When asked to recommend reviewers, many people’s first reaction is to “dig through their memory”: Who do I know in this field? Who has worked on something similar? Whose comments last time were solid? It is a quick move, but when the editor comes back with “Why recommend these people?” they often get stuck—unable to give specific grounds and only able to offer a vague impression.
The trouble does not stop there. Subject match is only the first gate; the second is conflict avoidance: same institution, recent collaborators, and advisor-advisee relationships all have to be ruled out. The third gate is stating the reasons clearly—recommending is not just naming people; it is letting the editor see why these people are suitable and where the grounds lie. The value of the MuchSci Talent Profile here is precisely that it turns “thinking of someone” into “being able to explain why it is this person.”
01 How an Explainable List Is Built
Start with the sequence of this action. Its biggest difference from “listing names from impression” is that every step leaves a documented basis.
▲ Figure 1 | It is not about writing down whoever comes to mind, but about every step leaving a documented basis
The first step is to define the field based on the manuscript’s topic. Do not rush to think of names; first write clearly what subfield the manuscript needs—whether it is at the methods level, the application level, or some interdisciplinary direction. Only when the direction is specific can “match” have a comparable standard. The second step is to retrieve candidates’ profiles and look at four types of fields: research areas in Basic Information, institution and position in Employment History, and projects led in Research Achievements.
The third step is to check subject match. See whether the projects led and the direction of achievements fit the manuscript’s subfield. In the profile, projects related to national-level major engineering will be specially marked; this kind of information is useful when judging “what scale of work he has done.”
The fourth step is to screen for conflicts of interest, and this step is the one that can least be skipped: same institution, recent collaborators, and advisor-advisee relationships must be filtered item by item. Only the fifth step is writing the recommendation rationale, and the rationale must land on specific fields—not “he is very suitable,” but “his research direction is a certain subfield, and in recent years he has produced results on a certain type of project.” If the rationale cannot land on fields, the recommendation is still at the level of impression.
02 Impression-Based Recommendations vs. an Explainable List
Place the two approaches side by side, and the difference becomes much clearer.
▲ Figure 2 | The difference is not in the candidates themselves, but in whether you can explain it clearly to the editor
The impression-based approach has five weaknesses that cannot be avoided. Professional grounds cannot be stated precisely—you can only say “I remember someone worked on a related direction”; sources cannot be traced—you cannot recall whether you saw it in a paper or at a conference; conflict screening relies on memory, and the more people there are, the easier it is to miss someone; when explaining to the editor, you can only answer “quite suitable”; and when someone else reviews it afterward, the result may differ from yours.
An explainable list addresses each of these points. MuchSci profiles fix the grounds in fields: professional grounds come from item-by-item comparison of research areas and projects led; sources are supported by the profile’s data source labels and profile identifier, so provenance is not lost; conflict screening filters item by item according to institution and employment history rather than impression; when explaining to the editor, you can directly point out which column the grounds fall in; and during later review, every step can be traced back to the original public materials and walked through again.
That is what “explainable” means in practice: if someone else checks using the same grounds, they can find the same things. This is being accountable both to the editor and to yourself.
03 One Recommendation Rationale, Supported by Three Layers
A rationale that can withstand follow-up questions is actually underpinned by three layers.
▲ Figure 3 | From top to bottom: the field matches, the employment history can be verified, and the source can be traced back
The top layer is the field criterion, answering “is the specialty a match?” It relies on research area fields and projects led. The middle layer is verifiable employment history, answering “is the current position true?” Employment history lists institution and position by date, and can be checked against the institution’s public information. The bottom layer is traceable sourcing, answering “where do the grounds come from?”, relying on the profile identifier and data source labels; records marked “pending verification” remind you that this item has not yet been fully compared.
If any one of the three layers is missing, the recommendation rationale cannot withstand follow-up questions: with only the field and no employment history, you cannot clearly say where the person is now or what they do; with only employment history and no source, others cannot verify whether what you say is correct. Only when all three layers are laid out does the recommendation change from “I think” to “checkable.”
One point about the eligibility design should be noted here: the Talent Profile includes only records related to national-level projects, and requires names, institutions, and employment history to match public materials precisely. Records for the same project under different names are marked “Pending Verification” and are not counted as experts. This “better to leave out than include wrongly” orientation is meaningful when used for recommending reviewers—having one or two fewer people on the list does not matter, but one extra person who does not match would be very hard to explain.
04 Functional Interface: Correspondence Between Profiles and Recommendation Rationales
The interface is under the “Talent Profile” tab of the Agent Assistant.
▲ Figure 4 | Basic Information and Education: a summary plus a timeline, with the profile identifier and data source written at the top
The top line is this mechanism’s public statement: profile identifier, data source, and an explanation of the inclusion scope. The criteria being written at the top of the profile means that when users write recommendation rationales, they can always explain how their grounds were derived. Below are Employment History and Research Achievements:
▲ Figure 5 | Employment History lists institution and position by date; Research Achievements are divided into Academic Contributions and Projects Led
Writing a recommendation rationale is actually annotating these columns. The research area in Basic Information corresponds to “direction match”; the institution and position in Employment History correspond to “current position verifiable”, and they are also the places that must be checked item by item when screening for conflicts of interest; the projects led in Main Research Achievements correspond to “what has been done recently”. Every sentence of the recommendation rationale should be able to point back to a certain column. Sentences that cannot be pointed back should preferably be deleted.
Two things need emphasis: conflict avoidance and confidentiality. The profile can help you lay out public information in order, but how conflicts of interest are defined and within what scope reviewer information may be used must follow the current rules of the journal or conference; this is not something the profile can decide.
05 Boundaries
① This article discusses functional mechanisms and application methods and does not constitute a promise of outcomes. The completeness of a profile depends on the availability of public materials; the inclusion scope is limited, and this does not mean that complete grounds can be given for all candidates.
② Structuring does not equal evaluation. A profile is the organization and presentation of public materials and does not constitute an evaluation of anyone’s abilities, a recommendation, or an endorsement; this article also does not constitute any review advice. Judgments involving specific individuals must be independently verified by the user, who bears the responsibility.
③ Fairness and confidentiality in review are subject to the current rules of the journal or conference. Only public materials are used, with no personal privacy data involved; the determination of conflict-of-interest avoidance, the use of reviewer lists, and confidentiality requirements should all follow the current rules and procedures of the relevant editorial office or conference organization.