{"id":"c1ea3ebc-48dc-4528-9da1-2137cd548f46","arxiv_id":"2505.09047","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"For right-eye monocular smart glasses, place the virtual image vertically centered and offset 8.7 to 23.7 degrees toward the ear, with a 15-degree field of view, for everyday text and glanceable content.","lead":"This paper reviews past studies on where to put the small screen in single-eye smart glasses and recommends placing it to the right of the wearer's straight-ahead gaze, about 9 to 24 degrees toward the ear, with a display width of 15 degrees. A generalist might read it because these concrete numbers give hardware designers a target that balances quick reading, low interruption, comfort, and social awkwardness.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The +8.7° inner-edge boundary is transferred from an unlit-combiner comfort study to the active-image recommendation without direct evidence; the exact 15°-FOV interval is therefore unsupported at its inner edge.","rationale":"The paper's headline contribution is a design guideline with specific angular bounds, and its correctness depends on each bound being grounded in evidence that supports that exact condition. The reader identified the same weakest point: Section 5.2.2's +8.7° is measured for an unlit combiner edge, not for the inner edge of an active virtual image, and Section 6 transfers it without an explicit test. I see no additional equally load-bearing flaw. The outer bound (+23.7°) is at least supported by the '+19.6° conservative' and 'up to 30°' active-display comfort studies, and the paper's own caveats about unpublished ongoing studies are stated. The concern is real but not fatal: it is a missing-evidence gap in an otherwise plausible synthesis, exactly the kind of thing a conditional verdict should flag. I agree with the reader's assessment and would not move the verdict to ACCEPT, REJECT, or UNVERDICTED. The paper is self-aware about the provisional status of its numbers, and the gap is localized to one boundary rather than undermining the whole argument structure.","tokens_in":22155,"tokens_out":2094,"duration_ms":18239,"concrete_test":"Run a within-subjects study (or re-analyze the authors' unpublished Kwok/Arora data, if complete) directly comparing a right-eye 15° active virtual image at inner-edge positions +8.7°, +10°, +15°, and +17.28° on a real waveguide or emulated see-through display, measuring interruption, comfort, reading performance, and social perception in a dual task. If comfort at +8.7° is statistically indistinguishable from larger offsets, the boundary supports the recommendation; if participants rate +8.7° as significantly more interruptive than +15°/+17.28°, the Section 6 interval is too close to the PPOG and the claimed 15° FOV is not supported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central recommendation (Section 6, Abstract) is a 15° FOV spanning +8.7° to +23.7° on the right eye. The left boundary +8.7° comes from Section 5.2.2, where Song and Arora [71,96] measured comfort of unlit optical-combiner edges in emulated glasses: edges should be outside -20.2° to +8.7° from the PPOG. Section 6 then uses +8.7° as the inner edge of the active display, stating it 'aligns with' Dowell's 8° cognitive-capture guideline, but no study in the paper directly tests an active virtual image whose inner edge sits at +8.7°.\n\nThis transfer is not obviously valid. First, the 5.2.2 studies used semi-transparent films to emulate combiners, not lit waveguide images; the percept is a static edge/transparency boundary, not a bright text region, so interruption, binocular-rivalry, and attentional-capture mechanisms differ. Second, an active display edge at 8.7° still leaves the image only 8.7° from the PPOG, far inside the ±25° eyes-only range cited from Sidenmark, and the interruption literature (Dowell et al. [27]) reports cognitive capture within an 8° radius for small HUD symbology, not a 15°-wide rectangular text field. Third, Chua et al.'s center-right position had its center near +17.28°, so its inner edge could have been closer to PPOG than +8.7°, but the paper does not report the relationship between Chua's position and the +8.7° boundary.\n\nThe consequence is specific: if the active-image inner edge actually should be farther out (e.g., closer to Chua's center at +17.28°, or beyond Dowell's 8° radius plus half-FOV), the recommended interval narrows or shifts, reducing usable FOV below the 15° claim. If it can be as close as +8.7°, the recommendation holds. The paper itself flags no direct test; Section 5.2.2 is explicitly about display-off comfort, and Section 6 combines it with an active-display comfort range (-24.6° to +19.6°) whose outer bounds come from different studies (Haynes, Song, Arora).