{"id":"56b709f9-9b5e-44ea-b724-92e50328ab9f","arxiv_id":"2611.17018","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In an online survey with imagined group scenarios, virtual cat and metaphor feedback were rated less embarrassing and more comfortable than direct text, without reducing perceived usefulness.","lead":"Researchers tested three ways of showing feedback in virtual reality group conversations: a direct text message, a playful seesaw metaphor, and a virtual cat. In a survey, people tended to rate the cat and metaphor as less embarrassing and more comfortable than text, though no one used the designs in a live conversation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"The reported inferential tests support the discomfort-reduction claim in only one of three scenarios, and no reported tests support the 'maintaining perceived usefulness and fairness' half of the central claim; the conclusion's scope exceeds its own data.","rationale":"The reader's CONDITIONAL verdict and its three conditions are sound, but my pass locates a different, less acknowledged gap. The vignette-to-live generalization problem identified by the reader is real yet explicitly acknowledged in Section 9 (b282), and the conclusion is carefully worded around perceptions; the authors have a standing defense there. The sharper problem is internal to the reported data. Reading Sections 5.2.1-5.2.2 closely, the inferential support for 'mitigate social discomfort' is significant for comfort in Scenarios 1 and 3 and for embarrassment in Scenario 1 only; Scenario 2 yields no reported significant contrast on either variable, and in Scenario 3 the Text-vs-Cat comfort gap is non-significant (p=.113) with no embarrassment omnibus effect. The cat probe is thus inferentially supported on embarrassment in exactly one scenario, and even the Scenario 1 comfort effect is small. The 'maintaining perceived usefulness and fairness' conjunct has no test behind it anywhere: usefulness is described via positive-response percentages and fairness via 'no clear differentiation,' and no equivalence analysis appears. The abstract's 'maintaining effectiveness' overclaims relative to a perception survey. None of this is acknowledged in the limitations, so it is not a disclosed limitation but an unstated scope gap. The concern does not question the direction of the effect or the integrity of the work: the randomized scenario ordering, the consistent descriptive pattern (Text_public worst on embarrassment and comfort in every scenario), the transparent limitations section, and the plausibly grounded design strategies all count in the paper's favor. The proposed concrete test is cheap, uses data already collected, and would settle whether the claim must be scoped down and which wording changes are required. This adds a scope-restriction condition to the reader's conditions, so the verdict remains CONDITIONAL rather than moving to ACCEPT or REJECT.","tokens_in":23560,"tokens_out":14150,"duration_ms":93800,"concrete_test":"Reanalyze the retained survey data by computing the full matrix of Friedman tests with Kendall's W and Bonferroni-corrected Wilcoxon post-hocs for all five dependent variables (usefulness, fairness, comfort, embarrassment, distraction) across all three scenarios under public delivery, explicitly reporting the currently absent Scenario 2 comfort/embarrassment tests and every usefulness/fairness test. Then run paired equivalence tests (TOST on the 5-point scale with a pre-specified margin, e.g., +-0.5) for Text vs Metaphor and Text vs Cat on usefulness and fairness under public delivery, and report 95% confidence intervals for all pairwise differences.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim (b297) is that metaphor- and cat-based 3D feedback 'can mitigate some of the social discomfort associated with public feedback while maintaining perceived usefulness and fairness.' For this to hold as a general finding across all three scenarios, the survey must show the probes reducing discomfort relative to public text without harming usefulness or fairness. The reported analysis supports only part of that. In Section 5.2.2, Friedman tests among the three public probes are significant for comfort in Scenario 1 (chi^2=6.561, p<.05) and Scenario 3 (chi^2=18.536, p<.001), and for embarrassment in Scenario 1 (chi^2=14.683, p=.001). No significant comfort or embarrassment contrast is reported for Scenario 2 (under-participation); in Scenario 3 embarrassment shows no omnibus effect, and the Text-vs-Cat comfort gap is non-significant (p=.113). The cat probe, one of the two claimed families, is inferentially supported on embarrassment only in the dominance scenario and on comfort only in Scenario 1. With 15 omnibus tests (3 scenarios x 5 DVs), only comfort in Scenario 3 and embarrassment in Scenario 1 survive a simple Bonferroni correction; the Scenario 1 comfort effect is small (Kendall's W is approximately 0.066 from chi^2=6.561, n=50, k=3), and no effect sizes or confidence intervals are reported anywhere. The other conjunct, 'maintaining perceived usefulness and fairness,' has no inferential support in the paper: Section 5.2.1 reports usefulness as descriptive positive-response percentages and fairness as 'no clear differentiation across scenarios' (b153). No Friedman or post-hoc test on usefulness or fairness under public delivery is reported, and 'maintaining' cannot be established from an unreported analysis. The abstract's stronger phrasing, 'maintaining effectiveness' (b5), additionally claims a behavior-level outcome that was never measured.