{"id":"9b57ae14-d4f5-4e3b-9f3e-ff1cfd19920c","arxiv_id":"2608.13532","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A co-design study with 11 people with disabilities shows that harassment protection in social VR must preserve users' social image, not just their safety.","lead":"People with disabilities helped design new anti-harassment tools for social virtual reality, revealing that many care more about keeping their social image than about using aggressive safety features. A smart generalist might read this to understand how safety features can backfire or be useless if they expose or embarrass the users they are meant to protect.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proximity effect is confounded by green-to-red color gradient: the danger-to-distance mapping was never counterbalanced, so the claim that proximity is a socio-spatial signal is not yet established.","rationale":"The reader's weakest assumption identified the same design feature, but treated it as a limitation that 'weakens' rather than invalidates the proximity result. I argue it is more load-bearing than the reader concluded: because the color gradient was perfectly confounded with distance in every session, the proximity manipulation has no clean operational definition. The acknowledged limitation in Section 5.4 does not resolve the confound; it simply states it. The paper's contribution explicitly includes proximity as a 'rich socio-spatial signal' and offers proximity-aware design implications, so this internal validity threat bears directly on a central part of the contribution. I am not calling for rejection: the qualitative self-presentation tradeoff claim is grounded in participant quotes and appears robust to the color issue, and the study's overall method is sound. However, the proximity component should be accepted conditionally on replication with a neutral or counterbalanced visualization, or the claims should be softened to acknowledge that distance and color were not separately measured. The proposed counterbalancing test is feasible, respects the study's existing apparatus, and would definitively settle whether the proximity gradient is an effect of spatial distance or of the red-to-green safety semantics.","tokens_in":29782,"tokens_out":11325,"duration_ms":124791,"concrete_test":"Re-run the scenario-based interview (Section 3.3.2) with the same protocol but counterbalance the color mapping across participants: one group sees the current green-to-red gradient, one sees the reversed gradient (intimate=green, public=red), and a third group sees neutral gray rings. If safety ratings and the 'personal distance as bottom line' theme do not differ across conditions, proximity itself is the driver; if they track the color (red rings always rated less safe), the reported proximity effect is a color artifact.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's quantitative and qualitative proximity findings are confounded by the color-coded visualization in Section 3.2.1: the four proxemic rings were always displayed with a green-to-red gradient, mapping the closest (Intimate) zone to red and the farthest (Public) zone to green. In the scenario-based interview (Section 3.3.2), participants rated safety at each distance in the presence of these colors, so the independent variable 'proximity' is perfectly correlated with hue (red=close, green=far). The significant effect of Proximity on perceived safety (Section 4.1.1) and the qualitative theme that Personal Distance is the 'bottom line' could therefore be artifacts of color semantics rather than physical distance. Section 5.4 explicitly acknowledges this possibility, but no counterbalancing, neutral visualization, or control condition was run, so the alternative explanation remains live. This matters because the paper's contributions include 'proximity as a rich socio-spatial signal' (Section 6) and the proximity-aware design recommendations in Sections 4.5 and 5.1. The self-presentation tradeoff claim is largely independent of this confound, but the proximity component of the central claim is not yet established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a co-design study with 11 people with disabilities (PWD) in social VR. The authors built a virtual cafe environment with four color-coded proxemic rings (Intimate, Personal, Social, Public), presented three scripted harassment scenarios (verbal, physical, environmental), and probed four protection mechanisms (Inform, Educate, Consent, Defense). They collected safety ratings at each distance and qualitative interviews, analyzing the data with ART ANOVA and thematic analysis. The paper claims that proximity operates as a rich socio-spatial signal through which PWD assess the target, intent, and severity of ableist harassment; that PWD hold divergent values (optimistic resilience vs. risk-averse caution); and that protection mechanisms function as a medium of self-presentation, creating a persistent tradeoff between safety and social image. Design recommendations follow: proximity-aware, intent-aware, context-aware, and