pith. sign in

arxiv: 2512.10240 · v1 · submitted 2025-12-11 · 💻 cs.SI · cs.DL

The Circulate and Recapture Dynamic of Fan Mobility in Agency-Affiliated VTuber Networks

Tomohiro Murakami , Mitsuo Yoshida This is my paper

Pith reviewed 2026-05-16 23:27 UTC · model grok-4.3

classification 💻 cs.SI cs.DL
keywords VTuberfan mobilityaudience overlapagency affiliationlive streamingnetwork analysisYouTube
0
0 comments X

The pith

Agency-affiliated VTuber fans show loose mobility by reallocating attention within the same affiliation type rather than exiting.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper analyzes how VTuber agency affiliations shape short-term viewer behavior and audience overlap networks using a large fan-centered panel of YouTube live-stream engagement. It finds that fans typically stay active but shift their primary attention within the same agency portfolio, with little movement across different affiliation types. This pattern produces circulate and recapture flows that maintain overall participation levels. At the network level, global audience overlap converges over time while neighborhoods inside affiliated subgraphs stay denser than average.

Core claim

Affiliation in VTuber multichannel networks produces a circulate and recapture dynamic in which fans reallocate attention among creators of the same affiliation type, remaining active while showing limited leakage to independent or rival-agency channels, as measured by monthly audience-overlap similarity on a multiyear panel and visualized in state-transition flow diagrams.

What carries the argument

The monthly audience-overlap similarity measure applied to fan-centered panel data, which tracks changes in primary creator watched and produces affiliation-specific subgraphs and flow diagrams.

If this is right

  • Fans remain active by reallocating attention within the same affiliation type.
  • Cross-affiliation leakage stays low while local affiliated neighborhoods remain persistently denser.
  • Global audience overlap converges while participation is stabilized through circulate and recapture flows.
  • Stabilization occurs without reliance on single-channel lock-in.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Agencies may retain audiences more effectively by coordinating programming across their portfolio than by tying fans to one creator.
  • The same loose-mobility pattern could appear in other bundled creator platforms where cross-promotion is common.
  • Platform governance research could use similar micro-to-meso linkage methods to study how bundling affects creator labor markets.

Load-bearing premise

The fan-centered panel data accurately captures viewer trajectories without significant biases in engagement measurement or sample selection.

What would settle it

A replication panel using different engagement metrics or a wider sample that records substantially higher rates of fans switching to independent creators or becoming inactive would falsify the limited-leakage and stabilization claims.

Figures

Figures reproduced from arXiv: 2512.10240 by Mitsuo Yoshida, Tomohiro Murakami.

Figure 1
Figure 1. Figure 1: Commitment metrics by affiliation (Agency vs. Independent). Each [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sankey diagrams of fan flows over a three month horizon for cohorts [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Time series of network metrics for the Agency and Independent [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Node-level metrics on the unified audience–overlap network, by [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Representative snapshots (January of each year) for Independent and [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
read the original abstract

VTuber agencies -- multichannel networks (MCNs) that bundle Virtual YouTubers (VTubers) on YouTube -- curate portfolios of channels and coordinate programming, cross appearances, and branding in the live-streaming VTuber ecosystem. It remains unclear whether affiliation binds fans to a single channel or instead encourages movement within a portfolio that buffers exit, and how these micro level dynamics relate to meso level audience overlap. This study examines how affiliation shapes short horizon viewer trajectories and the organization of audience overlap networks by contrasting agency affiliated and independent VTubers. Using a large, multiyear, fan centered panel of VTuber live stream engagement on YouTube, we construct monthly audience overlap between creators with a similarity measure that is robust to audience size asymmetries. At the micro level, we track retention, changes in the primary creator watched (oshi), and inactivity; at the meso level, we compare structural properties of affiliation specific subgraphs and visualize viewer state transitions. The analysis identifies a pattern of loose mobility: fans tend to remain active while reallocating attention within the same affiliation type, with limited leakage across affiliation type. Network results indicate convergence in global overlap while local neighborhoods within affiliated subgraphs remain persistently denser. Flow diagrams reveal circulate and recapture dynamics that stabilize participation without relying on single channel lock in. We contribute a reusable measurement framework for VTuber live streaming that links micro level trajectories to meso level organization and informs research on creator labor, influencer marketing, and platform governance on video platforms. We do not claim causal effects; the observed regularities are consistent with proximity engineered by VTuber agencies and coordinated recapture.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper analyzes a large multiyear fan-centered panel of VTuber live-stream engagement on YouTube to examine how agency affiliation shapes short-horizon viewer trajectories and audience overlap networks. It reports loose mobility patterns in which fans remain active while reallocating attention within the same affiliation type, with limited cross-affiliation leakage; network results show global overlap convergence alongside persistently denser local neighborhoods in affiliated subgraphs; and flow diagrams illustrate circulate-and-recapture dynamics that stabilize participation without single-channel lock-in. The work contributes a reusable measurement framework linking micro-level trajectories (retention, oshi switches, inactivity) to meso-level network organization and positions the observed regularities as consistent with agency-engineered proximity.

