Backcasting

​

What is Backcasting?

Backcasting is the practice of choosing a desired future end-point, then working backward to the present to test whether that end-point is feasible and what would have to change.

The major distinguishing characteristic of backcasting analyses is a concern, not with what futures are likely to happen, but with how desirable futures can be attained.

John Robinson wrote that in 1990, in the journal Futures. The end-point comes from a judgment about what is worth wanting. The walk back asks whether the physical world can support that picture, and which decisions would have to differ from the path already underway. The everyday picture is a meal you have already promised for six o’clock. You start from the table, and from how many people must be fed. If one burner cannot feed twelve, the evening is not feasible yet. You change the menu or the guest list before you buy more chairs. Scenario Thinking builds several plausible futures, often by reasoning forward from today, so a plan can survive more than one of them. Backcasting keeps an end-point that answers a stated problem, and it asks what must differ if that end-point is to be reached. Inversion Thinking walks backward from failure, so you can stay off the path that ruins the plan. Backcasting walks backward from a future you would accept.

​

Backcasting in 3 Depths

• Beginner: The cue is a promise with a clock, such as a paper due on Friday or dinner at six. Start from that moment, and name the one thing that must already be true an hour earlier.
• Practitioner: Write the end-point, write today’s system, and find one clash. Change the end-point or the path before you add a second goal. A second goal stacked on an impossible first one only lengthens the poster.
• Advanced: The end-point records what you are willing to want. Robinson expected useful backcasts to spread apart, so different goals can be compared. An early step that pays soon, and cannot serve the later goal, is a different bet from a step that stays usable later.

​

Origin

Backcasting enters the public record as a way to study energy futures when a forecast built on the current trend cannot show a break in that trend. In December 1982, John Bridger Robinson, at the University of Waterloo, published “Energy backcasting: A proposed method of policy analysis” in Energy Policy (volume 10, pages 337–344). The abstract says that arguments for soft energy paths were hard to illustrate with conventional forecasts. Those paths would not simply extend the supply trend of the time. The analyses used a method called energy backcasting: set policy goals, then determine how the goals could be met. In that article, Robinson proposed a specific form of the method. In 1996, Karl H. Dreborg, writing in Futures, said that Robinson introduced the term. He adds that Robinson’s 1982 paper credits the idea to Amory Lovins. Dreborg also describes a related Swedish line of energy studies from the 1970s, including work by Thomas B. Johansson and Peter Steen. He treats that line as close to Robinson’s account, with some differences in method. Robinson’s 1990 paper, “Futures under glass: A recipe for people who hate to predict,” is the fuller account, published in Futures (volume 22, pages 820–842). He was still at Waterloo, in the Department of Environment and Resource Studies, and the Swedish Energy Research Commission supported the work. He proposes the method for changes over 20 to 100 years. End-points, he writes, are usually set 25 to 50 years ahead, so that machines and habits have time to turn over. The paper sets out six steps, running from the purpose and the goals, through a description of the present system, to scenarios, and then to impacts. The scenario is checked for physical, technological, and socioeconomic feasibility. Where the picture contradicts itself, or where the impacts are severe, the scenario is revised. Robinson writes that a likelihood cannot be estimated inside the analysis. That likelihood would depend on whether the proposals were carried out. Dreborg’s 1996 paper asks when the method earns its place, and he lists five conditions. The problem cuts across many parts of society, and marginal fixes inside the current order will not do. The dominant trends are part of the problem, and they are often the base of a forecast. Much of the harm is an externality, a cost the price does not carry. The horizon is long enough to leave room for a deliberate choice. He also places the merit of a backcast in what philosophers call the context of discovery. The work should widen the options people are willing to discuss. Arrival of the end-point is a separate claim, and it needs its own evidence. In 2000, John Holmberg and Karl-Henrik Robèrt described backcasting as a planning frame for sustainability, developed with The Natural Step, a non-governmental organization. They argue for a small set of sustainability principles that do not cover the same ground twice. They recommend searching for early investments that combine two qualities. One is enough technical flexibility to support later steps that still meet those principles. The other is a return soon enough that someone will choose the step. In the same year, Mattias Höjer and Lars-Göran Mattsson, in Futures, argued that backcasting and forecasting are complements. Backcasting, on their account, fits where current trends lead toward an unfavourable state, and forecasting is what shows that this is the case. Later practice made the walk more participatory. Jaco Quist describes the Dutch Sustainable Technology Development programme, from early 1993 to 2002, as a setting where stakeholders helped build the end-point and the follow-up agenda. In 2024, the UK Government Office for Science, with SAMI Consulting, published a Futures Toolkit. Its backcasting workshop takes about 4 to 5 hours across two sessions, with about 8 to 16 people, and seven steps from the preferred future to an action plan.

​

Key Points

Backcasting pays when the future you want would require a break in the trend, and when you can test that future against the world as it is. It stalls when the end-point is a slogan, or when the trend is already a future you would accept.

​

Applications

Use Backcasting when a forecast, or a habit, keeps reproducing a future you have reason to reject. Write one end-point, one clash with today, and one owner for the next step.