pro-social mechanisms.","tokens_in":29998,"tokens_out":4823,"duration_ms":56822,"significance":"If the proximity findings are valid, the paper makes a valuable contribution: it is the first co-design study of PWD's protection needs in social VR, extends Hall's proxemics framework to hostile interaction, and offers concrete, actionable design recommendations. Methodological strengths include genuine co-design with PWD, dual coding with a reported Cohen's Kappa of 0.83, peer debriefing, a full study protocol in the appendix, and careful ethical safeguards. The self-presentation tradeoff finding is novel, richly supported by participant quotes, and largely independent of the proximity manipulation. However, the proximity-specific contribution is weakened by a confound between physical distance and the color-coded visualization, which I detail below.","major_comments":[{"comment":"The independent variable 'proximity' is confounded with the color-coded visualization. The four rings were always shown as Intimate=red, Personal=orange, Social=yellow, Public=green. The significant main effect of Proximity on perceived safety (F3,110 = 25.6, p < .001) and the qualitative theme 'Personal Distance as the Bottom Line' could therefore reflect red-danger/green-safe semantics rather than physical distance. Although Section 5.4 acknowledges this possibility, no counterbalancing, neutral visualization, or control condition was run, so the alternative explanation remains live. Because the paper's theoretical contribution ('proximity as a rich socio-spatial signal') and the Section 5.1 design recommendations hinge on this result, the claim as stated is not yet established. The authors should either provide additional evidence (e.g., re-coding interviews for explicit references to color vs. distance, or analyzing whether the pattern holds when participants describe the spatial layout), or substantially temper the proximity claims and reframe the findings as about the interpreted proxemic visualization rather than physical distance.","section":"Sec. 3.2.1, 4.1.1, 5.4"},{"comment":"The quantitative analysis is based on only 11 participants, with 12 ratings per participant (3 harassment types x 4 distances). The ART ANOVA is appropriate, but the post-hoc pattern--significant differences only between the intimate zone and other zones, plus public-vs-personal--is exactly what a red-to-green color priming effect would predict. Since Section 5.4 acknowledges the limitation but does not quantify or control it, the statistical claim in Section 4.1.1 should be presented as exploratory and explicitly hedged. The qualitative findings are more robust because participants articulated spatial reasoning in their own words, but even those interpretations occurred in the presence of the color-coded rings.","section":"Sec. 4.1.1 and 5.4"}],"minor_comments":[{"comment":"The abstract calls the fourth probe 'Combat' while the full text consistently uses 'Defense Mechanism.' Please reconcile the terminology.","section":"Abstract"},{"comment":"In the P3 quote, 'blocked or or expelled' contains a duplicated 'or.' Please correct the transcription.","section":"Sec. 4.2"},{"comment":"The phrase 'we realize that our sample' could be more precise in English; consider 'we recognize that our sample' to match standard scholarly phrasing.","section":"Sec. 5.4"},{"comment":"The caption reads 'they shared their experiences after each scenario'; it should be 'participants shared' for grammatical consistency with the rest of the caption.","section":"Fig. 3 caption"}],"recommendation":"major_revision","confidential_remarks":"The reader's report recommends acceptance, and I agree that the self-presentation tradeoff contribution is solid and novel. My concern is the proximity confound, which I believe prevents acceptance in the current form. A major revision that either supplies additional analysis to separate color effects from distance effects, or that explicitly reframes and tempers the proximity claims, would make the paper acceptable. I do not see fundamental circularity; the findings are grounded in participant data rather than assumed in the design. The discrepancy between the stated central claim and the current evidence for the proximity component is the main blocking issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Good paper with one real flaw. The novel, valuable contribution is the empirical finding that PWD treat protection mechanisms as a form of self-presentation and often prioritize social image over safety. That theme is well supported by participant quotes and holds together. The paper also does solid qualitative work: clear protocol, two coders with Kappa 0.83, peer debriefing, full appendix, and an ethics section that shows care. The four design probes are sensible and the co-design method fits the research question.