BryceTech Forecasting Retrospective Analysis and What It Says About Predicting Technology

Key Takeaways

• Bryce found forecast method and time horizon mattered more than most forecast attributes.
• Quantitative trend methods performed better for timing than opinion-based forecasting.
• The study warns against vague forecasts that lack dates, metrics, and testable events.

Bryce Tech Forecasting Retrospective Analysis and Its Defense Origin

The BryceTech forecasting retrospective analysis traces back to an August 13, 2012 final article prepared by Carie Mullins of The Tauri Group for the Office of the Secretary of Defense. The article, titled Retrospective Analysis of Technology Forecasting: In-Scope Extension, examined whether past technology forecasts could be evaluated in a systematic way and whether measurable patterns could explain why some forecasts succeeded and others failed. The document’s cover identifies the producer as The Tauri Group and notes that the team later became part of Bryce, formerly Tauri Group Space and Technology.

The defense connection matters because technology forecasting is rarely a neutral academic exercise in government settings. For the Department of Defense, forecasts influence research portfolios, acquisition plans, intelligence priorities, and the timing of investments in emerging capabilities. The 2012 study states that the Assistant Secretary of Defense for Research and Engineering wanted better tools and techniques for technological forecasting, including a performance baseline against which new forecasting methods could be compared.

The study also extended an earlier forecast-accuracy project. That earlier project analyzed 310 verified forecasts and found that quantitative methods were more accurate than other methods, forecasts about autonomous systems and computers performed better than forecasts in other technology categories, and the nine attributes studied did not produce a strong overall correlation with forecast accuracy. The 2012 project expanded the dataset to test those findings against a larger sample.

The work is best understood as an evidence test for technology foresight. It did not ask whether forecasters sounded persuasive, whether experts had impressive credentials, or whether a forecast fit a preferred strategic narrative. It asked whether a forecast had enough specificity to be checked and whether the predicted event occurred near the predicted time. That makes the study useful for defense and security planning, commercial technology strategy, and space-sector market forecasting, where long timelines and high uncertainty often encourage confident claims that remain hard to evaluate later.

BryceTech’s current public profile describes the company as an analytics and engineering firm serving science and technology clients, with work in government program support, business consulting, data analytics, and predictive models. That background helps explain why the forecast study still receives attention: it sits at the junction of defense planning, technology assessment, and empirical methods for judging claims about the future.

How the Forecast Sample Was Built

The 2012 study began with forecast documents from academia, industry, government, and other public sources. Analysts reviewed 300 documents and extracted 2,279 forecasts. Of those, 2,092 forecasts were specific, timely, complete, and relevant enough to be assessed. The team verified 1,055 forecasts using 2,016 verification documents.

A forecast had to pass a practical test before it could support accuracy analysis. It needed a technology, a predicted event, a time frame, and enough meaning to compare the claim with later evidence. Vague statements such as “robotics will become more important” are weak because almost any later development could be interpreted as confirmation. A statement such as “autonomous ground vehicles will perform a defined military logistics task by a specified year” gives evaluators a firmer basis for assessment.

The study sorted forecasts by objective attributes. Those attributes included methodology, technology area, and time frame. The report refers to nine objective attributes, and the later peer-reviewed article, A Retrospective Analysis of Technology Forecasting, describes factors such as forecast method, time horizon, technology area, prediction type, technology maturity, technology complexity, geographic origin of the forecast, geographic region forecasted about, and publication type.

This structure gave the analysis more value than a simple list of hits and misses. By coding forecasts in a comparable way, the team could test whether the source of a forecast, the technique used, the forecast horizon, or the technology category had a measurable connection to accuracy. The larger dataset also made the work more credible than anecdotal discussions that select famous successes and failures after the fact.

The method carried an unavoidable screening effect. Forecasts that lacked specificity could not be assessed, meaning that the study favored forecasts with measurable language. That was not a flaw in the ordinary sense. It was an important research boundary. A forecast that cannot be assessed later may still influence decisions, but it cannot contribute much to evidence about forecast accuracy.

What the Accuracy Findings Actually Showed

The central finding was narrow and powerful: forecast methodology and time horizon mattered more than most other coded attributes. The associated Technological Forecasting and Social Change article reports that, among the nine attributes examined, only methodology and time horizon had a statistically significant influence on forecast accuracy. Forecasts using quantitative methods were more accurate than forecasts using qualitative methods, and forecasts covering shorter horizons were more accurate than longer-range forecasts.