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"This paper investigates how public feedback in XR-based small-group conversation can be designed to be socially acceptable. Grounded in three strategies (incorporate humor, reduce social attention, use indirectness), the authors propose Text, Metaphor, and Cat probes for three conversational scenarios: dominance, under-participation, and negative emotion. They report an online survey of 50 participants who rated usefulness, fairness, comfort, embarrassment, and distraction under public and private visibility, plus 14 follow-up interviews. The manuscript's central claim is that metaphor- and cat-based 3D feedback can mitigate some of the social discomfort associated with public feedback while maintaining perceived usefulness and fairness. The reported inferential tests support that claim only for a subset of scenarios and dependent variables, and the usefulness/fairness conjunct is supported only descriptively.","tokens_in":23993,"tokens_out":11480,"duration_ms":74729,"significance":"The paper addresses a real gap in the CSCW literature: prior group-feedback systems often deliver public feedback without studying its social costs, and the XR-specific design space is underexplored. The probe-based method is a reasonable first step, the three scenarios cover recurring group-conversation problems, and the qualitative analysis produces concrete design directions (e.g., decoupling feedback from individuals, graduated delivery, personalization). The authors deserve credit for reporting exact chi-square and post-hoc p values and for an unusually explicit limitations section that acknowledges the vignette-based, receiver-only nature of the study. If the claims are re-scoped to the scenario-dependent evidence, the paper is a useful design-space contribution for future XR feedback systems; as currently worded, the headline claim exceeds what the data establish.","major_comments":[{"comment":"The conclusion that \"metaphor- and cat-based 3D feedback can mitigate some of the social discomfort associated with public feedback\" is broader than the reported inferential results. Significant Friedman effects among the three public probes are reported only for comfort in Scenario 1 (χ²=6.561, p<.05), comfort in Scenario 3 (χ²=18.536, p<.001), and embarrassment in Scenario 1 (χ²=14.683, p=.001). No significant comfort or embarrassment contrast is reported for Scenario 2, Scenario 3 embarrassment has no reported omnibus effect, and the Text-vs-Cat comfort comparison in Scenario 3 is non-significant (p=.113). With 15 omnibus tests across the three scenarios and five dependent variables, a simple Bonferroni correction leaves only the Scenario 3 comfort effect and the Scenario 1 embarrassment effect. Please re-scope the conclusion to the dominance scenario for both probe families and, for comfort only, to the metaphor probe in the negative-emotion scenario, or supply additional evidence for the broader claim.","section":"§5.2.2 and §10 (conclusion, b297)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The second conjunct of the central claim, \"while maintaining perceived usefulness and fairness,\" has no inferential support. Section 5.2.1 reports only descriptive percentages for usefulness and fairness, with no Friedman tests or post-hoc comparisons for these variables, and \"no clear differentiation\" is not a statistical demonstration of maintenance. Please report the corresponding tests for usefulness and fairness (ideally with equivalence or non-inferiority bounds if \"maintaining\" is meant literally), or remove that conjunct from the headline claim.","section":"§5.2.1 and §10 (b297)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"No effect sizes, confidence intervals, or Kendall's W values are reported for any of the significant effects. For the Scenario 1 comfort effect, W = χ²/(n(k−1)) = 6.561/(50×2) ≈ .066, which is small; this matters because \"some of the social discomfort\" in the central claim is a magnitude claim. Please report effect sizes and confidence intervals for the Friedman and post-hoc tests, and state explicitly whether the study is confirmatory or exploratory given the number of comparisons.","section":"§5.2.2","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The statement that \"public delivery elicited higher embarrassment and lower comfort than private delivery for each design\" is presented as a consistent result, but no inferential tests for the public-versus-private contrasts are reported; please either add the tests or label the observation as descriptive.","section":"§5.1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Section 4.2 describes participants as watching \"a short video depicting the situation,\" whereas Section 9 states they \"evaluated visual sketches combined with text descriptions\"; please clarify whether the probes were presented as videos, static images, or storyboarded text descriptions.","section":"§4.2 / §9","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The sentence \"The interview consisted of three parts, as shown in appendix C\" appears twice in consecutive sentences; please remove the duplicate.","section":"§6.2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"There is a stray ').' in \"56% in Scenario 3)., while Metaphor...