{"id":"95b9c841-2e9f-425f-85c0-c5ea860a7bb1","arxiv_id":"2606.21665","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"Large-scale analysis of Bluesky shows online participation is structured by low-vocality practices like liking that sustain high-intensity engagement, separate from posting which dominates low-intensity use.","lead":"The paper analyzes billions of Bluesky activity records and finds that high-intensity engagement often comes from liking rather than posting, while low-vocality actions sustain participation even when posting is rare. A smart generalist might read it to see why counting only visible posts gives a misleading picture of who is actually present on social platforms.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Sample construction and potential bias from registration/data availability patterns remain unvalidated for low-vocality capture","rationale":"The reader's weakest_assumption directly identifies the load-bearing empirical precondition for the strongest_claim. Because the supplied abstract provides no further methodological detail, the concern stands; full-text access would allow verification but does not remove the need for the concrete bias check.","tokens_in":1768,"tokens_out":314,"duration_ms":16397,"concrete_test":"Reconstruct the sampling frame from the methods section; compare registration-date distributions, total activity volume, and action-type ratios (likes vs posts) between the 80% sample and any available metadata on the excluded 20%; if the excluded cohort shows systematically lower intensity or different style, recompute the transition matrices and motif frequencies on a reweighted sample.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim—that differentiated low-vocality practices structure participation and require shifting from post-centered measures—rests on the monthly intensity/style dimensions being unbiased representations of all engagement. The abstract asserts a near-complete sample (>80% of registered users, >3B records) but supplies no description of sampling frame construction, exclusion criteria for the remaining ~20%, or tests for differential availability (e.g., whether non-sampled users skew toward zero-activity or unregistered low-vocality profiles). Without such checks, the observed concentration of vocal production and the attractor status of high-intensity likers could be artifacts of who registers and whose actions are logged.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper claims that analyzing over three billion activity records from a near-complete sample (>80% of registered Bluesky users) allows reconstruction of monthly user-level participation profiles distinguishing intensity (amount of engagement) from style (expression across actions like liking vs. posting). It reports that vocal production is concentrated, high-intensity engagement associates more with liking than posting, high-intensity likers and posters act as attractors in transitions, network-building redirects within active space, inactivity is a persistent boundary, and higher-order motifs show inactivity interrupting regimes with low-intensity liking sometimes preceding durable high-intensity engagement. The central conclusion is that differentiated low-vocality practices structure online participation, requiring a shift from post-centered measures to dynamic accounts of platform presence.","tokens_in":1892,"tokens_out":600,"duration_ms":23065,"significance":"If the sample and dimension construction prove robust, the work would be significant for computational social science by providing large-scale empirical evidence that low-vocality actions (liking, following) are central to sustained participation rather than mere supplements to posting. The scale of the dataset (>3B records) and use of transition patterns plus higher-order motifs represent strengths that could enable falsifiable, dynamic models of platform presence if the underlying measures are validated.","major_comments":[{"comment":"Data and Methods (inferred from abstract claims): The assertion of a 'near-complete sample accounting for more than 80% of registered users' supplies no description of the sampling frame construction, exclusion criteria for the remaining ~20%, or any tests for differential data availability/registration bias that could systematically under-represent low-vocality profiles. This is load-bearing for all downstream claims about low-vocality practices structuring participation.","section":"Data and Methods"},{"comment":"Intensity and Style Dimensions (abstract): The paper aggregates into monthly intensity and style dimensions and reports associations (e.g., high-intensity with liking) but provides no details on how these dimensions were defined, validated against external criteria, or tested for measurement error. Without such validation, the distinction between intensity and style and the claim that they are 'often conflated' cannot be evaluated.","section":"Intensity and Style Dimensions"},{"comment":"Transition Patterns and Motifs (abstract): The reported transition patterns (attractors, boundaries) and higher-order motifs lack any mention of error estimates, confidence intervals, robustness checks across aggregation windows, or sensitivity to the monthly aggregation choice. These are central to the attractor and 'inactivity as persistent boundary' claims.","section":"Transition Patterns and Motifs"}],"minor_comments":[{"comment":"The abstract is lengthy and packs many distinct findings into a single paragraph; breaking out the core empirical results more concisely would improve readability.