The report’s executive summary states that forecasts were generally more accurate than uninformed guesses. It also found that six of the eight methodologies were statistically more accurate than a theoretical random guess. Numeric trend forecasts performed better than opinion-based forecasts, short-term forecasts were more accurate than medium- and long-term forecasts, and clearly described forecasts offered more informational value.

The finding about expert judgment was more mixed than a simple “experts were wrong” reading would suggest. The peer-reviewed article found that expert sourcing produced the highest number of forecasts in which the predicted event occurred, but quantitative methods performed better on timing. That distinction is useful for technology strategy. Experts may know which technical goals are plausible, which research pathways are active, and which organizations have enough capability to pursue them. Quantitative analysis may better estimate the pace at which measurable performance, cost, production, or adoption trends are changing.

The study also found that computer, autonomous, and robotic technologies had stronger accuracy records than other technology areas. This result was observed in the dataset but not fully explained by it. A reasonable interpretation is that some technology families leave stronger quantitative trails. Semiconductors, computing systems, software performance, robotics components, and digital adoption patterns often produce measurable data series. Technology areas tied to social adoption, regulation, infrastructure, public funding, or complex physical systems may resist neat extrapolation.

The report also warned that a predictive model of forecast accuracy could not be built from the available attributes. Forecast accuracy appeared to depend on random components or attributes that the study did not capture. That result limits overinterpretation. The findings support better forecasting practice, but they do not make forecasting mechanical or risk-free.

Why Quantitative Methods Performed Better

Quantitative forecasting methods had an advantage because they forced forecasters to confront measured change. A trend line, cost curve, patent series, publication count, production rate, performance benchmark, or adoption curve imposes discipline on a prediction. The forecaster must specify what is being measured, how the measurement changes over time, and what pattern might continue or break.

In the Bryce/Tauri work, forecasts based on numeric trends outperformed opinion-heavy approaches. The report states that forecasts generated from quantitative trend analysis had statistically higher success rates than forecasts generated through other methodologies. The peer-reviewed article describes numeric trends as producing more accurate forecasts than other methods.

The result does not mean that simple extrapolation always works. It means that, in the study sample, forecasts grounded in measurable patterns performed better than forecasts grounded mainly in judgment. Extrapolation can fail when a technology hits a physical limit, loses funding, faces regulatory resistance, encounters supply-chain constraints, or becomes economically unattractive. The value of quantitative methods lies in making the assumption visible. When the assumption fails, analysts can identify what changed.

For space markets, this distinction is important. Launch price forecasts, satellite manufacturing forecasts, small-satellite demand projections, and ground-network capacity forecasts often depend on measurable industrial patterns. Launch cadence, payload mass, insurance claims, spacecraft orders, component lead times, and regulatory filings can all serve as signals. Yet a launch market forecast based only on order books may miss financing stress, customer concentration, licensing delays, or geopolitical restrictions. Quantitative data improves the forecast, but it does not remove judgment.

Modern technology-monitoring organizations often blend quantitative data with expert review. The Center for Security and Emerging Technology described a method using research clusters, publication data, and expert discussion to help track emerging technology trends. Its 2024 article, Staying Current With Emerging Technology Trends, cites the Bryce/Tauri retrospective work and states that methods combining quantitative analysis with human judgment are generally more successful.

That blended model matches the lesson from the Bryce work. Numbers provide structure. Experts provide interpretation. The weakest approach is neither fully quantitative nor well-informed by specialists. It is a loose narrative forecast with no testable event, no time frame, and no performance threshold.

Why Expert Judgment Still Had Value

The study’s finding about experts is often misunderstood. Experts were not useless. They were better at identifying which events would happen than at predicting exactly when they would occur. That pattern fits technology development because experts often understand feasibility, institutional interest, and research direction better than outsiders. Timing depends on financing, procurement, integration, policy, supply chains, and adoption.

The peer-reviewed article states that expert sourcing methods produced the highest number of forecasts in which events had been realized, indicating that experts were stronger at predicting if an event would occur and quantitative methods were stronger at predicting when.

