1% part08_networks.tex — Part VIII, Scientific Knowledge Networks (Agent L/I, 2026-11-03).2% Every corpus number is a macro from tables/numbers.tex or appears in a generated table (tables/make_tables.py).3% Result files: results/bibliometrics/*.csv (corpus v2, 2026-11-03 18:0x) and results/graph/*.csv (corpus v3 graph, 2026-11-03 18:44;4% tables tab_field_flows, tab_reference_coverage, tab_cited_age regenerated from them by tables/make_tables.py, Agent L/M).56\section{Scope and reading rules}7\label{sec:net-scope}89This part reports what the eligible corpus shows about the people, organisations, countries and venues that10produce computer-science research, about the distribution of attention over the works they produce, and about11the sources on which that research draws. It is the empirical counterpart of research questions RQ4 and RQ612and of the networks section of the methodology (Section~\ref{sec:m-networks}), and it tests the parts of13hypotheses H4, H5 and H6 that the current result files make testable. The citation-flow statistics between14taxonomy categories, which carry the main weight of H4, depend on the classification of Part~IX and are not15yet available; Section~\ref{sec:net-flows-categories} is a placeholder for them. The reader should16therefore expect a part that is complete on collaboration, geography, institutions, venues and aggregate17citation structure, and provisional on knowledge flows.1819Three reading rules apply throughout. First, every indicator is computed on the eligible corpus defined in20Part~III (Section~\ref{sec:data-eligibility}), so that ``computer science'' means the union of DBLP-indexed,21OpenAlex-primary-field and arXiv \texttt{cs} works after the exclusions of the PRISMA flow, and the indicators22inherit the boundary decisions and the affiliation completeness of that corpus. Second, the two periods of the23protocol are unequal in length and in completeness: the baseline runs from January 2018 to the end of24November 2022, the post period from December 2022 to the snapshot of October 2026 (the API increment of25Part~III is archived but not yet merged into the analytical corpus), and the last year of the post period is partial for the reasons set out in26Section~\ref{sec:data-dates}. Tables by year therefore carry 2026 as an incomplete row, and the fixed-window27citation statistics of Section~\ref{sec:net-citations} are complete only up to the publication years flagged in28Table~\ref{tab:citation_windows}. Third, the comparison of periods is descriptive. The protocol states the29hypotheses with a direction so that null and opposite results can be reported as such, and several of the30results below are of that kind; none of them is evidence of an effect of any single event, and the language of31this part is deliberately that of association and composition.3233The part is organised from the smallest unit of production outward. Section~\ref{sec:net-teams} describes34authorship and team structure; Section~\ref{sec:net-international} the international co-authorship network;35Section~\ref{sec:net-geography} the geography of output; Section~\ref{sec:net-institutions} institutions and36sectors, including the industry share that H6 concerns; Section~\ref{sec:net-venues} publication channels and37venue concentration; Section~\ref{sec:net-citations} the distribution of citations and its inequality;38Section~\ref{sec:net-flows} the inflows of knowledge from other fields and the age of the works cited;39Section~\ref{sec:net-boundaries} the boundary of the database itself, through the DBLP-only records and the40\texttt{created\_date} diagnostic; and Section~\ref{sec:net-hypotheses} collects the verdicts on H4--H6.4142\section{Authorship and team structure}43\label{sec:net-teams}4445The long-run growth of team size in science is one of the most replicated findings of scientometrics:46\citet{wuchty2007teams} documented the shift from solo to team production across fields over five decades,47\citet{lariviere2015teamsize} extended the series back to 1900 and showed that the citation advantage of teams48has grown with it, and \citet{milojevic2014teams} proposed a generative account in which a core of small,49``Poisson-like'' teams coexists with an expanding tail of large ones. Computer science has long been a field of50small teams by the standards of the laboratory sciences, which is why the question for this corpus is not51whether teams grow but whether the rate of growth or the shape of the distribution changes across the 202252boundary. The hypothesis set of the protocol does not take a position on team size, and the evidence is53reported for its own sake and as a control for the collaboration indicators that follow.5455Table~\ref{tab:collaboration} gives the annual series, and Figure~\ref{fig:collaboration} plots the mean and the56tail shares. The mean number of authors per eligible work rises from57\VolumeYearlyTwoZeroOneEightMeanAuthors{} in 2018 to \VolumeYearlyTwoZeroTwoTwoMeanAuthors{} in 2022 and58\VolumeYearlyTwoZeroTwoFiveMeanAuthors{} in 2025, a monotone increase at a pace that does not visibly change at59the boundary; the median stays at three authors in every year, so the growth of the mean is a growth of the60upper tail. The tail shares in the table make the same point: the share of works with five or more authors61rises steadily, and the share with ten or more, although small, roughly triples over the window. The62single-author share falls through 2022 and then rises again, sharply in 2025 and 2026. That reversal is not a change in63research practice. The column of works without any resolved country shows the same U-shape, and the64single-author and unaffiliated shares move together because the records that enter the corpus fastest after65publication (preprints, repository deposits, records not yet enriched with affiliations) are also those whose66authorship metadata is least complete; as those records are enriched and matched to their published versions67in later snapshots, both shares fall. The 2025 and 2026 rows of Table~\ref{tab:collaboration} must therefore68be read as a snapshot of the indexing process as much as of authorship, and the pre-2024 trend is the only69part of the series on which a substantive statement rests.7071\begin{figure}[htbp]72\centering73\includegraphics[width=0.95\textwidth]{fig_collaboration.pdf}74\caption{Team size, international co-authorship and industry participation by year. Mean authors per work and75tail shares from \texttt{team\_size\_yearly.csv}; the international share is the ICS of Part~IV (works with76two or more countries among works with at least one); the company share is over works with at least one77typed affiliation (\texttt{industry\_share\_yearly.csv}). 2026 is incomplete.