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive comments, which identify key areas where additional methodological detail will strengthen the paper. We will revise the manuscript to expand the Data and Methods sections with the requested information and checks while preserving the core empirical claims.","responses":[{"response":"We agree that the current description is insufficiently detailed. In the revised manuscript we will add a dedicated subsection describing the Bluesky data acquisition process, the construction of the sampling frame from platform registration and activity logs that yields coverage of more than 80% of registered users, the explicit exclusion criteria applied to the remaining accounts, and any available checks or discussions of potential registration or availability biases, including whether low-vocality profiles appear differentially affected.","revision_made":"yes","referee_comment":"[Data and Methods] The assertion of a 'near-complete sample accounting for more than 80% of registered users' supplies no description of the sampling frame construction, exclusion criteria for the remaining ~20%, or any tests for differential data availability/registration bias that could systematically under-represent low-vocality profiles. This is load-bearing for all downstream claims about low-vocality practices structuring participation."},{"response":"We will expand the Methods section to supply the precise operational definitions of the intensity dimension (total monthly actions) and style dimension (distribution of action types, e.g., proportion of likes versus posts), the aggregation formulas used, and any internal consistency or external validation steps performed. We will also report basic measurement-error diagnostics and discuss how these choices support the intensity-style distinction.","revision_made":"yes","referee_comment":"[Intensity and Style Dimensions] The paper aggregates into monthly intensity and style dimensions and reports associations (e.g., high-intensity with liking) but provides no details on how these dimensions were defined, validated against external criteria, or tested for measurement error. Without such validation, the distinction between intensity and style and the claim that they are 'often conflated' cannot be evaluated."},{"response":"We will add error estimates and confidence intervals to all reported transition probabilities and motif frequencies. We will further include sensitivity analyses that vary the temporal aggregation window (bi-weekly and quarterly) and report how the main attractor, boundary, and motif findings change or remain stable under these alternatives.","revision_made":"yes","referee_comment":"[Transition Patterns and Motifs] The reported transition patterns (attractors, boundaries) and higher-order motifs lack any mention of error estimates, confidence intervals, robustness checks across aggregation windows, or sensitivity to the monthly aggregation choice. These are central to the attractor and 'inactivity as persistent boundary' claims."}],"tokens_in":1574,"tokens_out":525,"duration_ms":26658,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The core observation here is that a near-complete sample of Bluesky activity reveals vocal posting is concentrated while low-vocality actions like liking sustain high-intensity participation, and that transitions and motifs point to inactivity as a selective barrier rather than a full reset.\n\nThe work does a solid job with scale: more than three billion records covering over 80 percent of registered users lets them aggregate monthly profiles and track shifts between intensity-style combinations. Distinguishing how much activity occurs from which actions express it is a straightforward and useful cut on the data, and the motif findings on how low-intensity liking can lead into higher regimes add some temporal structure.\n\nThe soft spots are in the methods details that are missing from the abstract. There is no description of how the intensity and style dimensions were defined or checked for robustness, and the stress-test concern about the remaining 20 percent of users is fair—without tests for differential registration or logging bias, the reported concentration of vocal behavior could partly reflect who shows up in the sample. Everything stays descriptive, so the attractor language for high-intensity likers and posters is interpretive rather than tested against alternatives.