That difference matters for strategic planning. A defense analyst may need to know whether a foreign military will deploy a capability at all. A commercial investor may need to know when revenue will materialize. A regulator may need to know when safety rules must be ready. Those are related questions, but they are not identical. A technology can be technically feasible and still arrive late, stall at demonstration, or fail to reach adoption at scale.

Expert judgment is also vulnerable to narrative pressure. Experts can overweight programs they know personally, assume continued funding, extrapolate from prototypes, or understate integration problems. Public experts may also speak in language shaped by institutional incentives. A government office may describe a technology as promising to support budget continuity. A company may describe a product as near-term to attract customers or capital. A research community may use strong language to draw attention to a field.

The Bryce approach reduces these risks by separating the claim from the claimant. It does not treat a forecast as more accurate because it came from a prestigious source. It asks whether the predicted event occurred and whether the timing was close. This approach is useful for space-sector claims about reusability, lunar infrastructure, satellite servicing, in-space manufacturing, optical communications, and orbital data centers. Each field contains serious technical work, but each also contains forecasts that mix prototype success with market timing.

Better expert forecasting would use clearer boundaries. Experts should be asked to separate technical feasibility from operational readiness, commercial adoption, and policy approval. They should give dates as ranges, attach confidence levels to claims, and identify observable milestones that would change their view. A forecast that says “technology X may be ready soon” has limited value. A forecast that identifies a flight demonstration, performance threshold, production rate, regulatory approval, or customer contract by a specified year gives decision-makers something to monitor.

Why Time Horizon Damaged Forecast Accuracy

Time horizon was the other statistically significant driver of accuracy. Shorter forecasts performed better because fewer unknowns had time to accumulate. The 2012 report states that short-term forecasts were more accurate than medium- and long-term forecasts. The peer-reviewed article also reports that forecasts with shorter horizons were more accurate than longer-horizon forecasts.

Longer forecasts face compounding uncertainty. A ten-year technology forecast must make assumptions about science, engineering, budgets, production capacity, regulation, customer behavior, competition, and macroeconomic conditions. Each assumption may look reasonable when viewed alone. The combined forecast can still fail because one weak assumption changes the path.

Energy forecasting offers a related example. A 2002 Annual Review of Environment and Resources paper on long-term United States energy forecasts found that forecasters from the 1950 to 1980 period often underestimated unmodeled surprises, including efficiency responses after the oil embargos of the 1970s. That paper focused on energy rather than defense technology, but it reinforces a similar lesson: long-range forecasts must model social systems as well as technical systems.

Space forecasting shows the same problem. A forecast about reusable launch vehicles can be grounded in engine performance and refurbishment cycles, but adoption also depends on capital availability, launch demand, insurance rules, range operations, reliability, and regulatory decisions. A forecast about satellite direct-to-device communications must account for spectrum rights, device compatibility, constellation replenishment, national regulators, telecom partnerships, and consumer behavior. A forecast about lunar resource use must separate engineering demonstration from sustained demand.

The time-horizon finding does not make long-term forecasting worthless. Long-range forecasts are useful when they frame scenarios, identify uncertainties, and guide monitoring. They become risky when users treat them as schedule promises. A good 20-year forecast should read less like a timetable and more like a map of possible pathways, dependencies, and early indicators.

This is where the Bryce study gives decision-makers a practical warning. Forecasts become more useful when they specify near-term checkpoints. A long-term claim about a 2040 capability should identify measurable signs expected by 2028, 2032, and 2036. If those signs fail to appear, the forecast should be revised rather than defended through vague reinterpretation.

How the Findings Connect to Emerging Technology Monitoring

Technology forecasting has shifted toward persistent monitoring. Instead of issuing a static forecast and revisiting it years later, analysts now track publications, patents, standards, investment flows, procurement records, company announcements, regulatory filings, and operational deployments. The National Academies article on persistent forecasting of disruptive technologies described forecasting methods and design features for systems that could help defense, homeland security, and intelligence users identify disruptive technologies.

The Institute for Defense Analyses article, Current and Potential Use of Technology Forecasting Tools in the Federal Government, examined technology forecasting tools for federal science and technology decision-making and found that tracking and summarization capabilities could help analysts sort, organize, and distill large amounts of information.

CSET’s 2024 work on emerging technology trends shows how this approach has developed. The article describes the use of a merged corpus of more than 259 million scientific publications, research clusters, and an Emerging Technology Observatory platform to support quantitative analysis, then adds facilitated expert discussion to interpret the results.

