile, the modern world grows safer, stabler, richer, and cleaner every year forward. But those real trendlines don’t sell papers, so the headlines continue to drive most political and economic activity, and we obsess over ever smaller risks and dangers. Nevertheless, the growing immediacy and ubiquity of our global media, and our constant attendance to headlines, even as it distorts our perception, does tend to make a safer and better world.

A Top Forecasting Community

A Top Forecasting Community

Forecasting is the best-known anticipation function, and the International Institute of Forecasters (IIF) and Scott Armstrong’s ForecastingPrinciples.com are two great practice communities.  As Paul Saffo says, to get good at forecasting we need to do it often, we should stick to subjects we can model well (mentally or formally), and we must follow up with post-forecast review and analysis. Saffo’s “Six Rules for Effective Forecasting,” Harvard Business Review (2007) (PDF) offers a great intro to anticipation practice. Saffo’s Six Rules are: 1) Define a Cone of Uncertainty, 2) Look for the S-Curve, 3) Embrace the Things that Don’t Fit, 4) Hold Strong Opinions Weakly, 5) Look Back Twice as Far as You Look Forward (better yet, look back as far as your time and resources allow), and 6) Know When Not to Make a Forecast (Know When You Are Maximally Ignorant). This is all excellent anticipation advice.

Unfortunately Saffo, whom I respect and consider a mentor, genuflects a bit to the popular but incorrect perspective that “forecasting is not about prediction”. While I agree it isn’t about prediction for evolutionary processes and events, which are the large majority of what we see in the world, I am quite sure forecasting is about prediction when we are talking about seeking to find and validate that critical 5% of developmental processes and events that are emerging ahead of us. In other words, I would argue his perspective is “95% right” in practice, but only 50% right in impact. Developmental processes and events, (think of globalization, information growth, Moore’s law, mobile, cloud services) few though they may be in number, are often so powerful as environmental factors, drivers and constraints that they rival the much more common evolutionary events in their impact on our future options and strategy. Finding relevant developmental (probable, predictable) processes and events is the central goal of the anticipation skill.

Qualitative forecasting is also called judgmental forecasting, or if it is aspirational, visioning. Qualitative forecasting can be an excellent start, but turning those into quantitative forecasts and predictions may be the most important for strategy, and once they are that specific their after-the-fact error can be easily reviewed (Skill 8), and adjustments made for future efforts. A great forecast or prediction always has some probability attached to it, and is shared in a critical internal or external community, which analyzes results to better calibrate future forecasts.

Particularly specific forecasts are called predictions. Their specificity makes them both particularly valuable and particularly risky. As a result, most of the professional foresight community does them rarely. That is their loss, as there are a host of obvious and nonobvious things we could say about any system’s future state, and it is the nonobvious predictions that can be most helpful in our clients strategy, plans, and actions. Some of us even try to convince our colleagues that “prediction is not what we do”. That is simply incorrect. Any futurist who offers a wildcard is saying that particular future is a low-probability, high-impact event. That’s making a prediction. If we are to improve at forecasting and prediction, we need to be honest about how much we already do, and try to always assign probabilities to our anticipation work. If all we can presently see ahead are low-probability predictions, it’s much better to communicate those than to do no prediction at all.

Flyvbjerg (2003)

Flyvbjerg (2003)

Reference class forecasting, developed by psychologists and economists Daniel Kahneman and Amos Tversky, is a clever strategy of measuring, predicting, and eliminating certain systemic biases in long-range forecasts. It starts by researching actual past outcomes in a reference class of similar actions to the one being forecast.  Program management expert Bent Flyvbjerg developed methods for its use in large projects and contracts, common in construction, development, and defense. The basics of his method can be found in Megaprojects and Risk (2003), but go to his papers for the latest. Flyvbjerg’s research shows, for example, that large public construction projects are usually underbid by a certain stable percentage, defense contracts underbid by another stable percentage, etc, for various types of bidders and various industries and countries. If you can find a good reference class for past forecasts in a particular country, industrial sector, and with a particular supplier, you can adjust your forecasts to eliminate that systemic bias and do better procurement and project forecasting and decision-making.