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper consolidates recent and prior research on monocular optical see-through head-worn display (OST-HWD) placement, with the aim of deriving a concrete design guideline for everyday eyeglasses. After reviewing interruption, comfort, performance, and social perception studies, the authors recommend a 15° horizontal field of view for the virtual image in the right lens, vertically centered and offset horizontally from +8.7° to +23.7° toward the ear relative to the principal position of gaze (PPOG), with glanceable content allowed up to +30° for short interactions. The paper also includes sections on field-of-view trade-offs, inactive optical combiner comfort, social perception of lens tint, overlay capabilities, and optical architecture considerations (freespace and flat lightguide).","tokens_in":22538,"tokens_out":7669,"duration_ms":67935,"significance":"If the recommended interval (+8.7° to +23.7°) is accepted, this paper would provide one of the few concrete, quantitative targets for positioning monocular OST-HWDs in everyday glasses, which is genuinely useful for hardware designers. The paper performs a valuable service by synthesizing aviation HUD, automotive HUD, and wearable HWD studies into a single framework, and it explicitly raises less commonly considered constraints such as the appearance of an inactive optical combiner and social perception of downward gaze. The recommendation is falsifiable and clearly stated, and the paper honestly distinguishes published from ongoing/unpublished work. However, the central numeric boundaries are less strongly supported than the prose implies, because the inner-edge boundary (+8.7°) comes from an unlit-combiner study, the 15° FOV relies in part on an unpublished captioning study, and the paper's own active-display comfort data indicate a more conservative outer bound (+19.6°) for extended reading.","major_comments":[{"comment":"The inner edge of the recommended active-image interval is set at +8.7°, but the source study (Song and Arora et al. [71,96]) measured comfort of unlit optical-combiner edges in emulated glasses, not of a lit virtual image with text. The paper presents no direct evidence that the comfort boundary for an inactive combiner edge transfers unchanged to the inner edge of an active display. Because the entire left boundary of the final recommendation rests on this transfer, this is a load-bearing gap; the authors should either mark +8.7° as a provisional boundary or report an active-display test of inner-edge placement near 8°–12°.","section":"Section 6, with Section 5.2.2"},{"comment":"The 15° FOV recommendation is supported primarily by an ongoing, unpublished captioning study (Kwok et al.) whose 9° offset was itself chosen from Arora et al.'s +8.7° combiner-edge boundary; the paper later treats +8.7° as an independent constraint, making the argument partly circular. Moreover, Section 6 attributes a 15° FOV suggestion to \"Martin et al. and Britain et al.,\" but the cited reference [109] is the same paper as [70] (Britain et al.), which actually reports a preferred 20°–30° FOV for captioning. The cited published work therefore does not support the 15° value. The authors should correct this attribution and state explicitly which parts of the 15° recommendation depend on unpublished results.","section":"Section 4 and Section 6"},{"comment":"Section 5.2.2 concludes that optical combiner edges should be placed \"outside the range of -20.2° and +8.7°\" from PPOG, meaning an edge at +8.7° is on the boundary of the forbidden zone rather than outside it. The final recommendation places the inner edge of the virtual image at exactly +8.7°. The paper should clarify whether +8.7° is an acceptable location for an active-image edge (e.g., because the 8.7° value includes a safety margin beyond Dowell's 8° cognitive-capture radius) or whether the recommended interval should begin strictly beyond +8.7° (e.g., at +9° or +10°).","section":"Section 5.2.2 and Section 6"},{"comment":"The paper's own active-display comfort data (Haynes, Song, and Arora) identify +19.6° as the comfortable outer bound for extended reading and +30° for shorter interactions, yet the final recommendation allows a 15° FOV extending to +23.7° for text-based tasks. Since text-based tasks are typically prolonged (reading, captioning), the 15° recommendation exceeds the stated extended-use bound without additional justification. The authors should either reconcile the +23.7° outer edge with the +19.6° extended-use threshold or restrict the 15° FOV to shorter interaction contexts.","section":"Section 5.2.1 and Section 6"}],"minor_comments":[{"comment":"References [70] and [109] appear to be the same paper (Britain et al., \"Preferences for captioning on emulated head worn displays while in group conversation\"); duplicate references and the unexplained \"Martin et al.