}78\label{fig:collaboration}79\end{figure}8081\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_collaboration}\endgroup8283The multi-institution share, the fraction of works whose authors list more than one distinct institution,84follows the mean team size through 2022 and then declines, with the same affiliation-completeness85explanation as the single-author reversal: an unaffiliated record cannot be multi-institutional. Here the86substantive literature gives a reason to expect genuine growth. \citet{jones2008multiuniversity} showed that87multi-university collaboration had become the fastest-growing form of team production in the United States,88with a citation premium concentrated in the top-tier institutions, and \citet{adams2013fourthage} argued89that a ``fourth age'' of research organised around international networks of elite institutions was90displacing the national systems of the previous one. The corpus is consistent with continued growth of91multi-institution work up to the point where the affiliation data become unreliable and is silent beyond it.9293Two cautions apply to all of these shares, and they are not specific to this corpus. The first is that94OpenAlex author counts are taken from the authorship lists of the surviving record after deduplication95(Section~\ref{sec:m-dedup}); where a preprint and its published version differ in author list, the published96version wins, which is the appropriate choice but one that makes the count depend on the linkage. The second97is that very large author lists in computer science come disproportionately from a few genres, namely98consortium reports, shared-task overviews and large model ``technical reports'', so that the ten-plus share99measures the prevalence of those genres as much as the size of ordinary research teams. The disruption100literature attaches a substantive interpretation to team size, with \citet{wu2019disruption} reporting that101small teams produce more disruptive work and large teams more developmental work; that interpretation is102not tested here, because the disruption index requires a citation graph with a long post-publication horizon103that the post period does not yet provide.104105\section{International co-authorship}106\label{sec:net-international}107108International collaboration has grown in every field and every country for which long series exist109\citep{wagner2015cooperation, wagner2017collaboration}, and the growth has been described as the emergence110of a self-organising global network with a core group of countries that is stable over time111\citep{leydesdorff2008core}. The protocol's H6 predicts, against this background, that the international112co-authorship share of computer science does \emph{not} rise in the post period. The prediction was motivated113by two considerations stated in the protocol discussion: the geopolitical restriction of some114bilateral channels after 2020, and the shift of a growing share of the field's output into115preprint channels and into countries whose output is predominantly domestic.116117The international collaboration share (ICS, Section~\ref{sec:m-bibliometrics}) is computed among works with118at least one resolved country, so that the affiliation-completeness problem of the previous section does not119enter the denominator. Its yearly values are in the last column of Table~\ref{tab:collaboration} and are120macro-reported here for the end points: \InternationalCollabYearlyTwoZeroOneEightShareIntl{} in 2018,121\InternationalCollabYearlyTwoZeroTwoZeroShareIntl{} in 2020, \InternationalCollabYearlyTwoZeroTwoTwoShareIntl{}122in 2022, \InternationalCollabYearlyTwoZeroTwoFourShareIntl{} in 2024 and123\InternationalCollabYearlyTwoZeroTwoFiveShareIntl{} in 2025. The series is flat to slightly declining: it124peaks in 2020, is lower in every subsequent year, and the post-period values are below every baseline value.125The null part of H6 is therefore supported in the direction stated, and the result is in fact stronger than126the hypothesis, which only predicted the absence of a rise. Two qualifications prevent a stronger reading.127The denominator grows faster than the numerator over the window128(\InternationalCollabYearlyTwoZeroOneEightN{} works with a country in 2018 against129\InternationalCollabYearlyTwoZeroTwoFiveN{} in 2025), and the composition of that growth, examined in the next130section, is dominated by countries with low international shares; a constant propensity to collaborate within131every country would produce exactly this aggregate decline. And the share is a share of works, not of132authors or of citations, so that the increasing size of international teams noted by133\citet{wagner2017collaboration} would not register in it.134135The structure of the network behind the share is summarised in Tables~\ref{tab:country_network} and~\ref{tab:country_pairs}, which are built from the edge list of country pairs with at least fifty joint works136in a period. The threshold is a reporting choice (Section~\ref{sec:m-bibliometrics}); it keeps the network to137the pairs whose weight is robust to small linkage errors, at the cost of excluding the long tail of weak ties138through which, in the account of \citet{leydesdorff2008core}, peripheral countries enter the core. On this139restricted network, the number of edges and the number of countries both increase from the baseline to the140post period even though the post period is about ten months shorter, so the network densifies at the141threshold. The ordering of countries by strength, the sum of a country's edge weights, changes at the top:142the United States has the largest strength in the baseline and China in the post period, with the United143Kingdom third in both. The twenty strongest pairs (Table~\ref{tab:country_pairs}) show where the change comes144from. The China--United States pair is the heaviest edge in both periods, but its weight is lower in the post145period than in the baseline in absolute terms, while the pairs of China with Hong Kong, Singapore, the United146Kingdom and Australia rise in weight, and Hong Kong and Singapore rise in rank; the China--Hong Kong pair147moves from fourth to second.148The share of the China--United States edge in the total weight of the network falls accordingly149(Table~\ref{tab:country_network}). A decline in the absolute volume of joint work between the two largest150producers, in a period in which the output of both grew, is the most distinctive structural feature of the151collaboration data, and it is consistent with the bilateral-restriction channel that motivated H6; it is152not, on its own, evidence about the cause, and the reader should note that Hong Kong and Macau carry their153own country codes in OpenAlex, so that part of the apparent reorientation of China's collaboration is a154reorientation toward its own special administrative regions.155156Most of the rest of the pair table is continuity. The United Kingdom--United States, Germany--United157States and Germany--United Kingdom pairs keep approximately their weights; the India--United States pair158rises; the Canada--China and Canada--United States pairs fall slightly; and the France--United States and159Australia--United States pairs fall more and drop in rank, two movements that the table records and this160part does not explain. The picture is that of a core group whose composition is stable, as161\citet{leydesdorff2008core} and \citet{glanzel2001national} found for science as a whole, with one edge162weakening and a cluster of East-Asian edges strengthening. Degree-based statistics of the network163(density, clustering, assortativity) are not reported, because at a fixed weight threshold they measure164the growth of the corpus as much as the structure of collaboration; the strength ordering and the pair165weights, which are counts, are the statistics that survive that objection.166167\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_country_network}\endgroup168\input{tables/tab_country_pairs}169170\section{The geography of production}171\label{sec:net-geography}172173The country-level composition of output is the part of this study where the counting convention matters174most, and the methodology (Section~\ref{sec:m-bibliometrics}) reports both. Under full counting a work with175authors in three countries is counted once for each; under fractional counting it is divided among them.176The two conventions give different levels, and the difference is itself informative, because a country whose177full-count share is much larger than its fractional share is one whose output is predominantly178collaborative \citep{gauffriau2008counting, waltman2015counting}. Table~\ref{tab:countries} gives the179fractional counts of the fifteen largest countries by year, Table~\ref{tab:country_shares} the corresponding180shares, Table~\ref{tab:country_concentration} the concentration of the distribution and the full-count shares181of the two largest producers, and Figure~\ref{fig:country_shares} the shares of the eight largest.182183\begin{figure}[htbp]184\centering185\includegraphics[width=0.95\textwidth]{fig_country_shares.pdf}186\caption{Country shares of eligible output under fractional counting, eight largest countries by cumulative187output. Each work is divided equally among its distinct author countries; works without a resolved country188are excluded. Source: \texttt{results/bibliometrics/countries\_yearly.csv}. 