Information and communications technologies (ICT) will greatly change the nature of anticipation in coming years. For one survey of what is coming, see Keller and von der Gracht’s ICT Tools in Foresight, Technological Forecasting & Social Change, June 2014. The authors propose that between now and 2020, typical ICT-aided foresight exercises will shift from their current use in scanning and data retrieval (Skill 1), to ICT for anticipation (Skill 2), strategy and decision-making (Skill 4) and execution (Skill 5). Yet over this time period we can also forsee how ICT will improve the other four of the eight skills as well. Having strong ICT skills should help this decade’s foresight consultants like never before. Let’s look briefly at a few examples.

Predictive analytics is an exciting new field that analyzes current and historical data to make quantitative predictions. One of its key features is simple mathematical modeling (correlational and causal models). It is a subset of data analytics, a term that has become popular with the rise of the modern web and big data. Siegel’s Predictive Analytics (2013), is an excellent beginner’s introduction to this rapidly emerging new field.

Siegel (2013)

Siegel (2013)

To consider how predictive analysis can help with strategy (Skill 4), let’s look at a specific example. My wife, who works at Google, was recently involved in a research study to find ways to improve the number of US women majoring in computer science. Initial research surfaced over twenty potentially important intervention variables, and there was conflicting research on each variable’s relative importance. Her team decided a logistic regression model and conjoint analysis, with good pre- and post-survey data, collected from a large group of women who had completed such majors, would allow them to cut through the noise and determine which variables were likely to be the most important. Such a model also ranks the variables in importance, without indicating causal relationships. They learned that parental encouragement (regardless of the parent’s profession or social status) and positive exposure to coding in high school (regardless of whether an AP computer science class or a much easier summer coding extracurricular activity) were the two most important factors. They can now design specific interventions emphasizing these two factors, and measure their efficiency of impact (ROI) versus other spending to incentivize US women doing computer science work in college. By publishing their research, they also invite others to use it, and to independently verify or falsify their findings.

The number of organizations doing policy and social interventions informed by this kind of predictive analytics work today is fewer than you might expect. Such work is often not expensive; it just requires an anticipative and analytical approach. But unless we have evidence-based champions in the leader suite, we often find it easier to stick to our traditional seat-of-the-pants heuristics, or maintain the do-nothing position that the system must be too complex to predict. We should not be surprised if such an approach gives mediocre results. Fortunately, as the increasingly intelligent web makes evidence-based anticipation even easier in coming years, and as more of us understand the greater performance gains available, prediction will increasingly be adopted in business strategy.

Thompson (2012)

Thompson (2012)

Two promising and presently underdeveloped group anticipation methods are real-time Delphi (online group estimation and forecasting), and prediction markets (group forecasting offering financial reward or other incentives to find the best predictors by subject area). Huunu and Zocalo are two early efforts in the prediction market space. See Sunstein’s Infotopia (2008), and Thompson’s Oracles: How Prediction Markets turn Employees into Visionaries (2012), for two good sources on real world experience with prediction markets today. As with group estimation (recall the Jelly Bean Estimation Challenge), studies have shown these platforms can provide more accurate forecasts and richer sets of alternative futures than those offered by individual experts. Even though it makes management more challenging, the data also show that leaders need measurable cognitive diversity on their anticipation teams if they want the best results on complex, poorly structured anticipation problems. See Scott Page, The Difference (2005), for evidence for this claim. There’s also good evidence that we need to mix predictive analytics methods (data and algorithms) with methods that rely on collective human judgment (Delphi and prediction markets) when assessing more abstract variables. Nate Silver’s impressive analytics techniques for predicting US presidential primaries in 2008 and 2012 were much less impressive when applied to World Cup predictions in 2014.

A few foresight consultants offer real-time Delphi platforms today, but their software and interfaces are primitive at present. Many firms have started and then abandoned internal prediction markets, as implementation requires careful participant training and facilitation. Unfortunately there are few successful commercial offerings today in this space. One non-commercial research platform of note is Philip Tetlock’s Good Judgment Project at the University of Pennsylvania and U.C. Berkeley. It is presently pioneering leading methods for social and political prediction, and publishing its findings, which are impressive. But today, public and private funding for collective forecasting research remains quite small. I expect we’ll need a better semantic web, and better-validated methods before we see real-time Delphi and prediction platforms take off in corporate environments.

Hubbard (2009)

Hubbard (2009)

Investing (asset management) is an organizational function where predictive quantitative models have made major advances. We’ll say a bit more about investing at the end of Chapter 7, as it is such a personally and professionally useful anticipation skill.