\" attribution in Section 6 should be corrected.","section":"References [70] and [109]"},{"comment":"Sections 9.2 and 9.3 refer to \"Figure 1 (left)\" and \"Figure 1 (right)\" as examples of freespace and flat lightguide architectures, but Figure 1 shows photographs of Google Glass and the Vuzix Z100; the intended figures appear to be Figure 14 and Figure 15. The cross-references should be fixed.","section":"Section 9.2 and 9.3"},{"comment":"The word \"contraints\" in Section 4 (paragraph on high FOV and all-day wear) is a typo for \"constraints.\"","section":"Section 4"},{"comment":"The paper inconsistently uses \"principal point of gaze\" in the abstract and \"Principal Position Of Gaze (PPOG)\" in Section 3.3; the terminology should be unified.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper is heavily self-cited, and several load-bearing results (Kwok et al., the vertical-offset social perception study, and to some degree the 15° FOV recommendation) are described as ongoing or under review; the editor may wish to verify how much of the contribution rests on unpublished work and whether the authors can be asked to weaken the precision of the headline recommendation accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a useful consolidation. The paper takes a scattered literature—aviation HUDs, automotive displays, reading studies, social perception, optics—and lands on a specific, actionable recommendation: right-eye display, vertically centered, +8.7° to +23.7° from gaze, 15° FOV, glanceable to 30°. That's a number a hardware team can design to. The writing is clear, the lit review is broad and mostly careful, and the optics/frame section is a nice addition that most HCI surveys omit. If you work on smart glasses or captioning, you'll want this on file.\n\nThe main soft spot is exactly the one the stress-test flags. Section 5.2.2 concerns comfort when the display is off: the -20.2°/+8.7° edge bounds come from an unlit combiner appearing as a smudge/edge. Section 6 then uses +8.7° as the inner edge of an active virtual image, citing Dowell's 8° cognitive-capture guideline. But no experiment in the paper tests a lit image with an inner edge at +8.7°. The mechanisms are plausibly different (static edge vs. bright text), and a 15°-wide image starting at +8.7° means the display spans nearly the entire ±25° eyes-only range. It may well be fine, but it's unsupported. That's not fatal if the paper is read as a provisional design guideline, and the authors do disclose that Kwok et al. is ongoing. Still, the recommendation is presented in the abstract with no caveat, which overstates the evidence.\n\nThere's also a mild circularity: Kwok's ongoing captioning study picked a 9° offset based on Arora's +8.7° boundary, and then the paper treats +8.7° as an independent constraint. The numbers are consistent, but they come from the same research program, and the unpublished parts are self-cited. That's not inappropriate—the work is legitimately cumulative—but it means external readers can't yet check the load-bearing pieces.\n\nWho is this for? HWD product designers, accessibility researchers building captioning glasses, and anyone writing a related design survey. It deserves a serious referee: it's a concrete, clearly argued position paper with real engineering relevance. I'd send it to peer review, and ask the authors to either add an active-image edge study or soften the abstract to say 'suggests' rather than 'recommends' until the transfer is tested.","headline":"A genuinely useful synthesis that gives HWD makers a concrete design target, but the left edge of the recommended window rests on a display-off combiner study that hasn't been tested for active imagery.","tokens_in":23231,"tokens_out":2744,"would_cite":true,"duration_ms":25583,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper consolidates human-factors studies to conclude that a monocular see-through display worn as everyday glasses should place its virtual image vertically centered and offset from +8.7° to +23.7° toward the ear, with a 15°…","keywords":["head-worn displays","optical see-through","monocular displays","display positioning","field of view","smart glasses","augmented