This monitoring approach addresses one weakness of single-point forecasting. A forecast made in 2012 may have been reasonable at the time, then become obsolete because of a funding change, a materials bottleneck, a new standard, or an unexpected commercial entrant. A persistent system updates the evidence base and makes forecast revision part of the process.

For policymakers, the Bryce lesson is not to replace judgment with data. It is to make judgment auditable. Analysts should record why a forecast was made, which data supported it, what assumptions mattered, and what evidence would weaken it. Forecast records should be revisited at set intervals. Failed forecasts should be retained rather than quietly deleted because they are the best training material for future methods.

For companies, the same principle supports better market planning. A satellite operator, launch provider, sensor manufacturer, or ground-network business can treat forecasts as living decision tools. Forecasting discipline improves when a company links projections to customer contracts, licensing milestones, production capacity, financing assumptions, and competitor behavior. It also improves when management asks whether earlier forecasts were accurate, why they missed, and whether the same bias appears in current planning.

Lessons for Space, Defense, and Commercial Market Forecasting

BryceTech is widely associated with space-sector analytics, but the 2012 retrospective has value beyond space. It gives any forecast user a way to interrogate claims about emerging technologies. The most useful question is not whether a forecast sounds plausible. The better question is whether it contains a testable event, a measurable performance condition, a date or time range, and a method that can be audited later.

Space and defense markets are especially exposed to weak forecasting because their technologies often require long development cycles, large capital commitments, government approval, and integration into systems of systems. A spacecraft component may work in a laboratory, fail environmental testing, pass flight qualification, reach low-rate production, and still miss commercial adoption because the end-user market changes. Treating those stages as one smooth path encourages schedule errors.

Forecast users should separate five different claims:

• Technical feasibility
• Prototype demonstration
• Operational deployment
• Commercial adoption
• Scaled economic effect

A forecast about laser communications, for example, may be accurate at the demonstration stage and wrong at the mass-adoption stage. A forecast about satellite servicing may correctly predict docking and inspection capability, then overstate the near-term market if customer willingness to pay remains limited. A forecast about in-space manufacturing may correctly identify a physics advantage, then misjudge cost, inspection, insurance, or return logistics.

The Bryce study also cautions against treating forecasts as brand assets. Companies often publish market forecasts to support fundraising, investor relations, or customer education. Government agencies publish roadmaps to align budgets and industry. Consultants publish forecasts to frame demand. None of those uses is improper, but each creates incentives. Retrospective evaluation helps distinguish a forecast designed for decision support from a forecast designed mainly for persuasion.

A mature forecast process should use numeric trends where available, expert judgment where needed, and explicit revision rules for both. It should track forecast age. It should distinguish baseline projections from scenarios. It should avoid treating a high-growth sector as though every component market will grow at the same rate. It should document the difference between total addressable market, serviceable available market, contracted revenue, and realized revenue.

The practical takeaway for the space economy is direct. Forecasts about launch demand, satellite manufacturing, Earth observation, positioning, navigation and timing, optical communications, defense space architecture, lunar services, and orbital infrastructure need evidence trails. A forecast that cannot be assessed later may still be interesting. It should not carry the same weight as a forecast that names a metric, a date, a market boundary, and a monitoring plan.

Methodological Limits and Risks of Overreading the Study

The Bryce/Tauri study has limits that matter. It evaluated forecasts that could be extracted, coded, and verified. Forecasts that were too vague, incomplete, or poorly bounded could not support accuracy analysis. That means the findings apply most directly to explicit technology forecasts with enough structure for retrospective assessment. They do not fully capture how vague forecasts influence budgets, public expectations, or strategic narratives.

The sample also depended on public or accessible documents. Private internal forecasts by companies, intelligence agencies, investors, and laboratories may differ in quality and language. Internal forecasts may use better data, but they can also carry stronger institutional bias. Public forecasts may be polished for persuasion. The study design could not fully resolve those differences.

The results should not be treated as a universal law. Quantitative trend analysis performed well in the sample, but a numeric method can still fail when the data series is short, the measurement is weak, or the trend is driven by temporary conditions. A misleading dataset can create false confidence. A model that fits the past can miss a regime change.