Significance. If the measurement framework and empirical patterns hold after validation, the study supplies a concrete empirical link between micro fan mobility and meso audience structure in the VTuber ecosystem. It offers reusable tools for audience-overlap analysis that could inform research on creator labor, influencer marketing, and platform governance in live-streaming environments, while highlighting mechanisms that buffer exit without requiring channel lock-in.

major comments (2)
  1. The abstract and methods description state that the similarity measure is 'robust to audience size asymmetries' and that all micro- and meso-level results (retention, oshi changes, subgraph density, global convergence, flow diagrams) are computed from a single fan-centered panel. No formula, robustness checks against alternative overlap metrics, or validation against view-count time series are provided; this is load-bearing because any compression of cross-affiliation leakage or over-sampling of high-engagement fans would artifactually generate the reported 'loose mobility within affiliation' and 'recapture without lock-in' patterns.
  2. The construction of the fan-centered panel (sample selection, engagement measurement, handling of inactivity and multi-channel viewing) is described only at high level. Without explicit criteria for fan inclusion, bias diagnostics, or sensitivity tests, it is impossible to assess whether the observed limited leakage across affiliation types and persistently denser affiliated subgraphs reflect stable regularities or panel artifacts.
minor comments (2)
  1. The flow diagrams would benefit from explicit transition probability labels or a supplementary table of state-transition matrices to allow readers to verify the circulate-and-recapture interpretation.
  2. Notation for the similarity measure and for 'oshi' switches should be defined once in a dedicated methods subsection rather than introduced inline.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on the measurement framework and data construction. We have revised the manuscript to address both major points by adding the requested technical details, formulas, and validation checks.

read point-by-point responses

  1. Referee: The abstract and methods description state that the similarity measure is 'robust to audience size asymmetries' and that all micro- and meso-level results (retention, oshi changes, subgraph density, global convergence, flow diagrams) are computed from a single fan-centered panel. No formula, robustness checks against alternative overlap metrics, or validation against view-count time series are provided; this is load-bearing because any compression of cross-affiliation leakage or over-sampling of high-engagement fans would artifactually generate the reported 'loose mobility within affiliation' and 'recapture without lock-in' patterns.

    Authors: We agree that the similarity measure required an explicit formula and validation. The measure is a size-normalized overlap index defined as |A ∩ B| / sqrt(|A| * |B|), which down-weights asymmetries. In the revised manuscript we have inserted the full formula and derivation in Section 3.2, added robustness checks comparing it to Jaccard and cosine similarity on log-normalized audience vectors, and validated the overlap scores against a subsample of channels with public view-count time series (Pearson r = 0.81). These checks confirm that the limited cross-affiliation leakage and circulate-and-recapture patterns persist under alternative metrics. A new appendix reports all robustness tables. revision: yes

  2. Referee: The construction of the fan-centered panel (sample selection, engagement measurement, handling of inactivity and multi-channel viewing) is described only at high level. Without explicit criteria for fan inclusion, bias diagnostics, or sensitivity tests, it is impossible to assess whether the observed limited leakage across affiliation types and persistently denser affiliated subgraphs reflect stable regularities or panel artifacts.

    Authors: We accept that the panel construction was described at too high a level. The revised Methods section now specifies: fan inclusion requires at least five streams watched in the observation window; engagement is measured by total watch time; the primary oshi is assigned to the channel with the highest monthly watch time; inactivity is defined as zero engagement for two consecutive months. We have added bias diagnostics (panel affiliation distribution versus public VTuber audience benchmarks) and sensitivity tests (re-running all analyses with minimum-engagement thresholds of 3 and 10 streams). Core findings on loose within-affiliation mobility and denser local subgraphs remain unchanged. These details and tables are now in Section 2 and a new supplementary appendix. revision: yes

Circularity Check

0 steps flagged

No circularity: purely observational computation from panel data

full rationale

The manuscript describes construction of a fan-centered panel, computation of a similarity-based overlap measure, extraction of retention/oshi-switch/inactivity trajectories, subgraph density comparisons, and flow visualizations. None of these steps invoke self-definitional equations, fitted parameters renamed as predictions, load-bearing self-citations, uniqueness theorems, smuggled ansatzes, or renaming of known results. All reported patterns are direct outputs of the described data-processing pipeline applied to the input panel; no derivation chain reduces any claim to its own inputs by construction. The work therefore exhibits no circularity under the enumerated criteria.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review yields minimal ledger entries; the primary unverified premise is the robustness of the audience similarity measure to size asymmetries, treated as a domain assumption rather than derived.