reality","cognitive capture"],"falsifier":"Empirical test: have participants wear a right-eye waveguide display with a lit 15° virtual image whose inner edge sits at +6°, +8.7°, and +11° while doing a text-reading task, and record annoyance and performance; if the +6° condition is not rated more annoying or more visibly edged than the +8.7° condition, the recommended left boundary is not supported.","tokens_in":21925,"feed_emoji":"👓","tokens_out":5789,"duration_ms":54801,"temperature":0.7,"pith_summary":"This paper consolidates experimental results on monocular optical see-through head-worn displays to answer where in the wearer's visual field the virtual image should sit in glasses worn all day. It argues that centering the display at the principal point of gaze is fastest but interruptive and socially awkward, and that an offset toward the ear balances interruption, comfort, performance, and social perception. For text-based tasks, it recommends a 15° horizontal field of view, vertically centered, placed from +8.7° to +23.7° toward the ear, with glanceable content extended up to +30°. If these numbers are right, they give waveguide-optics designers concrete angular targets for everyday smart glasses.","feed_headline":"Put smart-glasses text 9 to 24 degrees toward the ear","feed_subtitle":"Human-factors review finds a 15-degree field of view that balances interruption, comfort, and social perception.","key_machinery":"The reasoning rests on a set of angular thresholds from different studies: an 8° radius around gaze where centered imagery causes cognitive capture; the roughly 25° limit of eye-only gaze shifts; the -20.2° to +8.7° window where visible combiner edges annoy wearers even when the display is off; and the -24.6° to +19.6° comfort range for sustained reading. The recommendation interval is the intersection of these constraints: to the ear side of the visible-edge boundary, inside the sustained-reading comfort bound, and within eye-only reach.","core_discovery":"The paper's central claim is that for a right-eye monocular display worn like ordinary glasses, the virtual image should be vertically centered but horizontally offset so that it occupies +8.7° to +23.7° toward the ear, giving a 15° horizontal field of view for text. Glanceable content may extend to +30° for short interactions, while sustained reading sessions of thirty minutes or more should stay inside +19.6°. The paper also concludes that when the display is off, optical combiner edges should lie outside the -20.2° to +8.7° window around straight-ahead gaze. These bounds come from combining constraints on cognitive capture, eye-movement comfort, combiner-edge visibility, and social perception. The paper recommends the narrower +8.7° to +19.6° window for sessions of half an hour or more.","pith_inferences":["The paper's results are for right-eye wearers; by symmetry a left-eye display would mirror the window, but the social-perception and combiner studies do not test the mirrored case, so a left-eye recommendation remains an extrapolation.","The +8.7° inner edge comes from an inactive-combiner study; until a lit active image edge is tested at several offsets, the exact left boundary of the recommendation is provisional.","The 30° glanceable cap assumes eye-only shifts; if users are willing to move their heads, the practical positional envelope is larger, which could matter for glanceable notifications.","Future gaze-contingent designs could place optics more centrally and shift content only during use, but the paper's recommendation is for fixed, static positioning in the frame."],"forward_implications":["Waveguide optics for everyday glasses can be designed to center a 15° horizontal field of view at about +16° azimuth on the right lens.","Real-time captioning for deaf and hard-of-hearing users is best served by a 15° field of view at roughly 9° offset toward the ear, with larger fields causing eye and head strain and smaller fields providing too little context.","Content should be kept inside +30° toward the ear even for short glances, and inside +19.6° for reading sessions of 30 minutes or more.","A right-eyed display can still overlay graphics on the world ahead because the wearer can simply turn the head approximately 15° to bring the earward image into alignment.","Displays placed below eye level can feel socially awkward in face-to-face conversation, so the recommended earward, vertically centered position avoids that risk."],"supporting_citations":[{"why":"Supplies the 8° cognitive-capture radius: displays inside it can tunnel attention, setting the inner-edge constraint for the