How to speed up scientific and technological progress

How Humanity Could Speed Up Scientific and Technological Progress Toward Abundance

This page serves three goals. The first is futurism: asking what the next century could look like if discovery ran faster. The second is abundance and post-scarcity, the idea that food, energy, housing, and medicine could become so inexpensive to produce that rationing stops making sense, as discussed on Post-scarcity economy. The third is longevity: faster progress on aging research and on the basics of life, from protein structure to the genome, moves the date when longevity escape velocity might become realistic.

The page began as a single question: how could this theoretically possibly be done? Its answer is a thesis. Progress is a production process. It has inputs (people, money, tools, knowledge, and permission) and an output (new ideas that work). Each input can be deliberately increased or made more productive. Abundance therefore depends on treating the rate of progress itself as something to engineer, rather than as weather that simply happens to a civilization.

The problem: ideas are getting harder to find

Progress is not automatic. In "Are Ideas Getting Harder to Find?" (American Economic Review, 2020), economists Nicholas Bloom, Charles Jones, John Van Reenen, and Michael Webb measured research productivity, the output of new ideas per researcher. Their headline case is Moore's law. Keeping chip density doubling every two years now takes more than 18 times as many researchers as it did in the early 1970s. Research productivity for seed yields in corn, soybeans, cotton, and wheat falls at about 5 percent per year, and the gains in years of life saved from cancer and heart disease research show a similar decline. For the United States economy as a whole, research productivity has fallen by a factor of 41 since the 1930s.

The implication is stark. Growth has held steady only because societies kept pouring in more researchers. If that input stops growing, progress slows. If the input grows and each researcher also becomes more productive, progress can accelerate.

Lever one: people

The most direct input is the number of capable people working on hard problems. A study by Alexander Bell, Raj Chetty, Xavier Jaravel, Neviana Petkova, and John Van Reenen linked patent records for 1.2 million inventors to tax records. Children from families in the top 1 percent of income were ten times as likely to become inventors as children from below-median families, and the gap persisted among children with similar early math scores. The authors call the missing inventors "lost Einsteins." If that waste exists inside the wealthiest country, it is plausibly far larger across lower-income countries, where billions of people have little contact with laboratories, mentors, or research funding. Exposure, scholarships, remote collaboration, and open teaching materials are inputs that can be expanded.

A third people lever links directly to longevity. A researcher who stays healthy and sharp for an additional decade adds a decade of accumulated skill to the workforce at almost no training cost. Slowing aging is a goal of science and also an accelerator of science, a point developed on Classifying aging as a treatable condition.

Lever two: funding experiments

How money is allocated may matter as much as how much money exists. Several experiments illustrate the range.

• Fast Grants. In April 2020, a small team launched an emergency program for COVID-19 science with an application that took under 30 minutes and decisions within 48 hours. Over 2020 it raised more than 50 million dollars and made more than 260 grants. In a survey of recipients, 64 percent said the work would not have happened without the grant, and 57 percent said they spend more than a quarter of their time on grant applications. The organizers contrast this with standard federal review, which they report typically takes 200 to 600 days.
• Prizes. In 1714 the British Parliament offered up to 20,000 pounds for a practical method of finding longitude at sea, a sum that Royal Museums Greenwich estimates at more than 1.5 million pounds today. A self-taught clockmaker, John Harrison, eventually answered with marine timekeepers. Modern competitions such as the XPRIZE use the same logic: pay for results and let anyone compete.
• Focused research organizations. Convergent Research launches time-bound nonprofit teams, each built to produce one public good such as a tool or open dataset that unblocks a whole field and that neither a university lab nor a startup would naturally build.
• Funding people instead of projects. Pierre Azoulay, Joshua Graff Zivin, and Gustavo Manso compared Howard Hughes Medical Institute investigators, who receive long, flexible support that tolerates early failure, with similarly accomplished scientists on shorter, deliverable-based grants. The HHMI investigators produced high-impact papers at a much higher rate and moved into more novel lines of inquiry.

None of these models is a universal answer. The lesson is that funding design is itself a research question, and a portfolio of mechanisms can be tested and improved like any other technology.