2026 is incomplete.}189\label{fig:country_shares}190\end{figure}191192The dominant movement is the rise of China's share, which is already the largest in 2018 and grows in every193complete year of the window, under both counting conventions. The second is the decline of the United194States' share, which is monotone through 2025 under both conventions and recovers only in the incomplete1952026 row, where the fast channels in which American output is over-represented (preprint servers) weigh more196than they will once the year is fully indexed. The third is the growth of India and, more sharply in197relative terms, Indonesia, which enters the top five by the end of the window; the fourth is the slow198relative decline of the large European producers, Japan and Korea, whose absolute output is flat or rising199(Table~\ref{tab:countries}) while the total grows. The Russian series falls after 2021 and does not200recover. These movements are gradual and are visible in the baseline as well as in the post period;201nothing in the country shares changes direction at the 2022 boundary, and the growth of China's share is202if anything slower in 2024 and 2025 than in 2021 and 2022.203204The China part of H6 asks whether the share of works with China-affiliated authors rises in the post period205\emph{in the AI categories}. The aggregate rise is established by Table~\ref{tab:country_concentration},206whose column of works with a China-affiliated author is the quantity the hypothesis names; the category-specific version requires the207classification and is \pending{China share by category after classification}. The aggregate result is,208however, weaker evidence than it looks, because the rise is a continuation of a trend that predates the209window and is documented in every recent comparison of national output210\citep{bornmann2021growth}; a test of H6 that would be informative must compare the post-period211growth of China's share in AI categories with its growth in the rest of computer science, which is what the212pending analysis will do.213214Concentration across countries, measured by the Herfindahl--Hirschman index of fractional shares and by the215shares of the five and ten largest countries, rises through 2025 (Table~\ref{tab:country_concentration}).216The index is dominated by the largest share, so the rise is mostly China's rise; the top-ten share moves217less than the top-five share, which means that the countries ranked six to ten lose share to those ranked218one to five rather than to the tail. The number of countries with any output is stable. The widely219discussed concern that the geography of AI research is narrowing \citep{klinger2020narrowing,220ahmed2020dedemocratization} is a concern about institutions and compute rather than about countries, and the221country-level concentration measured here speaks to it only indirectly; the institutional evidence is in the222next section.223224The ratio of a country's full count to its fractional count, in the last two columns of225Table~\ref{tab:country_concentration}, exceeds one by the extent to which its works are co-authored with226other countries, and it is informative about collaboration intensity. China's ratio is lower than the United227States' in every year, which is the signature of a more domestic output: a larger fraction of228Chinese-affiliated works has only Chinese affiliations. Since China's share grows and the229United States' falls, the aggregate international share of Section~\ref{sec:net-international} is pulled230down by composition even if no country's propensity to collaborate changes. This is the mechanical231channel behind the null result on international collaboration, and it is the reason that result is232reported with the composition caveat rather than as a change in behaviour.233234\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_countries}\endgroup235\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_country_shares}\endgroup236\begingroup\renewcommand{\small}{\scriptsize}\setlength{\tabcolsep}{3pt}\input{tables/tab_country_concentration}\endgroup237238\section{Institutions and sectors}239\label{sec:net-institutions}240241\subsection{The sectoral mix of affiliations}242243OpenAlex types every institution as education, facility, company, government, healthcare, nonprofit,244funder, archive or other. Table~\ref{tab:institution_types} gives the distribution of typed affiliations over245these classes by year, and Table~\ref{tab:industry} the share of works with at least one company affiliation,246which is the statistic named by H6. The hypothesis predicts that the industry share rises in the post period247in AI categories. The prediction follows a line of work that has documented the growing weight of industry in248machine-learning research: \citet{frank2019citationgraphs} traced the shift of the AI citation graph toward249industry-authored and computer-science-internal sources, \citet{ahmed2023industry} reported that industry's250share of the largest models, of the most cited papers and of the field's PhD graduates had risen sharply in251the decade to 2022, and \citet{klinger2020narrowing} argued, in a preprint, that the thematic diversity of AI252research had narrowed as the private sector's share rose.253254The aggregate corpus does not show a rising industry share. The company share of works with a typed255affiliation is \IndustryShareYearlyTwoZeroOneEightShareCompany{} in 2018, rises to a maximum of256\IndustryShareYearlyTwoZeroTwoOneShareCompany{} in 2021, and then falls in every subsequent year to257\IndustryShareYearlyTwoZeroTwoFiveShareCompany{} in 2025, which is below the 2018 value; the incomplete2582026 row is at \IndustryShareYearlyTwoZeroTwoSixShareCompany{}. The absolute number of works with a company259affiliation keeps growing (\IndustryShareYearlyTwoZeroOneEightNCompany{} in 2018 against260\IndustryShareYearlyTwoZeroTwoFiveNCompany{} in 2025), so the falling share is a matter of the denominator261growing faster: the growth of the post-period corpus comes disproportionately from education-sector output,262and within that from the countries of Section~\ref{sec:net-geography} whose output is almost entirely263academic. Table~\ref{tab:institution_types} shows the same thing from the other side, with the education264share of affiliations rising through the window and the facility, government and company shares all giving265way. The aggregate part of H6 is therefore not supported; the sign is opposite to the one predicted. Whether266the category-specific version holds, that is, whether the company share rises within the AI categories while267falling in the aggregate, is exactly the kind of composition question that the aggregate cannot settle, and it268is \pending{industry share by category after classification}.269270The result is less at odds with the cited literature than it seems, for three reasons that matter for its271interpretation. \citet{ahmed2023industry} measure industry's presence at the frontier---the largest models,272the most cited papers, the leading conferences---and not its share of all output; a sector can dominate the273frontier of a field while its share of