Hope & Despair

Question 09 · Knowledge

Who shapes science now?

Science is no longer a small Western archive. It is a ten-million-work-a-year machine, increasingly open, increasingly Asian, increasingly collaborative, and still brutally unequal in who gets seen.

The lazy story is that science publishes more papers every year. True, but thin. OpenAlex lets us ask the better question: who is connected to the graph of knowledge, where are they, what fields dominate, what is open, and who gets attention?

Read that way, the answer is not one mood. The hopeful version is real: knowledge is more open, less Western, and more collaborative than it was in 2000. The despairing version is also real: visibility still concentrates around rich institutions, old prestige systems, and fields with money, labs, English, and citation machinery behind them.

Before
you read
  • OpenAlex counts scholarly works, not only journal articles. Its graph includes articles, books, proceedings, preprints, repositories, and other research objects where metadata exists. That breadth is the point, but it is broader than older journal-article statistics.
  • Country counts are country-attributed, not nationally exclusive. A paper with authors in China and Germany can count for both. That is correct for collaboration, but it means country bars are not slices of one pie.
  • Citations are an influence proxy, not a truth meter. Citation-normalized top-10% status is useful because it adjusts for field and year, but citations still follow language, prestige, network, and topic fashions.
  • Everything OpenAlex-derived here is snapshotted through the official API under CC0. The article uses grouped aggregates, not live requests at page load.
Fig. 1
The machine

How large is the visible research system?

OpenAlex records about 10.6 million scholarly works published in 2024, up from under 100,000 in 1900.

Start with the scale, because it changes the meaning of every other chart. In 1900, OpenAlex records about 92,000 scholarly works. By 2000 it records 3.8 million. By 2024, 10.6 million. Some of that is better indexing and broader metadata. But the broad fact is not subtle: science became an industrial-scale human activity.

That is hopeful if your problem needs more eyes, instruments, languages, and local knowledge. It is terrifying if your problem is judgment. The scarce resource is no longer a place to publish. It is attention, synthesis, replication, and knowing which claims actually changed the map.

Scholarly works published per year · World · OpenAlex records1900 - 2024
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Scholarly works published per year, World

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-12 checksum · 115405cc90d5…
Hope

More people, institutions, and fields can now contribute to the formal record of knowledge.

Despair

A literature this large can bury signal under output, incentives, and unreadable abundance.

Confusion

The count mixes real growth with better indexing; it measures the visible system, not pure discovery.

Fig. 2
The geography

Where is new science being produced now?

In 2024, China and the United States are nearly tied in OpenAlex country-attributed works.

This is the first big map change. In OpenAlex’s 2024 country grouping, China appears on about 1.28 million works and the United States on about 1.21 million. India and Indonesia follow at about 450,000 each, with the United Kingdom, Germany, France, Brazil, Italy, Japan, Canada, and Spain behind them.

Do not read this as a quality ranking. It is a presence ranking. Large countries, large university systems, publication incentives, and collaboration all matter. But presence matters. The old mental picture of science as a North Atlantic archive with everyone else reading along is obsolete.

Works by author-institution country · 2024 · millioncountry-attributed, not exclusive
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Works published in 2024, grouped by countries attached to authorship institutions. A multi-country paper can count once for each country represented.

Calculation. Group 2024 works by author institution country; show the top 12 in millions.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · ffa678c52eaa…
Hope

The production of formal knowledge is no longer monopolized by old Western research powers.

Despair

Raw output rewards scale, incentives, and indexing; it does not tell us whose work shapes the field.

Confusion

A collaborative paper counts for every represented country, so the bars show footprint, not ownership.

Fig. 3
The geography

How fast did the map move?

China’s country-attributed OpenAlex output rose about 23.6x from 2000 to 2024; India’s rose about 19.2x.