recommended window.","marker":"[27]"},{"why":"Systematically tests nine monocular display positions in a driving simulator and shows center-right is the best performance-comfort trade-off.","marker":"[49]"},{"why":"Order-picking study showing diagonals and far offsets hurt accuracy and comfort, ruling out corner placements.","marker":"[50]"},{"why":"Defines the roughly 25° eye-only gaze range used as the outer comfort boundary for the recommendation.","marker":"[63]"},{"why":"Measures where unlit optical combiner edges become annoying, producing the -20.2° to +8.7° forbidden zone adopted for the display's inner edge.","marker":"[71]"},{"why":"Long-form reading study showing 0–20° offsets are comfortable and pixels beyond 30° toward the ear cause eyestrain.","marker":"[94]"},{"why":"Combined left and right reading studies yielding the -24.6° to +19.6° comfortable pixel range that sets the outer bound for sustained use.","marker":"[95]"},{"why":"Reports captioning field-of-view preferences in group conversation that support 15° as a good compromise for text.","marker":"[109]"}],"fun_headline_variants":["Smart-glasses text: offset 9–24° toward ear","Everyday smart glasses: text best near the ear","15° FOV, 9–24° offset: smart-glasses sweet spot","For text on smart glasses, aim toward the ear","Glanceable smart-glass content: up to 30° earward"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the +8.7° inner-edge boundary, measured for the appearance of an unlit optical combiner in emulated glasses, applies unchanged to the inner edge of a lit virtual image on a real waveguide display.","fun_headline_variants_meta":{"raw":{"variants":["Smart-glasses text: offset 9–24° toward ear","Everyday smart glasses: text best near the ear","15° FOV, 9–24° offset: smart-glasses sweet spot","For text on smart glasses, aim toward the ear","Glanceable smart-glass content: up to 30° earward"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000227,"raw_usage":{"total_tokens":1476,"prompt_tokens":955,"completion_tokens":521,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":571,"completion_tokens_details":{"reasoning_tokens":431}},"tokens_in":571,"tokens_out":521,"duration_ms":5723,"temperature":1.0,"reasoning_tokens":431,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:41:45.089782+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Empirical test: have participants wear a right-eye waveguide display with a lit 15° virtual image whose inner edge sits at +6°, +8.7°, and +11° while doing a text-reading task, and record annoyance and performance; if the +6° condition is not rated more annoying or more visibly edged than the +8.7° condition, the recommended left boundary is not supported.","supporting_citations":[{"cited_title":"Perrault, Denys J","cited_arxiv_id":null,"evidence_quote":"Systematically tests nine monocular display positions in a driving simulator and shows center-right is the best performance-comfort trade-off."},{"cited_title":"Towards Finding the Optimum Position in the Visual Field for a Head Worn Display Used for Task Guidance with Non-registered Graphics","cited_arxiv_id":null,"evidence_quote":"Order-picking study showing diagonals and far offsets hurt accuracy and comfort, ruling out corner placements."},{"cited_title":"Coordinated eye and head movements for gaze interaction in 3D environments","cited_arxiv_id":null,"evidence_quote":"Defines the roughly 25° eye-only gaze range used as the outer comfort boundary for the recommendation."},{"cited_title":"Comfortably going blank: Optimizing the position of optical combiners for monocular head-worn displays during inactivity","cited_arxiv_id":null,"evidence_quote":"Measures where unlit optical combiner edges become annoying, producing the -20.2° to +8.7° forbidden zone adopted for the display's inner edge."},{"cited_title":"Effects of lateral eye displacement on comfort while reading from a video display terminal","cited_arxiv_id":null,"evidence_quote":"Long-form reading study showing 0–20° offsets are comfortable and pixels beyond 30° toward the ear cause eyestrain."},{"cited_title":"Looking from a different angle: Placing head-worn displays near the nose","cited_arxiv_id":null,"evidence_quote":"Combined left and right reading studies yielding the -24.6° to +19.6° comfortable pixel range that sets the outer bound for sustained use."},{"cited_title":"Preferences for captioning on emulated head worn displays while in group conversation","cited_arxiv_id":null,"evidence_quote":"Reports captioning field-of-view preferences in group conversation that support 15° as a good compromise for text."}],"review_version":1}