the field's total publications falls, and the two facts are compatible274with a frontier that is increasingly expensive to reach \citep{ahmed2020dedemocratization}. Second, the275OpenAlex institution type is assigned to the institution, not to the author's role, so a work by a276corporate research laboratory published with an academic co-author counts as both, and the share reported277here is the share of works with \emph{any} company affiliation, which is the most inclusive definition.278Third, the affiliation completeness problem of Section~\ref{sec:net-teams} affects the typed denominator:279records without affiliation data are excluded, and if industry-authored preprints are less often enriched280with affiliations than journal articles, the company share of recent years is understated. None of these281reasons reverses the aggregate finding, but together they confine it to what it measures.282283\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_industry}\endgroup284\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_institution_types}\endgroup285286\subsection{Institutional concentration and stability}287288The institution tables (Tables~\ref{tab:institution_concentration} and~\ref{tab:top_institutions}) list289institutions with at least two hundred works in a period, identified by their OpenAlex identifiers only; the290result files do not carry institution names, and this part refrains from resolving the identifiers by hand so291that every statement remains reproducible from the released artefacts. The number of institutions above the292threshold rises from the baseline to the post period, as does the number of works they account for, in both293counting conventions. The ratio of full to fractional counts among the largest institutions is close to two,294meaning that the average work of a large institution carries about one other institution, which is the295multi-institution pattern of Section~\ref{sec:net-teams} seen from the institution's side.296297Concentration among the listed institutions is low and stable: the ten largest account for a few per cent of298the listed fractional output and the hundred largest for less than a quarter, with both shares slightly299higher in the post period. This is a different picture from the one that the concern over the300``de-democratisation'' of AI research conveys \citep{ahmed2020dedemocratization}, and the difference is again301one of object: the distribution of \emph{all} computer-science output over institutions is wide, whereas the302distribution of the frontier of one sub-area may be narrow, and only the latter has been claimed to be303narrowing. The rank stability of institutions across periods is high: of the institutions listed in the304baseline, the great majority are also listed in the post period, and the Spearman correlation of their305fractional counts across the two periods, given in the note to Table~\ref{tab:institution_concentration}, is306close to nine tenths. Among the institutions listed in both periods, the twenty-five largest of the post period307(Table~\ref{tab:top_institutions}) have post-to-baseline ratios that cluster around or somewhat above the308ratio of the two periods' listed totals, with a few institutions growing considerably faster and two of the309largest full-count producers growing slower than the corpus. The institutional hierarchy of computer science, in other words, was not reshuffled in the310post period; it grew in place.311312\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_institution_concentration}\endgroup313\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_top_institutions}\endgroup314315\section{Publication channels and venue concentration}316\label{sec:net-venues}317318The venue is the unit at which the publishing system intersects the research system, and the two have been319changing in opposite directions in computer science for two decades: the publishing side has consolidated320into a small number of commercial and society publishers \citep{lariviere2015oligopoly}, while the research321side has moved a growing share of its first dissemination to arXiv322\citep{sutton2017arxivcs, lin2020arxivcs, lariviere2014arxiv}. The OpenAlex notion of source, on which the323tables of this section are built, follows the venue (a journal, a conference series, a book series, a324repository, an ebook platform) rather than the publisher, so what is measured here is concentration over325venues, and arXiv counts as one venue. The protocol's H5 anticipated this when it predicted that venue326concentration would be ``stable or lower because of preprint growth'' once the preprint source is treated327separately; the methodology (Section~\ref{sec:m-bibliometrics}) notes that the index with arXiv included328rises mechanically with preprint growth.329330Table~\ref{tab:venue_hhi} and Figure~\ref{fig:venue_concentration} give the yearly number of distinct sources,331the Herfindahl--Hirschman index over sources and the share of the largest source. The number of distinct332sources rises every complete year, from \VenueHhiYearlyTwoZeroOneEightNSources{} in 2018 to333\VenueHhiYearlyTwoZeroTwoFiveNSources{} in 2025, so the venue landscape is widening in the sense of variety.334The index, however, rises as well: from \VenueHhiYearlyTwoZeroOneEightHhi{} in 2018 to335\VenueHhiYearlyTwoZeroTwoFiveHhi{} in 2025 and \VenueHhiYearlyTwoZeroTwoSixHhi{} in the incomplete 2026336row, and the share of the largest source, which is arXiv in every year, goes from337\VenueHhiYearlyTwoZeroOneEightTopOneShare{} to \VenueHhiYearlyTwoZeroTwoFiveTopOneShare{} and then to338\VenueHhiYearlyTwoZeroTwoSixTopOneShare{} in 2026. The index is small in absolute terms (its upper bound is339one, reached when a single venue carries all works), and the squared share of arXiv accounts for most of it340in the later years; the 2026 value is an artefact of right truncation, since preprints are indexed within341days and journal and conference records months later (Section~\ref{sec:data-dates}), and will fall as the342year completes. The concentration result is therefore the one the protocol expected in mechanism and the343opposite of the one it expected in sign: concentration over venues, with arXiv counted as a venue, rises344because of preprint growth, and the version of the index without the preprint type, which H5 names, is345\pending{venue HHI excluding preprints}. The qualitative conclusion that does not depend on it is that the346rest of the venue landscape is not concentrating: the number of venues grows, and the second-ranked sources347in Table~\ref{tab:top_sources} hold shares that are an order of magnitude below arXiv's.348349\begin{figure}[htbp]350\centering351\includegraphics[width=0.95\textwidth]{fig_venue_concentration.pdf}352\caption{Number of distinct sources per year (left) and venue concentration (right): Herfindahl--Hirschman353index over sources, scaled by $10^4$, and share of the largest source, which is arXiv in every year.354Source: \texttt{results/bibliometrics/venue\_hhi\_yearly.csv}. 2026 is incomplete and inflated by the355faster indexing of preprints.