The 2024 bar is a snapshot. This is the motion behind it. Put each country’s OpenAlex footprint on a 2000 = 1x scale and the shift becomes obvious: China reaches about 23.6x by 2024, India 19.2x, and Brazil 9.4x. The United States, Germany, and Japan remain large research systems, but their visible output grew much more slowly from already large bases.

This is why the map feels different. The United States did not vanish; it grew. But China and India grew much faster, enough to change the center of gravity of the visible research system inside one generation.

Country-attributed OpenAlex works · multiple of 2000 output2000 - 2024
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. China country-attributed OpenAlex works as a multiple of its 2000 value.

Calculation. Divided China yearly country-attributed work count by its 2000 value.

The displayed line is monotone-cubic smoothed for legibility only — it never moves a data point.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · 671fcc2f14b0…
Hope

A much larger share of humanity now has institutions visible in the formal research graph.

Despair

Growth multiples can make smaller bases look explosive; they say nothing by themselves about influence or quality.

Confusion

This is country footprint, not exclusive national production; collaborations lift more than one line.

Fig. 4
The geography

Who gained share of the whole graph?

China’s share of all OpenAlex works rose from about 1.4% in 2000 to 12.0% in 2024.

The share chart asks a stricter question than the multiple: not just who grew, but who grew faster than the whole research system. China rises from about 1.4 percent of OpenAlex works in 2000 to about 12.0 percent in 2024. India rises from about 0.6 percent to 4.3 percent. Brazil rises from about 0.7 percent to 2.2 percent. The United States remains huge, the United Kingdom and Germany are roughly stable, and Japan’s share falls as the rest of the system grows around it.

This is the better chart for 'who gained ground.' The story is not American collapse. It is the end of an old monopoly.

Country-attributed works as share of all OpenAlex works2000 - 2024
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. China country-attributed works as a share of all OpenAlex works in the same publication year.

Calculation. Divided China yearly country-attributed work count by all OpenAlex works in the same year.

The displayed line is monotone-cubic smoothed for legibility only — it never moves a data point.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · 80a6809e9a79…
Hope

The formal record of science is less concentrated in one old center than it was in 2000.

Despair

A larger share of works is not the same as a larger share of funding, prestige, or agenda-setting power.

Confusion

Fast-rising shares can reflect real capacity, indexing changes, publication incentives, or all three.

Fig. 5
The fields

How did the field mix change?

Medicine remains the largest selected field, but computer science grew fastest: about fivefold since 2000.

A 2024 field ranking tells us what is large now. The time series tells us what changed. Medicine grows from about 0.56 million works in 2000 to 1.63 million in 2024. Social sciences rise from 0.44 million to 1.44 million. Engineering starts high and grows more slowly, from 0.76 million to 1.43 million. Computer science rises from 0.14 million to 0.71 million.

That shape is the modern research agenda in motion: health remains enormous; society and environment expand; computing becomes a general-purpose research language rather than one field among others.

Works by OpenAlex field · million per year2000 - 2024
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. OpenAlex works per year whose primary topic field is Medicine.

Calculation. Filtered works to primary topic field Medicine, grouped by publication year, 2000-2024, shown in millions.

The displayed line is monotone-cubic smoothed for legibility only — it never moves a data point.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · b44ff3c417c0…
Hope

The system is broad enough to work on disease, engineering, environment, society, culture, and computation at once.

Despair

Fields with money, institutions, and publication volume can dominate the visible archive and the public imagination.

Confusion

Field labels simplify messy interdisciplinary work; the same paper can belong intellectually to more than one frontier.

Fig. 6
The fields

Which fields accelerated most?

Among these large fields, computer science output is about five times its 2000 level.

Normalize each field to its own 2000 level and the hierarchy changes. Computer science is the clear accelerator, roughly five times its 2000 output by 2024. Environmental science and social sciences more than triple. Medicine almost triples. Engineering, already large in 2000, roughly doubles.

This is why the field story needs a transformation. Raw counts tell you what is biggest. Indexed growth tells you what is changing fastest.