Risk Management is a newer anticipation function, and a thriving professional community with its own literature and methods. As we’ve said, some of the most validated predictive models in the world are built by the reinsurers, like Munich Re, Swiss Re, and many others. This is to be expected, as they have so much money riding on their bets, and as they know, from good research, that anticipation works. There are many smart risk assessment, mitigation, and insurance strategies available today, even to small firms. A good primer is Tom Kendrick’s Identifying and Managing Project Risk (2009). Doug Hubbard’s, The Failure of Risk Management: Why It’s Broken and How to Fix It (2009), is a great recent work on the field. Risk Management is now big enough and well-funded enough that it’s been misapplied by many firms, just as strategic planning was poorly applied in the 1960s-1980s. When a field has grown up enough to be critiqued, and when those critiques are acted upon by business leaders, as some are doing with this book, that is progress of a sort!

Finally, Law & Security are two related anticipation functions that both seek to protect the firm’s assets, and guard against loss. Protection-oriented anticipation (law, security, and risk management) is as important to the firm as predictionoriented anticipation (predicting opportunities, as in investing, and predicting issues, threats, and risks as in risk management). Different personalities tend to be attracted each of these core functions of anticipation. Both of these functions help the firm survive, but perhaps the greatest goal of group anticipation, in a time of accelerating change, is to be able to occasionally see very worthy opportunities opening up ahead.

In the Evo Devo model, a good practice guideline for Anticipation, or developmental foresight, is to seek worthy destinations. Recall that anticipation is concerned with foreseeing where things seem to be headed regardless of our individual creative acts, uncovering probable future opportunities and dangers, and protecting and preserving what we value. The skills of anticipation come largely from the rational and logical side of cognition (forecasting), and with managing negative emotions (fears, vulnerabilities) related to risks and uncertainties (security). We can oversimplify a bit and say anticipation is managed primarily “from the head”. It is a conservative process, and it requires good data and models, and intellectual honesty.

One desirable anticipation practice is identifying environmental developments that are both increasingly inevitable and highly positive sum for all concerned, then getting on the right side of history with those expected developments in your mission, goals, and strategy. As a leader that values legacy, you might like your firm to be an instrument of the great developmental processes unfolding in your society in your era, at least the ones consistent with organizational goals, rather than fighting those emerging realities.

For example, in 1860s America you’d hope to be doing your small part to end legal human slavery, and today, ending it internationally and in immigration slavery and sex trafficking, rather than trying to oppose ending it, or profiting from it, in your business practices. In 2010s America, if you are a corporate leader, you’d like to understand how certain employee ownership, benefits, governance, sustainability, and CSR trends aren’t fads, but developmental trends that all the best companies must embrace as they get wealthier. If you have any influence over IT, you’d like to be doing your part to increase and cheapen broadband access, improve security, mobility, and transparency, and taking other actions to create the IT future that you know is inevitable, rather than siding with opportunistic monopolies working to slow down that access, as they try to maximize profits with old business models. If you are a law enforcement leader, you’d like to understand how your job now also should include crime prediction, crime prevention and community service. If you are a defense or intelligence agency leader, you’d like to understand how increasing global transparency, privacy and rights protection, humanitarian aid, and conflict prevention are among the inevitable new missions of all the best defense and intelligence agencies.

Let’s look at a global example in an area many foresight professionals care deeply about: world population. It’s easy to get scared about population, as until recently, it was a classic example of exponential growth. According to the American Museum of Natural History, it took 200,000 years for us to reach 1 billion, but just 200 to reach 7 billion. But since the 1960’s our marginal rate of population growth has been rapidly decreasing, as societies continue to develop. As statistician Hans Rosling of Gapminder says in Don’t Panic: The Truth About Population (2013), humanity hit Peak Child in 2000, stabilizing our global birthrate at roughly 130 million a year, due to a complex set of factors like mass electronic media showing people there are other ways to live, growing economic opportunities, women’s education, access to birth control, and the rising cost of raising a child in modern societies (children become liabilities, not assets once a certain level of technology and public health exists). Factoring out immigration, populations are now stable or declining in the great majority of countries. When I first started writing about this in 1999, I called it a “technological contraceptive,” and argued it was one of the great forces we were still not recognizing in global development policy. Wherever sufficient amounts of the right kind of technology goes, population sizes start shrinking. People shift their interests from having larger families to other pursuits. Pessimists aside, we know now that once we’ve reached peak population, our total population will start shrinking again (the “UN low” green curve below right).

