Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear author,

To ensure that the article meets the standards of excellence required for publication in Systems, please find attached a list of points that should be considered and incorporated into the study.

Comments for author File: Comments.pdf

Author Response

Please see the attachment. Many thanks.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript makes a genuinely useful contribution by integrating four evaluation logics that the literature has almost always treated in isolation, and it does so through a feedback-based vocabulary (reinforcing/balancing loops, shifting feedback dominance) that fits the phenomenon far more naturally than a linear cause-effect account would. The most original piece of the paper is the concept of Sustainable Institutional Capacity (introduced p. 10, l. 324–329, and developed in Section 3.5, p. 12, l. 421–442), because it reframes the usual "is evaluation good or bad" debate into a question about the point at which cumulative evaluation load stops paying off — something far more actionable for university leadership. The organizational architecture of the paper is also a real strength: Tables 1–3 (pp. 7, 9, 14) place the contribution against the existing literature with unusual honesty, and Section 7 (p. 26, l. 888–910) acknowledges the conceptual, non-calibrated nature of the model without hedging. That kind of self-awareness is rare and works in the paper's favor rather than against it.

Beyond the points already raised on originality and referencing, a closer reading turns up several issues that are worth flagging specifically, with their locations:

The relations numbered (1)–(8) throughout Section 4 (e.g., p. 15, l. 507–511 and l. 522–524; p. 17, l. 549–564; p. 18–19, l. 592 and l. 603; p. 20, l. 627–630 and l. 639–641) are causal chains written with arrows, not equations in the System Dynamics sense — there is no stock, no rate, no functional form. Labeling them "(1)" through "(8)" alongside a paper that repeatedly invokes Forrester and Sterman as its methodological anchor risks implying a level of formalization the model does not yet have. I would relabel them as "Loop R1", "Loop B1", etc., consistent with the loop names already used in Table 3 (p. 14).

Section 3.6 is explicitly titled "Conceptual Model Validity and Boundary" (p. 12, l. 443) and discusses theoretical plausibility and internal consistency, yet the reference list contains no citation to Barlas's foundational work on formal validity testing in System Dynamics (structure verification, dimensional consistency, extreme-conditions tests), which is the standard citation for exactly this discussion and sits alongside Sterman [24], Forrester [27] and Schaffernicht [28] in every comparable CLD paper I have reviewed. Its absence is conspicuous precisely because the section title promises what it doesn't cite.

There is an unexplained inconsistency across the four domain-specific figures. Figure 1 (p. 16) introduces "Sustainable Research Capacity (SRC)" directly inside the diagram, Figure 3 introduces "Sustainable Ranking-Response Level (SRL)" (p. 17), Figure 5 introduces "Sustainable Quality-System Capacity (SQSC)" (p. 19), and Figure 7 introduces "Sustainable Responsible-Assessment Capacity (SRAC)" (p. 21) — none of these four constructs is defined or even named in the running text or in the numbered relations that precede each figure, and none is explicitly connected back to the overarching SIC construct introduced later in Section 3.5. Are SRC/SRL/SQSC/SRAC meant to be the four domain-level components that sum to SIC, or independent constructs that happen to share a naming pattern? The paper needs one sentence tying these together, ideally where SIC is first defined (p. 12, l. 429).

The abstract (p. 1, l. 16–20) states that the framework explains "growth, adaptation, stabilization, or overextension" — but the actual behavior-over-time reference modes go considerably further: Figure 2 depicts outright collapse for the research-prestige logic (p. 16, stage 5, "Possible Collapse"), and Figure 6 depicts damped oscillation for the QA logic (p. 19). Neither "collapse" nor "oscillation" appears in the abstract's own list of possible outcomes, so a reader skimming only the abstract would underestimate how severe the paper's own reference modes actually are. Worth reconciling, and worth a sentence somewhere (Section 5.2, p. 22, l. 693–716, would be the natural place) explaining why research prestige is assumed to be more collapse-prone than the other three logics, since no argument is currently given for that asymmetry.