}356\label{fig:venue_concentration}357\end{figure}358359Tables~\ref{tab:source_types} and~\ref{tab:top_sources} describe the channels themselves. By source type,360journals carry about half of the listed works in both periods, conferences between a fifth and a quarter,361and repositories rise from about an eighth to about a fifth between the baseline and the post period, at362the expense of conferences, book series and ebook platforms. The twenty largest sources make the composition concrete. arXiv363is first in both periods and more than doubles its count; Lecture Notes in Computer Science is second in both;364SSRN, which is not a computer-science venue but hosts a growing number of works that OpenAlex classifies in the365field, moves from the lower part of the baseline list to third; Research Square and NeurIPS enter the366post-period list and the AAAI conference proceedings rise within it; the large open-access367journals (IEEE Access, Sensors, Remote Sensing) remain in the list but with lower counts in the post period368than in the baseline, a decline whose cause (editorial policy, reclassification by OpenAlex, or the shorter369period) the corpus does not identify. The370appearance of generalist repositories (SSRN, Research Square, OSF) among the largest computer-science371sources is a boundary effect of the OpenAlex topic classification rather than a change in where computer372scientists publish, and it is one of the reasons the strict-core sensitivity set of Part~III exists.373374\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_venue_hhi}\endgroup375\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_source_types}\endgroup376\begingroup\scriptsize\setlength{\tabcolsep}{3pt}\input{tables/tab_top_sources}\endgroup377378\section{The distribution of citations}379\label{sec:net-citations}380381\subsection{Levels, zeros and inequality within fixed windows}382383Citation distributions are heavy-tailed in every field, and the degree of skewness is remarkably similar384across fields once counts are normalised by the field mean \citep{seglen1992skewness, radicchi2008universality,385albarran2011skewness}. The mechanisms proposed for the tail are cumulative advantage, the ``Matthew effect''386by which recognised work attracts recognition \citep{merton1968matthew, price1976cumulative}, and its387network formalisation as preferential attachment \citep{barabasi1999scaling}; the mechanism proposed for the388mass of the distribution is a combination of a work's intrinsic fitness and an ageing function389\citep{wang2013longterm}. What a cross-period comparison can measure is not these mechanisms but the shape390of the distribution at a fixed exposure, and the fixed windows of Section~\ref{sec:m-citations} are the device391that makes the shape comparable: $c^{(36)}$ counts the citations received in the publication year and the two392following calendar years, and is complete for publication years up to 2023.393394Table~\ref{tab:citation_windows} and Figure~\ref{fig:citation_inequality} give the yearly statistics. The mean395of $c^{(36)}$ rises from \CitationWindowsYearlyTwoZeroOneEightMeanCThreeSix{} for works published in 2018 to396\CitationWindowsYearlyTwoZeroTwoTwoMeanCThreeSix{} for 2022 and397\CitationWindowsYearlyTwoZeroTwoThreeMeanCThreeSix{} for 2023, the last complete year; the median is one398citation in most years and two in 2022, so the mean is carried by the tail. The share of works uncited at399three years falls from \CitationWindowsYearlyTwoZeroOneEightShareUncitedThreeSix{} for 2018 to400\CitationWindowsYearlyTwoZeroTwoTwoShareUncitedThreeSix{} for 2022 and rises again to401\CitationWindowsYearlyTwoZeroTwoThreeShareUncitedThreeSix{} for 2023. Both movements are what a growing402citing population produces: each publication year is cited by the following two years' output, and that output403grows, so later cohorts receive more citations at equal exposure \citep{bornmann2015growth}. The reversal in4042023 is the first sign of right truncation in the window, since the third calendar year of the 2023 cohort405is 2025, for which the slower channels of indexing (and citation extraction) are not complete; it is406flagged in the table and is the reason the 2024 row, whose window is formally incomplete, is not used.407408The Gini coefficient of $c^{(36)}$ is the inequality statistic named, with a deviation in window length409recorded in Section~\ref{sec:m-bibliometrics}, by H5. It falls from410\CitationWindowsYearlyTwoZeroOneEightGiniCThreeSix{} for 2018 to411\CitationWindowsYearlyTwoZeroTwoTwoGiniCThreeSix{} for 2022, then rises to412\CitationWindowsYearlyTwoZeroTwoThreeGiniCThreeSix{} for 2023; the 12-month Gini follows the same path at a413slightly higher level, as expected from the larger share of zeros in the shorter window. Since the share of414uncited works moves with the Gini, and uncited works are the main contributor to inequality in a short window415(Section~\ref{sec:m-bibliometrics}), the fall through 2022 is to a large extent a fall in the share of zeros416rather than a change in the shape of the tail, and the 2023 uptick is the truncation effect just described.417H5 predicts that citation inequality is higher in the post period than in the baseline \emph{for AI418categories}. The aggregate series can test only the aggregate statement, and for the one post-period419publication year with a complete window (2023) the aggregate Gini is above the 2022 value and below every420baseline value before 2021, so the aggregate evidence is mixed and dominated by truncation; the category421version of the test, which is the one the hypothesis states, is \pending{Gini by category after classification}.422The reader should also weigh the finding of \citet{nielsen2021inequality} that global citation inequality has423risen over several decades when measured on long windows across all fields; the three-year windows used here424measure a different quantity, early attention, whose inequality is driven by zeros, and the two results are425not in contradiction.426427\begin{figure}[htbp]428\centering429\includegraphics[width=0.95\textwidth]{fig_citation_inequality.pdf}430\caption{Fixed-window citation levels and inequality by publication year, for years with a complete43136-month window. Left: mean $c^{(36)}$ and the 99th percentile divided by ten. Right: Gini coefficient of432$c^{(36)}$ and share of works uncited at 36 months. Source:433\texttt{results/bibliometrics/citation\_windows\_yearly.csv}.}434\label{fig:citation_inequality}435\end{figure}436437The field-weighted citation impact supplied by OpenAlex, reported for comparison in438Table~\ref{tab:citation_windows}, has a mean above one in every year, which says that the eligible corpus is439cited above the OpenAlex world average for works of the same type, year and subfield; its median is below440one in every year, which says that the typical computer-science work is not. The two statements together441are the skewness result once more. The mean FWCI declines slowly, though not monotonically, across cohorts and the median rises through4422022 and collapses for 2025 and 2026, where the exposure is too short for the indicator to mean anything;443the methodology's reservations about FWCI (Section~\ref{sec:m-bibliometrics}) are borne out and the444indicator is not used further.445446\begingroup\renewcommand{\small}{\scriptsize}\setlength{\tabcolsep}{3pt}\input{tables/tab_citation_windows}\endgroup447448\subsection{The most cited works and what they reveal about the database}449450Table~\ref{tab:top_cited} lists the ten works with the largest $c^{(36)}$ in each publication year from 2018451to 2025, with their total citations to date. The list is reported because research question RQ3 asks for the452most influential contributions and because the mandate requires that the manuscript identify them; it is453reported with its errors visible because the errors are instructive. Three kinds of entry appear. The first454is the expected one: the foundational works of the deep-learning decade are present with the citation counts455one would expect, among them \emph{Deep Contextualized Word Representations}, \emph{MobileNetV2} and456\emph{Squeeze-and-Excitation Networks} for 2018, \emph{BERT} for 2019, \emph{Momentum Contrast} and457\emph{A Simple Framework for Contrastive Learning} for 2020, the \emph{Vision Transformer} and the458\emph{Swin Transformer} for 2021, \emph{High-Resolution Image Synthesis