Technology Delphi
- Introduced by
- Helmer & Dalkey (RAND); national applications by NISTEP (Japan) and Cuhls (Germany)
- Year
- 1975
- Type
- Iterative structured expert-survey process
- Family
- Expert elicitation and forecasting
What does it do?
The technology Delphi is a large-scale, multi-round expert survey used to forecast the timing, importance, and feasibility of future technological developments.
Built on the classic Delphi principles of anonymity, iteration, controlled feedback, and statistical aggregation, it elicits judgements from hundreds or thousands of experts on a structured list of technology statements and converges them, round by round, into a collective forecast that informs national and organisational science and technology priorities.
The intuition behind the method
Suppose you want to know when a particular technology—say, affordable grid-scale energy storage—will become widely available, but no model can tell you and any single expert might be wildly off. The Delphi idea is to ask many experts independently and anonymously, show each of them how the group answered, and let them reconsider over several rounds. Anonymity strips away the dominance of loud personalities; feedback lets people learn from one another; iteration lets opinions converge or reveal genuine, lasting disagreement. The large-scale technology version scales this up to a whole nation's research community, producing a panoramic forecast of dozens or hundreds of technologies at once.
History
The Delphi method was developed at the RAND Corporation in the 1950s by Olaf Helmer, Norman Dalkey, and colleagues for technological and military forecasting, and was codified in Linstone and Turoff's 1975 handbook. Its most influential application became the large-scale national technology Delphi: Japan's NISTEP has conducted comprehensive technology Delphi surveys roughly every five years since 1971, covering hundreds of technology topics. Germany adapted the Japanese model in the 1990s Delphi '93 and '98 exercises, and Kerstin Cuhls's 2003 account documents how these large participative Delphi surveys evolved from pure forecasting into broader foresight processes embedded in research policy.
When to use it
Use a technology Delphi when you need structured, defensible forecasts of many technological developments, the relevant knowledge is dispersed across a large expert community, and face-to-face meetings would be impractical or distorted by group dynamics. It suits national or sectoral foresight programmes that must set research priorities across a broad technology landscape. The method assumes that aggregated, iterated expert judgement outperforms individual or unstructured group judgement, that anonymity and controlled feedback improve estimates, and that experts can meaningfully estimate timing and importance. It is less appropriate when expertise is thin, when the question is narrow enough for a quantitative model, when rapid answers are needed, or when the appearance of consensus might mask genuine and important disagreement.
Limitations and considerations
- Results are only as good as the panel and the statements; poorly framed statements or unrepresentative experts bias the whole forecast.
- The pressure toward consensus can artificially suppress valid minority views and overstate agreement.
- Large multi-round surveys are slow and expensive, and respondent fatigue causes attrition across rounds.
- Forecasts of realisation dates have a mixed accuracy record, and the method offers no causal mechanism behind its estimates.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Pools the judgement of very large, geographically dispersed expert communities that could never meet face to face.
- Anonymity and controlled feedback suppress dominance, status, and bandwagon effects that bias ordinary group forecasting.
- Iteration with statistical feedback lets opinions converge while also revealing where genuine, durable disagreement remains.
- Produces a structured, quantifiable forecast across many technologies at once, ideal for broad national priority-setting.
Applications
- National technology foresight programmes (e.g. Japan's NISTEP and Germany's Delphi exercises) forecasting hundreds of technologies.
- Sectoral foresight in energy, health, ICT, and manufacturing to identify priority research areas.
- Estimating the likely timing and feasibility of emerging technologies for R&D portfolio decisions.
- Eliciting expert consensus on policy options where empirical data are scarce.
- Supporting roadmapping by dating key technological milestones along a development path.
- Identifying obstacles, leading countries, and required actions for strategically important technologies.
How it works
How it works
Design the set of forecast statements and the questionnaire. Expert sub-committees draft concise statements describing future technological developments, and for each the survey asks structured questions—expected time of realisation, importance, feasibility, leading countries, and obstacles—so that responses can be aggregated quantitatively.
Recruit a large, knowledgeable, and suitably diverse expert panel. Unlike small expert groups, a technology Delphi enlists hundreds or thousands of specialists spanning the relevant fields, often selected through nomination chains, so the aggregated judgement reflects broad disciplinary coverage rather than a narrow committee.
Administer the first round anonymously. Experts answer the questionnaire independently and without knowing one another's identities or responses, which removes social pressure, status effects, and bandwagon dynamics that distort face-to-face group judgement.
Feed back the aggregated results and re-survey in subsequent rounds. Each expert receives statistical summaries of the group's responses—medians, ranges, and sometimes minority arguments—and reconsiders their own answer, repeating until opinions stabilise, so that controlled feedback drives convergence or clarifies persistent disagreement.
Aggregate the final responses into the forecast. Results are summarised statistically—median realisation dates, importance and feasibility rankings, and measures of consensus and dispersion—and analysed to identify priority technologies, expected timelines, and areas of expert disagreement for use in policy and strategy.
Common pitfalls
- Equating convergence with correctness, when apparent consensus may simply reflect anchoring on the group median.
- Drafting ambiguous or compound technology statements that experts interpret differently, corrupting the aggregation.
- Selecting a panel skewed toward one discipline or institution, so the 'collective' forecast reflects a narrow worldview.
- Letting round-to-round attrition go unmanaged, so later rounds rest on a shrunken and possibly self-selected sample.
Frequently asked questions
How is the large-scale technology Delphi different from a generic Delphi study?
A generic Delphi typically uses a small panel (often a few dozen experts) to reach consensus on a focused question in fields from healthcare to management. The large-scale technology Delphi is distinguished by its application: it is a national- or sector-scale technology-forecasting instrument enlisting hundreds or thousands of experts across an entire research landscape to date and rank a broad inventory of future technologies—exemplified by Japan's NISTEP and the German Delphi surveys—and is run as a flagship component of foresight and research policy rather than a standalone consensus exercise.
What are the four defining features of the Delphi technique?
Anonymity (experts respond without knowing one another's identities, removing social and status pressures), iteration (the survey runs over multiple rounds), controlled feedback (between rounds each expert receives statistical summaries of the group's responses and reconsiders), and statistical aggregation of the group response (results are summarised quantitatively, e.g. medians and interquartile ranges, rather than forced into unanimity). Together these distinguish Delphi from ordinary committee deliberation.
How accurate are technology Delphi forecasts?
Their record is mixed. Reviews of past national exercises show some technology realisation dates were broadly on target while others were significantly too optimistic or pessimistic, and the method cannot anticipate genuine discontinuities. Because of this, modern foresight treats the Delphi less as a precise prediction tool than as a structured way to mobilise expert knowledge, set priorities, and surface where experts agree and disagree—value that does not depend solely on hitting exact dates.
Related methods
Sources
Linstone, H. A., & Turoff, M. (Eds.). (1975). The Delphi Method: Techniques and Applications. Addison-Wesley.
ISBN: 9780201042948Copied.Cuhls, K. (2003). From forecasting to foresight processes—new participative foresight activities in Germany. Journal of Forecasting, 22(2-3), 93-111.
Open DOI referenceCopied.