Lever three: better tools

Tools raise the output of every researcher at once. The clearest recent case is protein structure prediction. Determining a single structure once required months or years of laboratory work. The 2024 Nobel Prize in Chemistry recognized the AlphaFold2 model, which has predicted the structure of virtually all 200 million proteins researchers have identified and has been used by more than two million people in 190 countries. Artificial intelligence applied to science, combined with lab automation and self-driving labs that plan and run experiments around the clock, could compress years of trial and error.

Falling input costs work the same way. The National Human Genome Research Institute estimates that sequencing a human-sized genome cost about 95 million dollars in 2001 and about 525 dollars by 2022, a decline of more than 180,000 fold. Computing and energy follow similar curves, as described on Energy costs and the path to abundance. When a measurement becomes 180,000 times less expensive, experiments that were unthinkable become routine.

Lever four: open knowledge

Knowledge that is slow to circulate, locked away, or lost has to be rediscovered. Preprint servers such as arXiv, bioRxiv, and medRxiv let results circulate in days. In the first ten months of the COVID-19 pandemic, more than 30,000 of roughly 125,000 COVID-19 papers appeared on preprint servers, against a typical six-month journal timeline in the life sciences. Open data, shared negative results, and permanent archives reduce duplicated and lost work. The threat of disappearing sources is covered on Link rot and the scarcity of reliable knowledge, permanent storage on Permaweb basics, and civilizational backups on Backing up civilization.

Speed limits we chose

Some delays are physical. Others are choices: environmental permitting that can take years for a transmission line, drug development that commonly spans a decade or more, and licensing rules that restrict who may practice, discussed on Occupational licensing as artificial scarcity. These rules protect real interests, so the goal is to reduce delay per unit of safety, not to remove all rules.

Operation Warp Speed shows the distinction. The Government Accountability Office found that vaccine development generally followed traditional practices: companies still gathered initial safety and immune response data in small groups before large phase 3 trials. The time savings came from running steps in parallel, overlapping trial phases with each other and with animal studies, and starting large-scale manufacturing during the trials. Parallel processing, accepted financial risk, and fast decisions saved months without skipping the safety evidence.

What one person or small group can do

The production view applies at small scale too.

• Pick a neglected problem where one more person matters, using tools such as Research brainstorming and Areas of research and study.
• Work in the open: publish notes, code, and drafts early.
• Build a tool, dataset, or tutorial that saves other researchers time.
• Share data, including failed experiments, in a durable archive.
• Write, teach, and mentor, especially people with no existing route into research, a form of Opportunity engineering.

Honest risks and limits

Faster progress is not automatically good progress. Dual-use research in biology and AI can enable harm as quickly as benefit, so safety capacity has to scale with speed. Gains can be distributed unevenly, flowing first to wealthy regions and owners of key tools. Hype can misdirect funding toward fashionable fields. Measurement is hard: patents, papers, and citations are imperfect proxies for useful ideas, so claims that a reform sped up science deserve skepticism until outcomes are measured.

A loose field called progress studies has formed around these questions, drawing on economics, history, and the science of science. Its central question matches this page: why progress happens, and how to deliberately produce more of it.

Discussion questions, essay ideas, and learning related AI prompt ideas

1. If research productivity keeps falling at about 5 percent per year, how many additional researchers would be needed each decade to keep growth steady?
2. Which input to the progress production function (people, money, tools, knowledge, or permission) is most underused today, and what evidence supports that view?
3. Essay idea: compare the Longitude Prize, Fast Grants, and Howard Hughes Medical Institute funding as three different theories of how to buy discovery.
4. Should regulators measure delay as a cost in the same way they measure risk? Use Operation Warp Speed as a case.
5. Ask an AI system to explain the idea production function from the Bloom, Jones, Van Reenen, and Webb paper in plain language with a numerical example.
6. Ask an AI system to list neglected research problems related to aging biology and rank them by the cost of the tools needed to start.
7. Ask an AI system to design a small experiment comparing two grant review methods, including how outcomes would be measured.

See also

• Post-scarcity economy
• Longevity escape velocity
• Time prices and measuring abundance
• Asteroid mining and why water comes first
• Lab-grown diamond in manufacturing, computing, and medicine
• Sciences
• Research