\"; please fix the punctuation.","section":"§5.2.3","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The limitations section contains a typo, \"there was s time gap\"; it should read \"there was a time gap.\"","section":"§9","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The Friedman statistics are reported without degrees of freedom; since all tests use k=3 conditions, please add df=2 (or the equivalent) to the chi-square reports.","section":"§5.2.2","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"It is a well-scoped probe study, not a breakthrough: the authors synthesize three known strategies (humor, reduced attention, indirectness), build three public-feedback probes per scenario, and compare them against direct text under public/private visibility. The mapping from strategy to probe is transparent, the scenario selection is standard, and the limitation section is genuinely candid. The empirical core is real but partial. The strongest consistent pattern is that public text produced the worst comfort/embarrassment ratings in all three scenarios; the inferential tests support the discomfort-reduction claim in Scenario 1 for both metaphor and cat, partly in Scenario 3 for metaphor, and essentially not in Scenario 2. The second half of the central claim, maintaining usefulness and fairness, is descriptive only and cannot be verified without the reported tests. The abstract's wording goes a bit beyond what the data show. There is a notable mismatch between the conclusion and the actual reported statistics: the supporting analysis is drawn mostly from Scenario 1, with no reported effect sizes or confidence intervals. This should not be rejected; it is a genuinely useful design-space exploration that deserves peer review. The authors openly acknowledge the vignette-to-live-XR gap, the receiver-only perspective, and the limited generalizability, so those are not new findings. What is missing is effect sizes and a more careful framing of the central claim.","headline":"Solid design-space exploration with honest methods, but the headline claim about reduced discomfort understates how thin the inferential support is outside Scenario 1.","tokens_in":24561,"tokens_out":705,"duration_ms":7970},"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"Public feedback in XR group conversations can be made less embarrassing by delivering it through visual metaphors or a virtual cat rather than direct text, while keeping it useful and fair.","keywords":["public feedback","social acceptability","embarrassment","extended reality","small-group conversation","embodied agent","visual metaphor","inclusive conversation"],"falsifier":"Run a live small-group study in XR with three real participants per session and deliver the three feedback forms publicly during an actual conversation, comparing the receiver's self-reported and behavioral signs of embarrassment (gaze avoidance, speech disruption, willingness to continue) against the survey predictions. If public text and public metaphor or cat probes produce indistinguishable embarrassment in the live setting, or if group members consistently fail to understand the metaphor and cat cues as feedback, the paper's central claim is refuted; a minimal version needs one scenario (a participant who dominates) and one comparison (text versus cat).","tokens_in":25046,"feed_emoji":"🐈","tokens_out":9841,"duration_ms":78938,"temperature":0.7,"pith_summary":"The paper claims that public feedback about group dynamics in extended-reality (XR) conversations can be made socially acceptable: wrapping the message in a visual metaphor or an embodied agent softens the embarrassment of being singled out while keeping the feedback useful. Drawing on three literature-derived strategies (incorporate humor, reduce social attention, use indirectness), the authors designed three feedback forms, direct text, metaphor probes (a tilting seesaw, a blooming sunflower), and a virtual cat, and tested them in three conversational scenarios: someone dominating, someone under-participating, and someone in negative emotion. In an online survey of 50 participants, public text drew the highest embarrassment (60–80%) and lowest comfort (26–32%) ratings in every scenario, while metaphor and cat probes received significantly better comfort or embarrassment ratings in at least one scenario each, with usefulness and fairness preserved. Fourteen follow-up interviews explained why: humor works when playful but not trivializing, attention is best reduced by decoupling the cue from the individual, and indirectness must stay interpretable. The stakes are practical: if this holds, group-conversation support systems in XR can deliver corrective feedback