with Latent Diffusion Models} and459\emph{Masked Autoencoders Are Scalable Vision Learners} for 2022, \emph{Segment Anything} and460\emph{YOLOv7} for 2023, and \emph{DeepSeek-R1} for 2025, together with the scientific-software and461bioinformatics-resource papers (SciPy, NumPy, STRING, SWISS-MODEL, iTOL, AlphaFold Protein Structure462Database, UniProt, KEGG) that the OpenAlex topic classification places in computer science and that are463among the most cited works of any field.464465The second kind of entry is a duplicate or misdated record of a canonical work: \emph{Bitcoin: A Peer-to-Peer466Electronic Cash System} appears as an SSRN record dated 2019 and again dated 2022, \emph{Generative467Adversarial Nets} as a 2023 book chapter, \emph{Batch Normalization} as a 2024 preprint, \emph{Learning468Multiple Layers of Features from Tiny Images} as a 2024 article, and \emph{Attention Is All You Need} as a 2025469preprint. These are not errors of this study's deduplication, which merges a preprint with its published470version within the window (Section~\ref{sec:m-dedup}); they are distinct OpenAlex works, with their own471identifiers and their own citation counts, created when a reference string resolved to a new record rather472than to the existing one, and dated by the record rather than by the work. Their $c^{(36)}$ is large because473the citations of a canonical work accrue to whichever record the citing works' reference strings resolve to.474The third kind is the inconsistent record: entries whose window count exceeds their total count, or whose475total count is zero while their window count is in the thousands, including several records of the Dagstuhl476DROPS series and of regional journals whose titles bear no relation to the citation volume attached to477them. These are, to the best of the author's understanding, artefacts of OpenAlex's reference matching,478where a frequently cited string has been mapped to a wrong target; they are recognisable precisely because479the two counts disagree, and \citet{alperin2024openalex} and \citet{culbert2025openalexcoverage} document480related inconsistencies in OpenAlex citation data at the aggregate level.481482The consequences for this part are limited but must be stated. The aggregate citation statistics of the483previous subsection are insensitive to a small number of misattributed records among hundreds of484thousands per year, except through the maximum and the 99th percentile, which Table~\ref{tab:citation_windows} reports and485which should not be interpreted. The identification of the most influential works for RQ3, by contrast,486cannot rely on a ranked list from the database without a record-level audit, and the audit must check the487consistency of window and total counts, the uniqueness of the work across records, and the agreement of the488record's date with the work's first appearance. That audit is part of the human-review checkpoint on RQ3 and489is \pending{record-level audit of the top-cited lists}; until it is done, the titles above are quoted as what490the database returns, not as a ranking endorsed by this study. The lesson for the field-normalised scores491of Section~\ref{sec:m-citations} is the same as the one \citet{waltman2016review} draw for all mean-based492indicators: a reference class containing one record with tens of thousands of misattributed citations has a493mean that is wrong for every work in the class, and the normalised scores of Part~IX should be computed494after the exclusion of records failing the consistency check; whether the current pipeline applies that495exclusion is \pending{NCS computed after exclusion of inconsistent records}.496497\begingroup\footnotesize\setlength{\tabcolsep}{4pt}\input{tables/tab_top_cited}\endgroup498499\section{Knowledge inflows and the age of cited work}500\label{sec:net-flows}501502\subsection{Where the references of computer science point}503\label{sec:net-flows-fields}504505Research question RQ4 asks how citation networks and interdisciplinary knowledge flows have evolved, and506H4 predicts that the share of citations crossing area boundaries rises in the post period. The methodology507defines two levels at which flows are measured: between the taxonomy categories of this study, through the508flow matrices of equation~\eqref{eq:flow}, and between the corpus and the rest of the literature, through the509external edge set of Section~\ref{sec:m-citgraph}, whose cited works are typed by their OpenAlex field. Only510the second level is available at the time of writing. Figure~\ref{fig:field_inflows} and511Table~\ref{tab:field_flows} are generated from the current graph tables of \texttt{results/graph/}512(\texttt{field\_flows\_yearly.csv}, corpus v3), and Table~\ref{tab:reference_coverage}, from513\texttt{reference\_coverage\_yearly.csv}, gives the denominators on which every share in this section rests:514the declared references of the eligible works of each citing year, the share of them that resolve to an515OpenAlex record, and the split of the resolved references into internal and external edges.516517\begin{figure}[htbp]518\centering519\includegraphics[width=0.95\textwidth]{fig_field_inflows.pdf}520\caption{Share of the resolved references of eligible works, by citing year, that point to works inside the521corpus and to works outside it by cited OpenAlex field (seven largest external fields). Generated by522\texttt{figures/make\_figures.py} from \texttt{results/graph/field\_flows\_yearly.csv} (corpus v3); the523values are in Table~\ref{tab:field_flows}.}524\label{fig:field_inflows}525\end{figure}526527Table~\ref{tab:reference_coverage} is read first because it conditions the rest. The number of declared528references per eligible work rises through the baseline and falls again in the last three citing years, the5292026 row being the shortest; this is the right-truncation of reference lists (recent records are indexed530before their reference lists are complete) rather than a change in citing practice, and it means that the531shares of the last two rows of every table in this section are computed on fewer references per work than532the earlier rows. The share of declared references that resolve to an OpenAlex record is nearly constant533from 2018 to 2024 and higher in 2025--2026; the recent cohorts carry fewer parsed references per work (the534reference-extraction channels of OpenAlex lag publication), and the references that are already parsed for535them resolve more often. The difference describes how the database was assembled, not what authors cite, and536it is the reason the flow statistics are expressed as shares of resolved references, as the methodology requires.537The internal share of resolved references rises steeply with the citing year, and this is the mechanical538feature of Figure~\ref{fig:field_inflows} that must be read before any other: the corpus starts in 2018, so a539work published in 2018 cannot cite inside it except at age zero, whereas a work published in 2026 has eight540years of corpus to cite. The share of references pointing to computer science outside the corpus, which is541mostly pre-2018 computer science, falls as the internal share rises.542543The two computer-science columns of Table~\ref{tab:field_flows} do not, however, sum to a constant. Their544sum, the share of references that stay within the field, rises in every citing year of the window, and its545complement, the share of references that cross into another OpenAlex field (column `Other fields'), falls in546every year, in the baseline years as in the post years and at a comparable pace on either side of the 2022547boundary. Engineering, the largest external field, accounts for the greater part of the decline;548mathematics, decision sciences, physics and psychology