On the Acknowledgments (p. 27, l. 951–952): the disclosed tool is named "ChatGPT (GPT-5.6 Sol, OpenAI)." I could not verify that this corresponds to a documented OpenAI release; I'd ask the author to double-check the exact model name and version against OpenAI's own documentation before proofs, since MDPI's AI-disclosure policy expects the specific tool identifier to be accurate and traceable.

Reference formatting against the Systems author guidelines. I checked the reference list (pp. 27–29) against MDPI's instructions for this journal. Two concrete deviations:

Author names throughout the 40 references are separated by commas (e.g., ref. 1, p. 27, l. 955–956: "Greenwood, R., Raynard, M., Kodeih, F., …"; ref. 13, p. 28, l. 979–980, the self-citation), whereas MDPI's required format separates each author entry with a semicolon ("Author 1, A.B.; Author 2, C.D. …"). This affects every entry and should be corrected globally before typesetting.

Reference 22 (CoARA, p. 28, l. ~998) gives only "Coalition for Advancing Research Assessment (CoARA). Coalition for Advancing Research Assessment. 2023." with no URL and no access date, while refs. 37, 38 and 39 (same page) correctly follow the required website format "Available online: URL (accessed on Day Month Year)." Ref. 22 should be brought in line with the others.

Separately from formatting, the literature review is thorough within institutional logics, rankings, QA and responsible-assessment research, but it does not engage with the existing System Dynamics literature on higher education management (enrollment, faculty workforce, funding dynamics), which is the closest methodological precedent to what this paper attempts and its absence is more noticeable given how central the SD framing is to the paper's claimed contribution.

None of these points bear on the soundness of the core argument, which I still find convincing; they are the kind of revision that strengthens a conceptual paper's credibility with readers who work in the SD tradition specifically, without requiring new data or a different study design.

On the quality of English. The text is grammatically correct, and the AI-driven polishing mentioned by the author in the Acknowledgments is evident; however, the result is stylistically very uniform and repetitive: the same pattern of qualification (“may…,” “can…,” “could therefore…”) is repeated in almost every paragraph of the article (see, for example, p. 4, lines 173–180; p. 6, lines 234–238; p. 22, lines 683–686), which gives the text as a whole a formulaic tone and makes it difficult to distinguish when the author is asserting something with conviction and when he is simply considering possibilities. It would be advisable to vary the register and shorten some particularly long sentences (e.g., p. 12, l. 429–434, or p. 23, l. 746–750), where a single sentence links three or four subordinate ideas. This is not a problem of comprehension—the argument follows easily—but rather one of fluency; the style, being so uniform, can come across as overly formulaic or mechanically edited.

Author Response

Please see the attachment. Many thanks.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript is well structured and presents a relevant conceptual contribution by integrating research prestige, rankings, quality assurance, and responsible assessment within a System Dynamics framework, while also introducing the concept of Sustainable Institutional Capacity (SIC). The theoretical framework is coherent and the discussion of reinforcing and balancing feedback mechanisms is clearly developed. However, a few minor revisions would strengthen the paper. In particular, the authors could provide a brief clarification of how the literature used for the theoretical synthesis was selected, since the framework is based on a purposive and iterative review process. The SIC concept could also be developed slightly further by specifying its main dimensions or possible indicators, such as workload, administrative burden, resource availability, staff capacity, organizational learning, and mission alignment. In addition, the contextual variability and boundary conditions of the cross-system feedback mechanisms could be explained more explicitly, as the manuscript acknowledges that these mechanisms may differ across institutional and national settings. Finally, some repetition could be reduced and the figure captions could more clearly indicate that the behavior-over-time trajectories are theoretical reference modes rather than empirical or predictive results.

Author Response

Please see the attachment. Many thanks.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I thank the authors for their thorough and constructive response to my previous comments. The revisions substantially strengthen the manuscript, and I am satisfied with how each point was addressed.