in front of everyone without making the recipient the object of public discomfort.","feed_headline":"A seesaw or a virtual cat softens the sting of public feedback","feed_subtitle":"Direct text embarrassed recipients most; metaphor and cat probes kept usefulness while easing discomfort.","key_machinery":"The carrying mechanism is a set of design probes: three feedback forms, direct text (the baseline), a metaphor, and a virtual cat agent, each rendered as 3D content in XR and applied to three conversational scenarios (domination, under-participation, negative emotion). The metaphor probe maps participation imbalance onto a tilting seesaw and emotional distress onto a blooming sunflower; the cat probe conveys reminders through behavior (knocking objects over, turning away and looking back, headbutting for petting). These forms operationalize three literature-derived strategies for reducing embarrassment: incorporating humor, reducing social attention (by redirecting gaze to an object or agent rather than the individual), and using indirectness (conveying the message through a symbol or action rather than explicit words). The empirical work is carried by a two-phase design: an online survey in which 50 participants watched short video vignettes and rated each probe on usefulness, fairness, comfort, embarrassment, and distraction under both public and private visibility, analyzed with Friedman tests and Wilcoxon signed-rank post-hoc comparisons; and 14 follow-up interviews using thematic analysis to interpret the ratings and elicit design extensions.","core_discovery":"On the paper's own terms, the central discovery is that the social cost of public feedback is not fixed by its visibility; the form the feedback takes shapes how much exposure the recipient feels. Across three scenarios (someone dominating, someone under-participating, someone in negative emotion), the direct text baseline shown publicly received the highest embarrassment ratings (60–80% agreement) and the lowest comfort ratings (26–32%) in every scenario. The seesaw metaphor (tilting toward the dominant speaker, rising on the under-participant's side), the sunflower metaphor for emotional support, and the cat agent's behaviors (knocking objects over, turning away and looking back, headbutting for petting) fared better: Friedman tests found significant differences among the three public probes on comfort in Scenario 1 ($\\chi^2 = 6.561$, $p < .05$) and Scenario 3 ($\\chi^2 = 18.536$, $p < .001$), and on embarrassment in Scenario 1 ($\\chi^2 = 14.683$, $p = .001$); post-hoc comparisons showed text rated significantly less comfortable than metaphor and cat in Scenario 1, less comfortable than metaphor in Scenario 3, and more embarrassing than both in Scenario 1. Perceived usefulness and fairness stayed positive across the probes, so the reduction in discomfort did not come at the cost of effectiveness. Follow-up interviews traced the effect to the three design strategies and showed that each requires calibration: humor that is playful yet appropriate, attention reduction achieved by disconnecting the cue from the individual, and indirectness balanced against interpretability.","pith_inferences":["If the survey perceptions transfer to live settings, the classic trade-off between public and private feedback partly dissolves: the operative design variable is not just who sees the cue but how face-threatening its form is, so an indirect public cue could combine shared awareness with individual comfort.","Because only the receiver's perspective was measured, an open question is whether bystanders actually decode a seesaw tilt or a cat's gesture as feedback about a specific person; if they do not, the comfort gain may come at the cost of the collective awareness that makes public feedback effective, and a spectator-decoding study would settle it.","The specific probes suggest a general principle: participation and emotional state can be rendered as ambient, object-level conditions rather than statements about people; the metaphors participants themselves proposed (water level, hourglass, thermometer, environmental lighting) could be tested to see whether the comfort benefit attaches to indirectness itself or to particular objects."],"forward_implications":["Public feedback does not have to be abandoned in XR group-conversation tools: metaphor- and agent-based forms preserve the perceived usefulness and fairness of public delivery while lowering the embarrassment it causes.","The three strategies function as design levers, but each carries a documented trade-off: humor must stay appropriate to the context, attention reduction requires actually decoupling the cue from the individual, and indirectness must remain interpretable.","Feedback visibility should adapt to context: participants preferred public delivery for positive reinforcement and familiar groups, and private delivery for sensitive situations like stress-driven under-participation, so fixed visibility rules will not fit every conversation.","Gradual delivery (starting with a soft cue and escalating only if the behavior continues) and letting users choose or personalize the feedback agent were the concrete mechanisms