Technology area differences also require care. The study observed stronger performance for computers and autonomous or robotic technologies, but it did not fully explain the cause. Digital technologies may produce cleaner data and smoother improvement curves. They may also attract more forecasts from analysts familiar with quantitative methods. Without isolating those factors, the finding should guide caution rather than certainty.

The inability to build a predictive model of forecast accuracy is one of the most important results. The report states that forecast accuracy appeared to be influenced by a random component or by an attribute not captured in the study. In practice, this means forecast users should avoid treating any single scoring system as a substitute for careful review.

A forecast evaluation process should score more than outcome accuracy. It should also assess clarity, evidence quality, update discipline, scenario design, and decision relevance. A forecast can miss the exact date but still help decision-makers prepare if it correctly identifies dependencies and warning signs. Another forecast can hit a date by chance and still be poorly reasoned. Retrospective analysis should improve judgment, not create a mechanical scoreboard detached from decision value.

Building Better Forecasts After the Bryce Findings

A better technology forecast begins with a defined claim. The forecast should state the technology, the event, the performance level, the expected date or date range, and the evidence behind the estimate. It should avoid language that can be reinterpreted after the fact. It should separate what is observed from what is inferred.

The Bryce findings suggest a practical template. Start with available quantitative indicators. Add expert judgment to interpret the indicators. Identify assumptions. Define monitoring signals. Set review dates. Record changes. This discipline makes the forecast useful even when the future departs from the original estimate.

For space-sector analysts, a forecast about a launch vehicle should distinguish test flight, licensed operation, repeatable cadence, customer acceptance, and profitability. A forecast about a satellite constellation should distinguish manufacturing throughput, launch deployment, service activation, regulatory approval, subscriber adoption, and replenishment cost. A forecast about defense space capabilities should distinguish research, demonstration, procurement, operational integration, doctrine, and allied interoperability.

Decision-makers should also resist false precision. A forecast that gives a single year may look more authoritative than a forecast that gives a probability range, but the single-year estimate may hide uncertainty. A responsible forecast might say that a capability has a 60% probability of operational demonstration between 2029 and 2032, subject to funding, regulatory approval, and three named technical milestones. That phrasing is less dramatic, but it is more useful.

The Bryce work also supports forecast portfolios. Instead of relying on one forecast, organizations should maintain competing forecasts using different methods. A quantitative trend forecast, an expert elicitation forecast, a scenario forecast, and a procurement-based forecast can be compared. Agreement across methods increases confidence. Disagreement identifies assumptions worth testing.

The most valuable institutional practice may be routine retrospective review. Forecasts should be archived with their assumptions and revisited later. Misses should be analyzed without embarrassment. Good forecasting cultures treat error as data. Bad forecasting cultures quietly replace old forecasts with new narratives.

Summary

The Bryce Tech forecasting retrospective analysis remains useful because it disciplined a field often dominated by confident claims. Its strongest finding was not that the future can be predicted with certainty. Its strongest finding was that some forecast practices are more useful than others. Quantitative trend methods performed better than opinion-based approaches in the dataset. Shorter time horizons performed better than longer ones. Expert judgment helped identify events that later occurred, but data-driven methods helped more with timing.

The study also showed that forecast clarity matters. A forecast that names a technology, event, time frame, and performance condition can be evaluated. A vague forecast can influence decisions without ever facing a real test. That distinction is important for defense and security, space markets, artificial intelligence, robotics, energy systems, biotechnology, and commercial technology planning.

The 2012 report should not be read as a promise that numerical forecasting can solve uncertainty. The report itself found that a predictive model of forecast accuracy could not be developed from the attributes studied. That result is a warning against overconfidence. Better forecasting does not eliminate uncertainty. It makes assumptions visible, errors traceable, and revisions more disciplined.

For space and defense organizations, the practical standard is simple. Forecasts should be specific, measurable, time-bounded, evidence-based, and revisited. Claims about emerging technologies should separate technical feasibility from operational deployment and scaled adoption. The more expensive the decision, the less tolerance there should be for vague timing, undefined markets, and unsupported extrapolation.

Appendix: Useful Books Available on Amazon

• Superforecasting: The Art and Science of Prediction
• The Signal and the Noise
• Prediction Machines
• Thinking, Fast and Slow
• Technological Revolutions and Financial Capital
• Forecasting: Methods and Applications

Appendix: Top Questions Answered in This Article

What Was the Bryce Tech Forecasting Retrospective Analysis?