decline slowly; the social sciences are stable; and549medicine is the only field among the seven whose share rises, so that by the end of the window it is the550second largest external destination after engineering. On this evidence the field-level version of H4 is551not supported: the boundary-crossing share of computer-science references toward other fields does not rise552after 2022, it continues to fall, and no change of pace is visible at the boundary.553554Two reservations qualify the direction, though not the absence of a break. The first is a classification555boundary. The eligible corpus is the CS core of Part~III, that is, works with a computer-science topic556whatever their primary field, so a post-2018 engineering work with a computer-science topic is an internal557edge while its pre-2018 counterpart is an external edge typed as engineering; as the corpus lengthens, part558of what used to be counted as a flow to engineering or mathematics is re-counted as internal, and the559decline of the external shares is to that extent mechanical. The field-flow file does not type internal560edges by the cited work's primary field, so the size of this effect cannot be removed here; a flow table561that types every resolved reference, internal or external, by primary field would be the correction, and it562is noted for the director. The second is that H4 concerns boundaries \emph{within} computer science563(systems toward AI, security toward AI, human--computer interaction toward AI), and a field whose sub-areas564converge on a common set of methods may well cite less outside the field, not more, as its internal565frontier absorbs the attention that used to go to engineering or mathematics. A falling external share is566therefore compatible with H4 and with its negation, and only the category flow matrices can decide between567them. What the field-level table does establish is a boundary condition for the interdisciplinarity568literature: the diversity indices of Section~\ref{sec:m-diversity}, whether Rao--Stirling or569Leinster--Cobbold, are computed over a reference distribution whose external component is shrinking570smoothly, so that any change they detect at the 2022 boundary will come from the internal component.571\citet{porter2009interdisciplinarity} found, for six fields over three decades, that interdisciplinarity572measured on references rose slowly and mostly through neighbouring fields; the computer-science columns show573the neighbouring fields (engineering, mathematics, decision sciences) losing share and a non-neighbouring574one (medicine) gaining it, and they do not show the acceleration that the foundation-model narrative would575predict.576577\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3pt}\input{tables/tab_reference_coverage}\endgroup578579\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{2.5pt}\input{tables/tab_field_flows}\endgroup580581\subsection{The age of cited work}582\label{sec:net-age}583584The age of the works a field cites is the simplest measure of how fast its knowledge turns over, and the585literature disagrees on its direction. \citet{evans2008narrowing} reported that online availability narrowed586citation toward fewer and more recent works; \citet{verstak2014giants}, in a preprint based on Google Scholar,587reported the opposite, a growing share of citations to older articles; \citet{parolo2015attention} measured588an accelerating decay of attention to individual papers as the volume of publication grew; and589\citet{chu2021slowed} argued that in very large fields the growth of publication itself ossifies the canon,590because new work cannot be read widely enough to displace it. Computer science is the field in which the591turnover is expected to be fastest, and \citet{frank2019citationgraphs} found that the AI citation graph had592become more self-referential and more recent-leaning over the 2010s.593594Figure~\ref{fig:cited_age} and Table~\ref{tab:cited_age} give the distribution of the age of cited works, the595citing year minus the cited work's publication year, for the resolved references of eligible works, read596from the corpus v3 file \texttt{results/graph/cited\_age\_yearly.csv}, which is built on the same edge sets as597Table~\ref{tab:reference_coverage}. The mode of citation age is two years in every citing year (visible in598the figure, where every curve peaks at the same abscissa); the mean age falls in every citing year from 2018599to 2026; the median falls over the baseline years and is constant from 2021 onward; the share of600references to works at most five years old rises by roughly the amount by which the share of references to601works more than ten years old falls; and the share of references to works at most two years old rises less,602and mostly during the baseline years. The shift is steady and does not accelerate at the 2022 boundary; the603post-period curves in the figure continue the movement of the baseline curves, and the year-to-year changes604of the mean in the table are of the same order before and after it, if anything smaller after. A small share of references point to605works dated after the citing work, which arises from preprints cited before their published version carries606a later date and from year-only dates; this share falls to zero over the window as the corpus's own dates607become first-appearance dates, and it is excluded from the age statistics.608609\begin{figure}[htbp]610\centering611\includegraphics[width=0.95\textwidth]{fig_cited_age.pdf}612\caption{Distribution of the age of cited works (citing year minus cited year, 0 to 30 years) for five613citing years. Generated by \texttt{figures/make\_figures.py} from614\texttt{results/graph/cited\_age\_yearly.csv} (corpus v3); summary statistics in Table~\ref{tab:cited_age}.}615\label{fig:cited_age}616\end{figure}617618The evidence therefore sides with \citet{evans2008narrowing} and \citet{parolo2015attention} rather than619with \citet{verstak2014giants} for this field and this window, with one reservation that the reference620coverage of OpenAlex imposes. If the probability that a reference string resolves to an OpenAlex record is621higher for recent, DOI-bearing works than for older ones, as \citet{culbert2025openalexcoverage} report in622their comparison with Web of Science and Scopus, then the resolved references are biased toward youth, and a623constant true age distribution would appear to become younger as the resolvable share rises.624Table~\ref{tab:reference_coverage} is the control. The resolved share is nearly constant from 2018 to 2024,625so over those years a resolution bias toward recent works would shift the level of the age distribution but626not its trend; the fall of the mean and of the over-ten-year share between 2018 and 2024 is therefore not an627artefact of a rising resolvable share. The 2025 and 2026 rows are different: their resolved share is higher628and their reference lists shorter (right-truncation), and the small further decline of the mean age in those629two rows should not be read on its own. The age shift is reported here as observed within the resolved630references, not as corrected for the composition of the unresolved ones, which cannot be typed.631632\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_cited_age}\endgroup633634\subsection{Knowledge flows between categories}635\label{sec:net-flows-categories}636637The flow matrices $F_t$ between the categories of the ontology, the cross-area shares $X_{\alpha,t}$ of638equation~\eqref{eq:cross-area} and the yearly modularities $Q_{\mathrm{cat}}$ and $Q_{\mathrm{area}}$ of639equation~\eqref{eq:modularity} are the statistics on which H4 turns and the ones that would show whether640systems, security and human--computer interaction cite the AI categories