participants proposed for keeping public feedback comfortable."],"supporting_citations":[{"why":"Supplies the overt text-directive baseline messages used in Scenarios 1 and 2 and the finding that subtle directives are more positively accepted than overt ones.","marker":"[72]"},{"why":"Motivates the choice of a virtual cat as the embodied agent for emotional support in XR.","marker":"[52]"},{"why":"Shows the seesaw as an intuitive visual metaphor for identifying who dominates a conversation; the seesaw probe builds on it.","marker":"[59]"},{"why":"Provides evidence that explicit social attention amplifies awkwardness, grounding the reduce-social-attention strategy.","marker":"[11]"},{"why":"Documents the spotlight effect, the basis for predicting that redirecting attention lowers perceived exposure.","marker":"[27]"},{"why":"Documents humor as a remedial strategy in embarrassing interactions among equal-status people, grounding the humor strategy.","marker":"[23]"},{"why":"Establishes embarrassment as a distinct emotional response to public evaluation, motivating the embarrassment measure and the humor strategy.","marker":"[36]"},{"why":"Shows that embarrassment can be elicited in virtual meetings, justifying the study of social discomfort in XR group conversations.","marker":"[77]"}],"fun_headline_variants":["Seesaw and cat feedback take the sting out of public criticism","Metaphor-based 3D feedback eases small-group conversation discomfort","Direct text embarrasses most; metaphor and cat probes soften feedback","Virtual cat and seesaw make public feedback socially acceptable"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The load-bearing premise is that what participants say they would feel while watching short video vignettes and imagining themselves in an XR conversation matches what they would actually feel as the feedback receiver in a live multi-user XR conversation, a premise the authors acknowledge was not tested because participants evaluated visual sketches with text descriptions rather than a real-time interactive setting.","fun_headline_variants_meta":{"raw":{"variants":["Seesaw and cat feedback take the sting out of public criticism","Metaphor-based 3D feedback eases small-group conversation discomfort","Direct text embarrasses most; metaphor and cat probes soften feedback","Virtual cat and seesaw make public feedback socially acceptable"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":0.000136,"raw_usage":{"cost":0.0001363544,"is_byok":false,"cost_details":{"upstream_inference_cost":0.0001363544,"upstream_inference_prompt_cost":3.29784e-05,"upstream_inference_completions_cost":0.000103376},"total_tokens":1728,"prompt_tokens":1160,"completion_tokens":568,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":997,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":506}},"tokens_in":1160,"tokens_out":568,"cost_source":"openrouter_usage","duration_ms":6644,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":506,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-13T07:21:46.770011+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"Run a live small-group study in XR with three real participants per session and deliver the three feedback forms publicly during an actual conversation, comparing the receiver's self-reported and behavioral signs of embarrassment (gaze avoidance, speech disruption, willingness to continue) against the survey predictions. If public text and public metaphor or cat probes produce indistinguishable embarrassment in the live setting, or if group members consistently fail to understand the metaphor and cat cues as feedback, the paper's central claim is refuted; a minimal version needs one scenario (a participant who dominates) and one comparison (text versus cat).","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the overt text-directive baseline messages used in Scenarios 1 and 2 and the finding that subtle directives are more positively accepted than overt ones."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Motivates the choice of a virtual cat as the embodied agent for emotional support in XR."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows the seesaw as an intuitive visual metaphor for identifying who dominates a conversation; the seesaw probe builds on it."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides evidence that explicit social attention amplifies awkwardness, grounding the reduce-social-attention strategy."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the spotlight effect, the basis for predicting that redirecting attention lowers perceived exposure."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents humor as a remedial strategy in embarrassing interactions among equal-status people, grounding the humor strategy."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes embarrassment as a distinct emotional response to public evaluation, motivating the embarrassment measure and the humor strategy."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that embarrassment can be elicited in virtual meetings, justifying the study of social discomfort in XR group conversations."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}