It was a retrospective evaluation of past technology forecasts produced by The Tauri Group, later associated with BryceTech. The 2012 report examined thousands of extracted forecasts and assessed more than 1,000 verified forecasts to identify patterns linked to forecast accuracy.

Why Did the Department of Defense Support This Work?

The Department of Defense has strong incentives to improve technology forecasting because research, acquisition, and intelligence decisions depend on anticipating technical change. The study helped create a baseline for comparing forecasting methods and future analytical tools.

What Was the Main Finding of the Study?

The main finding was that forecast methodology and time horizon had the strongest measurable relationship with accuracy. Quantitative methods performed better than qualitative methods, and shorter forecasts performed better than longer forecasts.

Did Experts Perform Poorly in the Bryce Analysis?

Experts were not useless. Expert sourcing had value in identifying events that later occurred, but quantitative methods performed better on timing. This suggests that expert judgment and data analysis should be combined rather than treated as substitutes.

Why Are Short-Term Forecasts More Accurate?

Short-term forecasts face fewer unknowns. Longer forecasts must account for technical progress, funding, policy, customer adoption, supply chains, regulation, and unexpected events. Each added year increases the chance that at least one assumption will fail.

What Makes a Technology Forecast Assessable?

A forecast is assessable when it identifies a technology, a predicted event, a time frame, and enough detail to verify later. Forecasts lacking dates, performance metrics, or clear event definitions are harder to evaluate and less useful for decision-making.

How Does the Study Apply to the Space Economy?

Space markets depend on long development cycles, regulation, capital access, and government demand. Forecasts about launch, satellites, lunar services, optical communications, or in-space infrastructure should separate technical demonstration from operational use and scaled market adoption.

Why Did Quantitative Forecasts Perform Better?

Quantitative forecasts require measurable indicators, such as performance trends, cost curves, production rates, publication counts, or adoption data. These indicators make assumptions visible and create a stronger basis for checking whether the forecast remains plausible.

Can the Bryce Findings Predict Which Forecasts Will Succeed?

The study did not produce a complete predictive model of forecast accuracy. It identified patterns associated with better accuracy, but it also found that unexplained or random components affected outcomes.

What Is the Best Practical Lesson From the Study?

The best lesson is to make forecasts testable. A useful forecast should state the claim, evidence, timing, assumptions, and revision triggers. Organizations should preserve old forecasts and review them later to improve future judgment.

Appendix: Glossary of Key Terms

Assessable Forecast

An assessable forecast is a prediction specific enough to be evaluated later. It identifies a technology, event, time frame, and relevant performance condition. Without these elements, analysts cannot reliably determine whether the forecast succeeded or failed.

BryceTech

BryceTech is an analytics and engineering firm that works with science and technology clients. Its public work includes space-sector analysis, government program support, data analytics, and forecasting-related services.

Expert Sourcing

Expert sourcing uses knowledgeable specialists as the source of forecast judgments. In the Bryce/Tauri research, this approach had value for identifying events that later occurred, but it performed less well on timing than quantitative trend methods.

Forecast Accuracy

Forecast accuracy measures how closely a prediction matches later events. In technology forecasting, it usually requires checking whether the predicted event occurred and whether it occurred near the predicted time.

Forecast Methodology

Forecast methodology refers to the method used to create a forecast. Examples include quantitative trend analysis, expert judgment, scenario methods, source analysis, and models. The Bryce/Tauri work found methodology had a statistically significant connection to accuracy.

Forecast Time Horizon

Forecast time horizon is the length of time between the forecast date and the predicted event date. Short horizons usually produce better accuracy because fewer assumptions have time to change.

Persistent Forecasting

Persistent forecasting is an ongoing approach to tracking technologies over time. It uses repeated monitoring rather than one-time predictions, allowing analysts to update judgments as new evidence appears.

Quantitative Trend Analysis

Quantitative trend analysis uses numerical data to estimate future change. Relevant data can include cost, performance, production, publication, patent, investment, or adoption trends.

Retrospective Analysis

Retrospective analysis evaluates past claims after outcomes are known. In forecasting, it helps measure which methods, time horizons, or source types produced more reliable predictions.

Technology Forecasting

Technology forecasting estimates the timing, performance, diffusion, or operational use of technologies. It supports decisions in research funding, procurement, commercial strategy, regulation, and risk management.