Field output growth · 2024 as multiple of 2000selected OpenAlex fields
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Selected OpenAlex fields, with 2024 output divided by 2000 output.

Calculation. Computed 2024/2000 output multiple for selected large fields.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · dbefd69e3e75…
Hope

New fields and cross-cutting tools can scale quickly through the research system.

Despair

Acceleration can reflect publication incentives and hype as much as genuine discovery.

Confusion

A fast-growing field may still be smaller than a slower-growing field; growth and size answer different questions.

Fig. 7
The opening

Where is research most open to read?

In large fields, open-access shares now commonly sit above 60%.

The access story is more interesting by field than in total. In 2024, large OpenAlex fields such as earth and planetary sciences, agricultural and biological sciences, biochemistry, environmental science, mathematics, physics, neuroscience, and medicine all sit around the low-to-high sixties or low seventies in open-access share.

That is a real change in the social contract of science. It means a student, patient advocate, engineer, journalist, or small-city researcher can often reach the paper without a subscription wall. It does not make the paper understandable, trustworthy, translated, or reusable. But the locked door is opening.

Open-access share by field · 2024 · % of field outputfields with at least 100,000 works
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Open-access share of 2024 works by OpenAlex primary topic field. Only fields with at least 100,000 works are shown.

Calculation. Divide open-access 2024 field counts by total 2024 field counts; show fields with at least 100,000 works.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · 9af8de668469…
Hope

The frontier is less locked away than it used to be; access is becoming normal in many major fields.

Despair

Open to read is not open to understand, reproduce, translate, or build on.

Confusion

Open-access metadata is imperfect, and fields differ in preprint, repository, and journal culture.

Fig. 8
The collaboration

Is science becoming more international?

The share of works with institutions in more than one country rose from about 5.0% in 2000 to 12.6% in 2024.

The graph is knitting together. In 2000, only about one in twenty OpenAlex works had authorship institutions from more than one country. By 2024 it is about one in eight. That is not a majority. Most research is still nationally contained. But the direction is clear.

International collaboration is not automatically virtuous. It can reproduce hierarchy, with rich-country labs setting agendas and poorer-country partners supplying sites, samples, or local access. But it also means problems can be attacked at the scale they actually exist: pandemics, climate, food systems, migration, oceans, and AI are not domestic objects.

Internationally co-authored works · World · % of OpenAlex works2000 - 2024
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Share of OpenAlex works with authorship institutions in more than one country.

Calculation. Divided yearly works with countries_distinct_count > 1 by all yearly works, 2000-2024.

The displayed line is monotone-cubic smoothed for legibility only — it never moves a data point.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · 7b7db5c37eee…
Hope

More of science is now made across borders, which fits the scale of the problems science is asked to solve.

Despair

Collaboration can still be unequal: one institution writes the agenda while another supplies the terrain.

Confusion

Country counts depend on institutional metadata, so unaffiliated or poorly parsed works are easier to miss.

Fig. 9
The missing world

Is the Global South entering the visible graph?

The share of works with a Global South institution rose from about 4.5% in 2000 to 30.4% in 2024.

This is the most hopeful line in the piece. OpenAlex’s Global South institution flag appears on about 4.5 percent of works in 2000 and about 30.4 percent in 2024. That is not equality. It is a large opening in the formal map of knowledge.

The hard part is what the line does not say. A paper can include a Global South institution while the money, journal prestige, equipment, language, citation network, and theory-making power sit elsewhere. Participation is not control. Still, a world where more researchers can appear in the graph is different from one where they are invisible.

Works with a Global South institution · World · % of OpenAlex works2000 - 2024
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Share of OpenAlex works with at least one authorship institution marked Global South.

Calculation. Divided yearly works with a Global South authorship institution by all yearly works, 2000-2024.

The displayed line is monotone-cubic smoothed for legibility only — it never moves a data point.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · 32b126b1c7fd…
Hope

The visible research system is far less exclusively rich-world than it was at the start of the century.