more after 2022 than before. They641require the primary category of every work and are produced by642\texttt{knowledge\_graph/02\_category\_flows.py} once the classification of Part~IX is complete. The result643files \texttt{category\_flows\_yearly.csv}, \texttt{cross\_area\_share\_yearly.csv} and644\texttt{modularity\_yearly.csv} do not exist at the time of writing, and this subsection is645\pending{category flows after classification}. When it is filled, it should report, for each citing area, the646cross-area share by year with the composition-adjusted variant of Section~\ref{sec:m-networks}, the two647modularities with the self-share, and a comparison of the baseline and post means with the caveat that the648post-period citing years have shorter reference lists in the incomplete 2026 row.649650\section{The boundary of the database}651\label{sec:net-boundaries}652653\subsection{What DBLP sees that the corpus does not}654655The corpus is the union of several sources resolved to OpenAlex records (Section~\ref{sec:data-linkage}),656and a DBLP record without an OpenAlex counterpart cannot enter it, because the citation, affiliation and657topic data on which every indicator of this part depends are OpenAlex's. Table~\ref{tab:dblp_only} counts658these records by DBLP type and year. They are a small fraction of DBLP's annual output, and their659composition is specific: journal articles with a DOI that OpenAlex has not indexed or has indexed under a660different identifier, conference papers mostly without a DOI, which are the hardest to link by title, and661doctoral theses, which DBLP indexes from a few national repositories and OpenAlex largely does not. The662``data'' type, DBLP's record of datasets and software, grows across the window and is almost entirely663unmatched, which is a reminder that the research outputs of computer science that are not papers are664outside the corpus altogether. For 2026 the conference share of the unmatched records rises sharply, which665is the DBLP-side manifestation of the right truncation discussed in Section~\ref{sec:data-dates}: DBLP666indexes a proceedings volume as a whole when it appears, OpenAlex matches it paper by paper over the667following months, and the gap is largest in the most recent year. The missing records bias the indicators668of this part toward the parts of computer science that publish with DOIs in journals and large conference669series, that is, away from theses, workshop papers and national venues; the direction of the bias is known670and its magnitude is bounded by the table.671672\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_dblp_only}\endgroup673674\subsection{Why the creation date does not measure indexing lag}675676The research plan proposed to estimate the indexing delay of OpenAlex from the \texttt{created\_date} of677each record and to correct the recent months of the publication series for it. Part~III678(Section~\ref{sec:data-dates}) records why the proposal was abandoned; Table~\ref{tab:indexing_lag} gives679the evidence in annual form. For every publication year up to 2024, the median lag between publication and680record creation is of the order of years, it is larger for January than for June publications by almost681exactly five months, and more than nine tenths of the year's records were ``created'' more than 180 days682after publication. A genuine indexing delay cannot have that shape: it would be measured in weeks or683months, would not depend on the month of publication within the year, and would not place nearly all684records beyond the 180-day mark. The shape is that of a single creation date for the bulk of the database,685with the lag equal to the time between each work's publication and that date, which places the date in the686second half of 2025. For works published in 2025 and 2026 the field behaves differently: a large share of687records is created within 90 days of publication, and the median lag collapses to days by mid-2026. The688field therefore measures the delay only for records created after the re-ingestion, and only for them689would a lag profile be meaningful; a profile estimated from the whole table, or from the years that the690structural-change analysis needs for its pre-trend diagnostics, would be an artefact of database691maintenance. This is the reason the structural-change analysis truncates the series rather than correcting692it (Section~\ref{sec:m-truncation}), and it is the reason the only honest estimator of indexing lag is the693comparison of two snapshots of the same database, which the released raw snapshots make possible for a694future study.695696\begingroup\renewcommand{\small}{\footnotesize}\setlength{\tabcolsep}{3.5pt}\input{tables/tab_indexing_lag}\endgroup697698\section{Hypotheses H4--H6 at this stage}699\label{sec:net-hypotheses}700701The protocol states its hypotheses with a direction so that null and opposite results are reported as702such. On the evidence of this part, the verdicts are the following, each limited to the component of the703hypothesis that the available tables test.704705H4 (convergence) is not testable in its stated form, which concerns citation flows between areas of the706ontology and the modularity of the area network; those statistics await the classification. Its field-level707counterpart, the share of computer-science references that cross into other OpenAlex fields, falls through708the window with no change of pace at the boundary (Table~\ref{tab:field_flows},709Section~\ref{sec:net-flows-fields}), part of the fall being a mechanical consequence of the corpus710boundary; it neither supports nor contradicts the within-field convergence the hypothesis describes.711712H5 (concentration) has two components. The citation-inequality component is stated for AI categories and is713pending; its aggregate counterpart gives a Gini of three-year citations that falls through the baseline and714rises for the one complete post-period cohort, a movement dominated by the share of uncited works and by715right truncation, and therefore not interpretable as a change in concentration. The venue-concentration716component predicted a stable or lower index once preprints are treated separately; the index with arXiv717included rises, for the mechanical reason the protocol anticipated, and the index without the preprint type718is pending. The non-arXiv venue landscape widens.719720H6 (geography and collaboration) has three components. The international co-authorship share does not721rise, as predicted, and in fact declines slightly, with a composition explanation that the country shares722support. The China share rises in the aggregate under both counting conventions, as predicted, but as the723continuation of a baseline trend, and the category-specific version is pending. The industry share does724\emph{not} rise in the aggregate: it peaks in 2021 and declines in every post-period year, which is the725opposite of the predicted direction, with the category-specific version pending and with the frontier-level726measures of the literature it was drawn from measuring a different object.727728What the part adds beyond the hypotheses is a set of structural observations that were not predicted and729that the pending analyses will have to accommodate: the fall in absolute joint output between China and the730United States during a period in which both grew; the stability of the institutional hierarchy; the rise of731generalist repositories among the largest sources of computer-science records; the steady shift of732references toward younger work without acceleration at the boundary; and the record-level artefacts in the733most-cited lists, which make any ranking of influential works conditional on an audit that this part has734specified but not performed.735736