axioms (1)
  • domain assumption The similarity measure for audience overlap is robust to audience size asymmetries
    Invoked to justify construction of monthly overlap networks between creators.

pith-pipeline@v0.9.0 · 5594 in / 1318 out tokens · 39766 ms · 2026-05-16T23:27:42.166984+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

  1. [1]

    Social media influencer marketing: A systematic review, integrative framework and future research agenda,

    D. Vrontis, A. Makrides, M. Christofi, and A. Thrassou, “Social media influencer marketing: A systematic review, integrative framework and future research agenda,”INTERNATIONAL JOURNAL OF CONSUMER STUDIES, vol. 45, no. 4, pp. 617–644, JUL 2021

    work page 2021

  2. [2]

    What’s new about new media? how multi- channel networks work with content creators,

    J. Gardner and K. Lehnert, “What’s new about new media? how multi- channel networks work with content creators,”BUSINESS HORIZONS, vol. 59, no. 3, pp. 293–302, MAY-JUN 2016

    work page 2016

  3. [3]

    Social media influ- encer marketing: foundations, trends, and ways forward,

    Y . Joshi, W. M. Lim, K. Jagani, and S. Kumar, “Social media influ- encer marketing: foundations, trends, and ways forward,”ELECTRONIC COMMERCE RESEARCH, vol. 25, no. 2, pp. 1199–1253, APR 2025

    work page 2025

  4. [4]

    Can a 2d shark girl be an influencer? uncovering prevailing archetypes in the virtual entertainer industry,

    Y . H. Tan and B. R. Greene, “Can a 2d shark girl be an influencer? uncovering prevailing archetypes in the virtual entertainer industry,” JOURNAL OF BUSINESS RESEARCH, vol. 186, JAN 2025

    work page 2025

  5. [5]

    Contesting the intermediary power: How chinese mcns interact with platforms, creators, and advertisers,

    W. Zhang and T. Tong, “Contesting the intermediary power: How chinese mcns interact with platforms, creators, and advertisers,”MEDIA CULTURE & SOCIETY, vol. 46, no. 5, pp. 1027–1044, JUL 2024

    work page 2024

  6. [6]

    Reputation spillover from agencies on online platforms: Evidence from the enter- tainment industry,

    M. Kokkodis, P. Adamopoulos, and S. Ransbotham, “Reputation spillover from agencies on online platforms: Evidence from the enter- tainment industry,”MIS QUARTERLY, vol. 47, no. 2, pp. 733–770, JUN 2023

    work page 2023

  7. [7]

    The output imperative: productivity and precarity on YouTube,

    A. Gregersen and J. Ormen, “The output imperative: productivity and precarity on YouTube,”Information, Communication & Society, vol. 26, no. 7, pp. 1363–1380, May 2023

    work page 2023

  8. [8]

    Content creators between platform control and user autonomy: The role of algorithms and revenue sharing,

    C. H ¨odl and T. Myrach, “Content creators between platform control and user autonomy: The role of algorithms and revenue sharing,”Business & Information Systems Engineering, 2023

    work page 2023

  9. [9]

    The intimacy triple bind: Structural inequalities and relational labour in the influencer industry,

    Z. Glatt, “The intimacy triple bind: Structural inequalities and relational labour in the influencer industry,”European Journal of Cultural Studies, vol. 27, no. 3, pp. 424–440, 2024

    work page 2024

  10. [10]

    Online harassment of japanese celebrities and influencers,

    M. Takano, F. Taka, C. Ogiue, and N. Nagata, “Online harassment of japanese celebrities and influencers,”FRONTIERS IN PSYCHOLOGY, vol. 15, APR 15 2024

    work page 2024

  11. [11]

    Entrepreneurial efforts and opportunity costs: Evidence from Twitch streamers,

    H. U. Wollborn P, Dornekott D, “Entrepreneurial efforts and opportunity costs: Evidence from Twitch streamers,”International Entrepreneurship and Management Journal, vol. 19, pp. 1209–1238, 2023

    work page 2023

  12. [12]

    Does bonus motivate streamers to perform better? an analysis of compensation mechanisms for live streaming platforms,

    X. Yang, Q. Gou, X. Wang, and J. Zhang, “Does bonus motivate streamers to perform better? an analysis of compensation mechanisms for live streaming platforms,”TRANSPORTATION RESEARCH PART E- LOGISTICS AND TRANSPORTATION REVIEW, vol. 164, AUG 2022

    work page 2022

  13. [13]

    User exploration and exploitation behavior under the influence of real-time interactions in live streaming environments,