\n\nThis is for computational social science groups that track platform metrics and want platform-specific patterns on a newer service. A reader working on engagement measurement would find the scale and the low-vocality emphasis worth seeing. It deserves peer review because the data volume is real and the questions are on-topic, even though the validation steps will need attention in revision.","headline":"The paper uses a large Bluesky dataset to separate engagement intensity from style and shows liking dominates high-intensity use while posting is more common at low intensity.","tokens_in":2402,"tokens_out":373,"would_cite":false,"duration_ms":14184,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Online participation on Bluesky is structured by low-vocality practices like liking rather than by posting alone.","keywords":["online participation","low-vocality engagement","Bluesky","user intensity","engagement style","social media analytics","platform presence","activity transitions"],"falsifier":"A replication on another platform that finds posting frequency predicts long-term retention and re-entry better than liking frequency would falsify the claim that low-vocality practices primarily structure participation.","tokens_in":2658,"feed_emoji":"📱","tokens_out":674,"duration_ms":19773,"temperature":0.7,"pith_summary":"The authors reconstruct participation profiles from over three billion activity records covering more than 80 percent of registered Bluesky users. They separate monthly user behavior into an intensity dimension that measures how much engagement occurs and a style dimension that captures how it is expressed through different actions. Vocal posting turns out to be highly concentrated while high-intensity users tend to favor liking over posting, and low-intensity users show the reverse pattern. Transition data indicate that high-intensity likers and posters act as attractors within the active space and that inactivity forms a selective barrier to re-entry. The work therefore argues that platform presence must be tracked dynamically rather than through visible posts alone.","feed_headline":"Liking drives sustained engagement more than posting on Bluesky","feed_subtitle":"Billions of actions show high-intensity users favor low-vocality behaviors and that post counts miss most participation patterns.","key_machinery":"The two-dimensional decomposition of participation into intensity (volume of actions) and style (distribution across posting, liking, and following) applied to monthly user aggregates.","core_discovery":"By aggregating three billion records into monthly user-level observations and separating intensity from style, the study establishes that vocal production is highly concentrated, high-intensity engagement associates most strongly with liking, posting-oriented participation appears more often among low-intensity users, network-building redirects users inside the active space, and inactivity serves as a persistent boundary, while higher-order motifs show that low-intensity liking can precede durable high-intensity regimes and that inactivity interrupts rather than erases prior patterns.","pith_inferences":["Platform designers could test whether surfacing low-vocality signals improves retention compared with post-centric feeds.","The intensity-style decomposition offers a template for re-analyzing existing datasets from other platforms without new data collection.","Computational models of user churn might incorporate motif-level interruption patterns to forecast re-entry probabilities more accurately."],"forward_implications":["Vocal production remains highly concentrated even in a near-complete user sample.","High-intensity engagement correlates more strongly with liking than with posting.","Posting-oriented behavior occurs more frequently among low-intensity users.","High-intensity likers and posters function as attractors that retain users within active participation.","Network-building actions redirect users inside the active space while inactivity selectively limits re-entry."],"fun_headline_variants":["Bluesky high-intensity engagement tied to liking not posting","Low-vocality practices structure online participation on Bluesky","Posting concentrated while liking sustains high engagement on Bluesky","Inactivity blocks re-entry but liking can lead to intense Bluesky use","Intensity and style separate vocal and low-vocality participation on Bluesky"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The near-complete sample of more than 80 percent of registered users and the monthly aggregation into intensity and style dimensions accurately reflect all forms of engagement without systematic bias from platform data availability or user registration patterns.","fun_headline_variants_meta":{"raw":{"variants":["Bluesky high-intensity engagement tied to liking not posting","Low-vocality practices structure online participation on Bluesky","Posting concentrated while liking sustains high engagement on Bluesky","Inactivity blocks re-entry but liking can lead to intense Bluesky use","Intensity and style separate vocal and low-vocality participation on Bluesky"]},"model":"grok-4.3","cost_usd":0.007603,"raw_usage":{"total_tokens":3429,"prompt_tokens":722,"num_sources_used":0,"completion_tokens":83,"cost_in_usd_ticks":76028000,"prompt_tokens_details":{"text_tokens":722,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2624,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":722,"tokens_out":83,"duration_ms":21997,"temperature":1.0,"reasoning_tokens":2624,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-26T12:16:18.936906+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A replication on another platform that finds posting frequency predicts long-term retention and re-entry better than liking frequency would falsify the claim that low-vocality practices primarily structure participation.","supporting_citations":[],"review_version":1}