\n\nThe soft spot is the proximity claim. The virtual environment always rendered the four zones with a green-to-red gradient, mapping intimate distance to red and public to green. That means hue and distance are perfectly correlated, and the significant safety-rating differences and the 'personal distance is the bottom line' theme could be partly driven by color semantics. The authors acknowledge this in Section 5.4 but ran no counterbalancing or neutral visualization. The qualitative quotes about embodied discomfort from closeness suggest the effect is not purely an artifact, but the quantitative result is definitely weakened, and the paper's contribution of 'proximity as a rich socio-spatial signal' is not as firmly established as the conclusion implies. This is fixable in a revision: rerun with neutral rings or a reversed gradient, or temper the proximity claims.\n\nMinor: the abstract says 'Defense' while the body says 'Defense Mechanism' — trivial. The sample is small and English-speaking, but the authors are explicit about that limitation.\n\nVerdict: worth engaging with. The social-image tradeoff finding is new to the accessibility and social VR literature and should see the light. A serious referee should ask for the proximity claims to be reined in or the confound addressed, but this is not a desk reject. I would accept with revisions.","headline":"A well-conducted co-design study with a genuinely new finding about social image, but the proximity claim is weakened by a color confound the authors acknowledge yet do not resolve.","tokens_in":30498,"tokens_out":2047,"would_cite":true,"duration_ms":22812,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"For disabled people in social VR, protection tools are judged by the social image they project, not just the safety they provide.","keywords":["co-design","people with disabilities","social virtual reality","harassment","ableism","proxemics","protection mechanisms","self-presentation"],"falsifier":"Observe behaviour in a live social VR community: if users who asked for subtle, image-preserving mechanisms instead activate conspicuous blocking and reporting when harassment actually occurs, the tradeoff claim would need revision.","tokens_in":29599,"feed_emoji":"🛡️","tokens_out":4043,"duration_ms":42276,"temperature":0.7,"pith_summary":"This paper reports a co-design study with 11 people with disabilities (PWD) who use social virtual reality (VR), asking what protection mechanisms they want against ableist harassment. The central finding is that protection mechanisms are not neutral safety utilities: they are social gestures through which PWD manage how others see them. Participants read the distance of an approaching avatar as a signal of whether harassment is targeted, intentional, and severe, and they chose mechanisms partly to preserve a desired social image, sometimes prioritizing that image over their own safety. The paper argues that effective protection design must therefore be proximity-aware and image-aware, protecting PWD without forcing disability disclosure or confrontation.","feed_headline":"Disabled VR users trade safety to protect social image","feed_subtitle":"People with disabilities weigh how safety tools make them look, not just how well they block harassment.","key_machinery":"The organizing device is Hall's Proxemics Theory, which divides interpersonal space into Intimate, Personal, Social, and Public distance zones. The study visualizes these as four colour-coded rings around the participant's avatar in a custom virtual cafe built for the study, then stages three harassment scenarios (verbal, physical, environmental) at each distance and presents four protection probes: Inform, Educate, Consent, and Defense. This setup lets the researchers map how harassment perception and protection preferences shift with proximity, and how each mechanism's visibility changes the social cost of using it.","core_discovery":"On the paper's own terms, the discovery is that in social VR, disability-targeted harassment is interpreted through a spatial-social lens, and protection mechanisms function as a medium of self-presentation. As a harasser's avatar moves from public to intimate distance, PWD read proximity as progressively disclosing the target, the ableist intent, and the urgency of response; the Personal Distance acts as a 'bottom line' where harassment becomes intolerable. Beyond proximity, PWD weigh behavioural effort and bystander reactions to judge severity. When choosing protections, PWD balance two competing values: personal safety and the social image they project, at times explicitly downplaying harassment or preferring subtle, system-level interventions to avoid looking oversensitive, vulnerable, or preachy.","pith_inferences":["Beyond the paper: the safety-image tradeoff likely generalizes to other stigmatized groups in embodied social spaces; the same co-design method could test whether women, queer users, or racialized users trade safety for social image when choosing moderation tools.","Beyond the paper: the finding that environmental harassment is rated safer partly because it is harder to recognize suggests a detection gap: future systems could flag memetic disability avatars even when the target user has not yet perceived the mockery.","Beyond the paper: if the colour-coded zones primed danger, then