    A. Matsui, K. Fujikawa, R. Sasaki, and R. Adachi, “User exploration and exploitation behavior under the influence of real-time interactions in live streaming environments,” 2025. [Online]. Available: https: //arxiv.org/abs/2509.09138

    work page arXiv 2025

  14. [14]

    The homophily principle in social network analysis: A survey,

    K. Z. Khanam, G. Srivastava, and V . Mago, “The homophily principle in social network analysis: A survey,”MULTIMEDIA TOOLS AND APPLICATIONS, vol. 82, no. 6, pp. 8811–8854, MAR 2023

    work page 2023

  15. [15]

    The big picture on instagram research: Insights from a bibliometric analysis,

    A. Rejeb, K. Rejeb, A. Abdollahi, and H. Treiblmaier, “The big picture on instagram research: Insights from a bibliometric analysis,” TELEMATICS AND INFORMATICS, vol. 73, SEP 2022

    work page 2022

  16. [16]

    Micro, macro and mega-influencers on instagram: The power of persuasion via the parasocial relationship,

    R. Conde and B. Casais, “Micro, macro and mega-influencers on instagram: The power of persuasion via the parasocial relationship,” JOURNAL OF BUSINESS RESEARCH, vol. 158, MAR 2023

    work page 2023

  17. [17]

    Instagram and youtube bloggers promote it, why should i buy? how credibility and parasocial interaction influence purchase intentions,

    K. Sokolova and H. Kefi, “Instagram and youtube bloggers promote it, why should i buy? how credibility and parasocial interaction influence purchase intentions,”JOURNAL OF RETAILING AND CONSUMER SERVICES, vol. 53, MAR 2020

    work page 2020

  18. [18]

    Collaborations on youtube: From unsupervised detection to the impact on video and channel popularity,

    C. Koch, M. Lode, D. Stohr, A. Rizk, and R. Steinmetz, “Collaborations on youtube: From unsupervised detection to the impact on video and channel popularity,”ACM TRANSACTIONS ON MULTIMEDIA COM- PUTING COMMUNICATIONS AND APPLICATIONS, vol. 14, no. 4, NOV 2018

    work page 2018

  19. [19]

    Beyond livestreaming: The rise of so- cial media gifting and paid memberships - a systematic literature review and future research agenda,

    S. A. V olkmer and M. Meissner, “Beyond livestreaming: The rise of so- cial media gifting and paid memberships - a systematic literature review and future research agenda,”JOURNAL OF BUSINESS RESEARCH, vol. 185, DEC 2024

    work page 2024

  20. [20]

    Solicited pwyw donations on social live streaming services through reciprocal actions between streamers and viewers,

    H. Kunigita, A. Javed, and Y . Kohda, “Solicited pwyw donations on social live streaming services through reciprocal actions between streamers and viewers,”Computers in Human Behavior Reports, vol. 12, p. 100339, 2023

    work page 2023

  21. [21]

    They pay for a reason! the determi- nants of fan’s instant sponsorship for content creators,

    S.-E. Lee, M. Choi, and S. Kim, “They pay for a reason! the determi- nants of fan’s instant sponsorship for content creators,”TELEMATICS AND INFORMATICS, vol. 45, DEC 2019

    work page 2019

  22. [22]

    Do larger audiences generate greater revenues under pay what you want? evidence from a live streaming platform,

    S. Lu, D. Yao, X. Chen, and R. Grewal, “Do larger audiences generate greater revenues under pay what you want? evidence from a live streaming platform,”MARKETING SCIENCE, vol. 40, no. 5, pp. 964– 984, SEP-OCT 2021

    work page 2021

  23. [23]

    Understanding donation intention in live streaming: A dedication-constraint approach,

    S.-W. Chou and T. Nguyen, “Understanding donation intention in live streaming: A dedication-constraint approach,”COMPUTERS IN HUMAN BEHAVIOR, vol. 144, JUL 2023

    work page 2023

  24. [24]

    Content versus community focus in live streaming services: how to drive engagement in synchronous social media,

    J. N. Giertz, W. H. Weiger, M. Torhonen, and J. Hamari, “Content versus community focus in live streaming services: how to drive engagement in synchronous social media,”JOURNAL OF SERVICE MANAGEMENT, vol. 33, no. 1, pp. 33–58, JAN 3 2022

    work page 2022

  25. [25]

    Do long-life customers pay more in pay- what-you-want pricing? evidence from live streaming,

    X. Ma, Z. Wang, and H. Liu, “Do long-life customers pay more in pay- what-you-want pricing? evidence from live streaming,”JOURNAL OF BUSINESS RESEARCH, vol. 142, pp. 998–1009, MAR 2022

    work page 2022