Despair

Being present in the graph is not the same as setting the agenda or receiving equal credit.

Confusion

OpenAlex’s Global South flag is institution-based; author identity, funding control, and leadership are harder questions.

Fig. 10
The institutions

Who produces the visible record?

Education institutions dominate, but hospitals, facilities, governments, companies, and nonprofits all appear at scale.

OpenAlex is useful because it does not stop at papers. It connects works to institutions, and those institutions have types. In 2024, education institutions appear on about 5.5 million works. Healthcare and facilities each appear around one million. Government, companies, nonprofits, funders, archives, and others form the rest.

That is the modern research system: universities, hospitals, national labs, companies, field stations, repositories, funders, and infrastructure. Treating science as only universities misses how medicine, AI, energy, defence, agriculture, and climate knowledge are actually made.

Works by institution type · 2024 · millionauthor institutions
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Works published in 2024, grouped by type of authorship institutions attached in OpenAlex.

Calculation. Group 2024 works by authorship institution type; show counts in millions.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · 7b3e2ca6adfa…
Hope

Knowledge production is a broad infrastructure, not just a campus activity.

Despair

Corporate, government, and funder power can shape what questions are asked and what stays invisible.

Confusion

Institution-type metadata is coarse; the same organization can play several roles in the research system.

Fig. 11
The attention

Who gets highly cited?

Among large 2020 producers, smaller rich research systems can have a higher top-cited share than the biggest producers.

Volume and attention are different maps. Using 2020 works so citations have time to accumulate, OpenAlex’s citation-normalized top-10% flag puts Australia, the Netherlands, Switzerland, Italy, the United Kingdom, China, Spain, Germany, Canada, South Korea, the United States, and Poland near the top among countries with at least 75,000 country-attributed works.

This does not mean these countries are simply 'better’. Highly cited work follows field mix, language, collaboration networks, prestige, journal placement, and rich-country visibility. But it does show the key point: the country that produces the most works is not automatically the country whose works receive the most attention per work.

Top-cited share by country · 2020 works · % in top citation percentile bandcountries with at least 75,000 works
Source · OpenAlex · CC0 1.0
How this number was made

What it measures. Share of a country-attributed 2020 works that OpenAlex marks as being in the top 10 percent of citation-normalized percentile. Countries shown have at least 75,000 2020 works.

Calculation. Divide top-10-percent cited 2020 works by all 2020 works for each author-institution country.

source · OpenAlex (OpenAlex) license · CC0 1.0 vintage · 2026-06-13 checksum · c5662f50d997…
Hope

Influence is not only a brute-volume game; smaller systems can punch above their output.

Despair

Citations still reward prestige, English, networks, and fashionable fields, not just truth or usefulness.

Confusion

Citation influence should be read with field, collaboration, and time-lag context; it is a signal, not a verdict.

Science is becoming more open and more global. The question is whether power is moving as fast as participation.— Hope & Despair
I
The multi-disciplinary read

One question, read by the fields that actually study it. Each voice is given both its most hopeful and its darkest honest reading — with a note on how settled the evidence underneath is.

A student outside a rich universityconfidence · high

the hopeful readThe open-access charts matter most. Much more of the literature is reachable without a campus login.

the darker readReachable is not equal. Language, training, compute, journals, and networks still decide who can use the work.

A researcher in the Global Southconfidence · medium

the hopeful readParticipation in the visible graph has changed dramatically since 2000.

the darker readParticipation can be subordinate. Credit, authorship order, funding, and agenda-setting power may still sit elsewhere.

A science funderconfidence · medium

the hopeful readThe graph shows where new capacity exists and where collaboration could solve problems too large for one country.

the darker readFunding can amplify existing prestige rather than build neglected institutions, languages, and local questions.