​

Case Study

A documented public use of Backcasting is the Novel Protein Foods study inside the Dutch Sustainable Technology Development programme, which ran from early 1993 to 2002. It took a factor-20 challenge: look for technologies that might cut an environmental burden about twentyfold, across needs such as nutrition, water, mobility, and housing. The protein study was cofinanced by major Dutch food companies and ran from 1993 to 1996. In the early 1990s the Netherlands had a large intensive livestock sector. Manure, energy use, and feed crops grown abroad added to the load. So did the weak conversion of plant protein into meat protein. Quist lists the effects as acidification, climate change, toxic effects in water, and nutrient overload in soils and surface water. Meat was also a major source of protein, and of taste, habit, and status. Firms in the meat business had built alliances with a strong say over policy and research, and meat alternatives were off the food-innovation agenda. Quist, citing Aiking and colleagues (2006), notes that a major Dutch food firm had tried soy-based texturized vegetable protein in the late 1960s, that the attempt failed, and that industry interest cooled. The question was whether a different end-point could be specified, and whether anyone would carry it. The programme’s idea was a meat-like food from non-animal protein. It would need to be close enough to high-quality meat that people would buy it in volume, and its environmental burden would need to be much lower. A small group from business and research was consulted first, because resistance from the livestock sector was expected. A consulting firm then named the category novel protein foods. It set out 20 candidate pairings of a protein source and a process. The fuller study set two demands: the foods should show a factor-20 environmental improvement against pork as it was then, and they should be attractive to consumers and to producers. Seven research groups, from universities and institutes, did the work, and funders sat on an advisory board. A societal panel met three times, each time for a day and a half. The elaborated vision said that these foods, if they met the factor-20 bar, could replace 40 percent of Dutch meat consumption in 2035, with a 5 percent substitution by 2005. Those shares were the picture the team chose to walk back from. They were a chosen end-point, and the study does not report them as food people later bought. Life-cycle analysis, a tally of environmental burden across production, compared seven options with pork. The comparison assumed 1995 production and ordinary crop growing. Pork was indexed at 100, and the seven totals ran from 3.1, for a spirulina option, to 21.4, for an option made from modified lupine. A factor-20 cut would sit near 5 on that index. The spirulina option cleared that bar, and the lupine option did not. Nutrient loading, aquatic eco-toxicity, acidification, and global warming made up the main themes in the score. A separate pathway table then drew one possible route toward the vision. Its 2035 column pairs a 40 percent share for the new foods with a 60 percent share for meat, and a 55 percent lower environmental strain for the two together. Those cells are a projection drawn inside the study, and they record the path the team wrote down. The same analysis found a gap on the plate, even though most of the production techniques already existed in ordinary food processing. The technological research still found that the team could not yet offer products attractive enough to bring a significant drop in meat consumption. Consumer research said that taste and convenience would decide a purchase in 2035. The environmental index and the shopping basket did not clear the same bar. Ten years after the 1996 close, Quist and colleagues looked for follow-up, and the 2013 handbook chapter reports that search, which he dates to his 2007 thesis and to a 2011 paper. The trace is clusters of activity, beginning with a multidisciplinary programme called Profetas, which brought research groups together with several large food companies. New development work included a meat alternative from dairy proteins, introduced by a major Dutch dairy firm. Smaller firms that already sold vegetarian protein foods extended their lines. The environment ministry took the topic into policy on sustainable consumption. The agriculture ministry took it into policy on supply chains. Environmental organizations became more positive about meat alternatives. Quist calls that follow-up quite successful at the level of those niches. He also reports that the wider programme did not get significant follow-up from every backcasting study it ran. Backcasting explains the order of this project. The end-point broke with the livestock trend, the walk back produced a comparison and a list of missing knowledge, and the agenda named owners. The boundary is the one he states in general. A study of this participatory kind may lead to follow-up, and it does so only in some studies. The index from 3.1 to 21.4 scores options against 1995 pork. A later change in what the country ate would need a different count.

​

Boundaries and Failure Modes

Backcasting fails when the trend is already acceptable, when the end-point cannot be tested, and when a workshop is treated as the change. Höjer and Mattsson’s condition is the first boundary. The method is mainly apt where current trends lead toward a state you have reason to reject. If a forecast already lands where you are willing to go, a dramatic end-point adds a goal nobody needed. Dreborg’s horizon matters in the same place. Robinson’s usual end-point sits 25 to 50 years out, so that capital and habits can turn over. A choice that must be made on Thursday gains nothing from a decorative end-point in 2050. The second boundary is the feasibility test. Robinson’s method revises the scenario when physical flows contradict the goal, or when social and economic impacts are severe. A sheet that says “forty percent,” and never asks who can make the food or whether anyone will buy it, skips that revision. In the protein study, the environmental index and the consumer product did not clear the same bar. An end-point that survives only by ignoring one of those tests is unfinished. The common misuse is to read the backcast as a forecast run in reverse, and then to treat the drawing as a result. Robinson’s product is a feasibility reading and a set of implications. On his account, likelihood sits outside the analysis, because it depends on whether anyone carries out the proposals. Quist found niche activity after the protein study, and he found other studies in the same programme with no significant follow-up. The 40 percent share and the 55 percent strain figure were drawings of a path. A later national count would be a different document.

​

Common Misconceptions

Three confusions travel with the name. A backcast gets mistaken for a prediction, any backward look gets mistaken for the method, and a vivid target gets mistaken for a result.

​

Related Concepts

These neighboring pages cover plausible futures, the backward look from failure, the long horizon, the choice of where to play, later effects, and the links inside a system.

​

One-Line Takeaway