So we’re presently nearing peak human population, and it’s obvious to anyone who takes a universal view that from that point forward the population of biological humans will be increasingly less important to the future of the planet. There are still people who like to think that we might keep exploding in numbers (the “UN high” red curve below right) but that’s just not going to happen. A few developed societies, worried about who will support the growing elderly, have paid women to have have children, in the form of direct payments (Australia) or greatly ramped up benefits (Germany, Japan), but this hasn’t stabilized their crashing population numbers.

These nations should be more foresighted. It’s obvious who will support our aging, shrinking population of people. It’s our technology, which will continue to accelerate in its abilities and intelligence, and continue to escape resource needs as as it dives ever further into inner space. The ratio of machine to biological minds will continue to grow, exponentially in the favor of machines. The fraction of biological minds on Earth just a century from now may be quite small, and it will be negligible a century later.

The era of biological humans controlling the planet is ending, we can see it clearly ahead, and we just don’t want to admit it yet. So we keep generating these silly estimates like the red curve at right. The red forecast is two kinds of wrong: the well-meaning kind, a self-preventing prophecy intended to scare us into getting much more aggressive about population control, as we should, and the ignorant kind, an unwillingness to see that human beings are rapidly and irreversibly being superseded by our machines, which will provide for all our basic needs just a few generations from now. It is the ignorant dimension of the red forecast that keeps us from having the dialog, and the smart investments in machine intelligence and machine productivity, and distributing the benefits of that productivity in things like a basic income, that we should be having right now. It’s obvious that machines are both our future and our “salvation”, we just are too arrogant to want to admit it.

World Population Forecasts, 2010-2100 Three UN Projections (Wikipedia, 2010)

World Population Forecasts, 2010-2100
Three UN Projections (Wikipedia, 2010)

How far ahead will our own population peak be? And what can we do to reduce it today? Rosling tells us the “pin code” for global population today is 1114. He means there are 1 billion humans in the Americas, 1 billion in Europe, 1 billion in Africa, and 4 billion in Asia. By 2100, median UN projections now expect our global population pin code to be 1145. The Americas and Europe have both stopped growing, other than through immigration, and just one billion more people will be added in Asia, which has almost achieved population stabilization. In Bangladesh, for example, there are now 2.2 births per two adults, down from 6 in the 1970s. In a worst case scenario, if present trends continue, three billion more humans may join the planet in Africa, taking us to a peak of 11 billion people by 2100. In a best case scenario, if we choose to get serious now about curbing population growth in Africa, and very secondarily in Asia, we might end up with something around 1124, with a peak of 8.3 billion people by 2050.

Rosling’s analysis makes clear that it’s really just the birth rate in underdeveloped countries in Africa, and a few underdeveloped countries in Asia, that we should be focusing on today. Even developing countries like Brazil and Iran have seen their birth rates drop. So when we talk about population reduction, we should be talking about development efforts in those two places. That’s where our moral choice lies. Anyone who cares about sustainable human development, and providing the maximum resources to all the world’s children as they increasingly seek developed world lifestyles should be looking at helping Africa and Asia today with women’s rights and education, paid jobs, access to birth control, and a range of targeted technological development activities. That’s where the problem is coming, and we can easily anticipate it today.

To solve this still-growing problem, we need opinion leaders, NGOs and corporate and political leaders to focus us on curbing birth rates in key countries in Africa and Asia today. We can’t do it in a patronizing way, either. Those countries won’t accept edicts from us, and why should they, we developed countries didn’t curb our own populations early on. Instead they need targeted development policies and opportunities which create the results we seek, in a way that is acceptable to most of the countries offered those opportunites. All of us have a chance now to lead on this predictable developmental issue. Whether we do or not will have billions of consequences in the quality of human life and our global environment for the remainder of this century.

When you’ve been lucky enough to discover anything important and developmental coming your way, you can either get in front of the parade early and take credit for its inevitable arrival, or get forced or shamed into it later. How you respond to emerging inevitabilities is your choice, but you will be more respected as a leader, and accomplish much more when you can identify (learn and anticipate) and then align your strategy, execution, and influence behind the worthiest developmental destinations, those offering the most benefit to the most stakeholders, in your company, your industry, your profession, and your personal life.

Share your Feedback

Better Wording? References? Data? Images? Quotes? Mistakes?

Thanks for helping us make the Guide the best intro to foresight on the web.