II
The two strongest cases

Strongest case for hope

The hopeful case is strong: knowledge is multiplying, more of it is open, Asia and the Global South are far more visible, and cross-border science is rising. OpenAlex makes a public graph of a system that used to be mostly locked behind commercial indexes.

Strongest case for despair

The despairing case is that visibility is not equality. The graph can expand while power remains concentrated: rich institutions, English-language journals, citation networks, funders, companies, and prestige systems still shape what gets counted, cited, trusted, and turned into policy.

III
What would change this verdict

This verdict would move toward hope if Global South participation kept rising together with leadership measures: corresponding authorship, first authorship, local funding, open data, local-language dissemination, and top-cited work across neglected fields. It would move toward despair if participation rose only as junior partnership while citations, money, infrastructure, and agenda-setting stayed concentrated.

IV
Methodology

Everything done to the numbers between the source and the chart, disclosed in full. Where a choice is editorial rather than measured, we say so.

OpenAlex API
The article uses official OpenAlex group_by queries, snapshotted under data/sources/openalex. The page never calls the API at runtime.
Country attribution
Country bars group works by authorships.institutions.country_code. A multi-country paper can count for every represented country, so the bars measure footprint in the graph, not exclusive ownership.
Country growth multiple
The country-growth chart divides each country’s yearly country-attributed work count by its own 2000 value, so 1x means that country’s 2000 output. It shows pace, not absolute size.
Country share
The country-share chart divides each country’s yearly country-attributed work count by all OpenAlex works in that publication year. Because country counts are non-exclusive, shares should be read as footprint, not mutually exclusive slices.
Field time series
Field lines filter works by primary_topic.field.id, group by publication year, and show selected large fields from 2000 through 2024. The article stops at 2024 because 2025 and 2026 are still subject to indexing lag.
Field growth multiples
The field-growth bar divides each selected field’s 2024 output by its 2000 output. This answers what accelerated fastest, not what is largest.
Open access by field
The field openness chart divides 2024 is_oa:true field counts by all 2024 field counts and shows large fields with at least 100,000 works.
International collaboration
The collaboration line divides yearly works with countries_distinct_count > 1 by all yearly works from 2000 to 2024.
Global South participation
The Global South line divides yearly works with authorships.institutions.is_global_south:true by all yearly works. It is institution-based, not author-identity-based.
Citation influence
The attention chart uses 2020 works marked by OpenAlex as citation_normalized_percentile.is_in_top_10_percent:true, divided by all 2020 works for countries with at least 75,000 country-attributed works.
V
Sources & references
  1. OpenAlexopenalex.org · CC0 1.0 · vintage 2026-06-13Used for works, country attribution, fields, open access, collaboration, Global South participation, institution types, and citation-normalized top-cited shares.
Cite this page

Reference

Hope & Despair. “Who shapes science now?.” Hope & Despair, 2026. https://hopeanddespair.world/questions/who-shapes-science.

BibTeX

@misc{hopeanddespair:who-shapes-science,
  author       = {{Hope & Despair}},
  title        = {{Who shapes science now?}},
  year         = {2026},
  howpublished = {\url{https://hopeanddespair.world/questions/who-shapes-science}},
  note         = {hopeanddespair.world, updated 2026-06-14}
}

Original prose, code, and chart designs are released under CC BY 4.0 ↗ — reuse freely with attribution. Source datasets keep their own licences, listed above. Add your access date when citing.

VI
Revision history
  1. Mobile-first chart system + redesign branch snapshot View diff →

Generated from the commit history — every edit to a number, a word, or a chart is tracked in the open. Browse the full history & diffs on GitHub →

in plain words

If you only remember one thing, remember this: science is becoming more open and more global, but power is not the same thing as participation.

The old map of knowledge was simpler: a few rich countries, elite universities, subscription journals, and everyone else at the edge. That map is no longer true.

The new map is harder. More people are inside the graph. More work is open. More countries appear. But citation, funding, infrastructure, language, and agenda-setting still decide which parts of the graph become the story of science.