the study's quantitative proximity gradient may overstate how much distance alone drives perceived threat; a replication with neutral spatial markers would isolate the effect, and the paper's own limitation section invites this.","Beyond the paper: the preference for system-level, depersonalized interventions opens a concrete design target, a 'safety credit' or reputation signal, but such aggregation risks new surveillance and bias against neurodivergent or otherwise atypical users, a tension the participants did not fully resolve."],"forward_implications":["Proximity-based protection design: future safety systems should trigger graded responses that match distance, from ambient norm-setting at Public distance to automatic blocking and muting at Personal and Intimate distances.","Intent-aware, not just action-triggered, moderation: participants want systems that aggregate behavioral signals (repeated gestures, historical safety record) to tell hostile intent from accident or in-community joking.","Social-cost offloading: protection should be framed as system-level action rather than user-initiated confrontation, so the burden of intervention falls on the platform, not the PWD.","Discreet and privacy-preserving interventions: mechanisms should avoid visual clutter, use non-alarming colors, and handle conflicts in private bubbles that do not expose the PWD to an audience.","Normalizing disability representation: platforms could populate spaces with NPCs carrying disability signifiers to make disability presence routine and reduce the 'troll' misreading of disability-representing avatars."],"supporting_citations":[{"why":"Supplies Hall's Proxemics Theory and the four distance zones that structure the entire co-design apparatus.","marker":"[27]"},{"why":"Prior diary study documenting disability-signifier harassment in social VR and the positive interactions those signifiers invite; it defines the harassment types and motivates the need for protection.","marker":"[78]"},{"why":"Prior work on avatar disability signifiers and ableist hate that establishes the baseline problem of representation triggering harassment.","marker":"[2]"},{"why":"Foundational framing of harassment in social VR and its governance challenges, used to motivate the proximity-based social lens.","marker":"[6]"},{"why":"Meta's Personal Boundary, the current proactive safety feature that blocks all interaction, serves as the design baseline the co-design critiques.","marker":"[63]"},{"why":"Prior study on women's envisioned AI companions for harassment mitigation, informing the paper's AI-powered moderation design implications.","marker":"[19]"}],"fun_headline_variants":["Disabled VR users trade safety for social image in harassment tools","Co-designed VR anti-harassment: Disabled users weigh safety and image","Proximity signals ableist intent in VR harassment of disabled users","VR harassment: How disabled users balance safety and social standing","For disabled VR users, safety tools can damage social image"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Participants' reactions to scripted harassment in a researcher-built VR scene with colour-coded proxemic rings stand in for how they would perceive and respond to real harassment in live social VR.","fun_headline_variants_meta":{"raw":{"variants":["Disabled VR users trade safety for social image in harassment tools","Co-designed VR anti-harassment: Disabled users weigh safety and image","Proximity signals ableist intent in VR harassment of disabled users","VR harassment: How disabled users balance safety and social standing","For disabled VR users, safety tools can damage social image"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001113,"raw_usage":{"total_tokens":4607,"prompt_tokens":885,"completion_tokens":3722,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":501,"completion_tokens_details":{"reasoning_tokens":3644}},"tokens_in":501,"tokens_out":3722,"duration_ms":27650,"temperature":1.0,"reasoning_tokens":3644,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:47:58.947419+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Observe behaviour in a live social VR community: if users who asked for subtle, image-preserving mechanisms instead activate conspicuous blocking and reporting when harassment actually occurs, the tradeoff claim would need revision.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Hall's Proxemics Theory and the four distance zones that structure the entire co-design apparatus."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior work on avatar disability signifiers and ableist hate that establishes the baseline problem of representation triggering harassment."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Foundational framing of harassment in social VR and its governance challenges, used to motivate the proximity-based social lens."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Meta's Personal Boundary, the current proactive safety feature that blocks all interaction, serves as the design baseline the co-design critiques."}],"review_version":1}