{"sustainability_index":{"claim":"Calendar audit score","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Higher = healthier *calendar*. Composite of four operational indicators with the weights above. Not a clinical wellbeing score.","paper_url":null,"kind":"operational_definition","formula":"score = 100 - (1 - intensity_low_pct) * 30 - (1 - capture_rate) * 30 - after_hours_pct * 25 - weekend_pct * 15; clamp to [0, 100]. Weights sum to 100. (Corrected 2026-07-02: the previous formula counted capture twice — capture_rate rewarded +20 while cannibalization_rate, which is 1 - capture_rate over the same totals, penalized -20 — one signal wearing ~40 of the weight points.)","disclaimers":["Not a clinical wellbeing assessment.","Not a measure of individual performance — cohort-only.","Composite weights are internal calibration, not from a single paper."],"scientific_anchor":{"claim":"Burnout is a clinical syndrome with three dimensions; diagnosis requires the Maslach Burnout Inventory or equivalent instrument.","citation":"Maslach & Leiter, World Psychiatry 2016","grade":"A","effect_note":"Calendar/messaging signals are stressor proxies, NOT a substitute for the MBI. Use them to flag patterns worth investigating, never to label a person.","paper_url":"https://doi.org/10.1002/wps.20311","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"sustained_intensity_distribution":{"claim":"Sustained-intensity tier distribution","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"An operational tier, not a clinical burnout diagnosis.","paper_url":null,"kind":"operational_definition","formula":"For each user, count triggers tripped over the window: (meeting-load avg > 55%), (after-hours avg > 45 min/day), (>=2 weekend-active days), (focus-capture < 50%). Tier: 2+=high, 1=med, 0=low. Distribution = count of users in each tier.","disclaimers":["Calendar data alone cannot diagnose burnout (which requires the Maslach Burnout Inventory or similar).","A 'high' tier is a cohort-level signal to investigate, not a per-person label.","Trigger thresholds (55% meeting load, 45 min after-hours, 50% capture) are internal calibration."],"scientific_anchor":{"claim":"Burnout is a clinical syndrome with three dimensions; diagnosis requires the Maslach Burnout Inventory or equivalent instrument.","citation":"Maslach & Leiter, World Psychiatry 2016","grade":"A","effect_note":"Calendar/messaging signals are stressor proxies, NOT a substitute for the MBI. Use them to flag patterns worth investigating, never to label a person.","paper_url":"https://doi.org/10.1002/wps.20311","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"deep_work_shortfall":{"claim":"Deep-work shortfall (hours)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"When p90 > 10h, the meeting-load problem is systemic.","paper_url":null,"kind":"operational_definition","formula":"For each user, sum over weekdays of max(0, target_minutes_per_day - deep_completed_minutes), divided by 60. Default target = 120 min/day. Distribution reported as p50/p75/p90 across the cohort.","disclaimers":["Not a sleep-debt or recovery metric. Sleep-tracker data is not ingested."],"scientific_anchor":{"claim":"Sustained high-effort focused work appears bounded to a few hours per day — a soft, inferential upper bound, not a hard peer-reviewed limit.","citation":"Macnamara, Hambrick & Oswald, Psychol Sci 2014 (meta-analysis of the deliberate-practice → performance effect); the ~4h/day figure itself originates in Ericsson, Krampe & Tesch-Römer, Psychol Rev 1993","grade":"C","effect_note":"IMPORTANT: Macnamara et al. 2014 is a variance meta-analysis — it finds the deliberate-practice → performance effect is real but smaller than Ericsson claimed (~4% education, ~18% sports, ~21% music). It does NOT establish a daily sustained-focus ceiling; the ~3–4h/day figure is Ericsson 1993's, and deliberate practice is not identical to deep knowledge work. We therefore use the ~4h/day figure only as an inferential soft bound and treat our 120 min/day deep-work floor as an internal operational default that sits well below it.","paper_url":"https://doi.org/10.1177/0956797614535810","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"cited"},"capture_rate":{"claim":"Focus-capture rate","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Below 60% means meetings are cannibalizing planned focus.","paper_url":null,"kind":"operational_definition","formula":"sum(deep_completed_minutes) / sum(deep_planned_minutes), averaged over users with any planned minutes. deep_completed is the surviving block's minutes MINUS a Slack-activity discount: 10 min per message the user sent inside a focus block, capped at 50% of the block's minutes (both internal calibration; applies only when Slack is connected).","disclaimers":["'Completed' is block survival discounted by demonstrated in-Slack activity — not a claim of verified attention."],"scientific_anchor":{"claim":"Task-switching leaves attention residue that degrades the next task; interruption also makes work faster but more stressful.","citation":"Leroy & Glomb, Organization Science 2018 (attention residue); Mark, Gudith & Klocke, CHI 2008 (speed/stress)","grade":"A","effect_note":"Leroy & Glomb 2018 demonstrate attention residue across four experiments. Mark 2008 shows interrupted work is faster but more stressful. The popular '~23 min resumption' figure is HCI folklore, not cleanly sourced to a peer-reviewed result; Resett uses ~23 min only as an internal operational planning estimate, never as a cited result.","paper_url":"https://doi.org/10.1287/orsc.2017.1184","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"meeting_trend_delta":{"claim":"Meeting-load trend (% change)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Negative = lighter calendar week-over-week.","paper_url":null,"kind":"operational_definition","formula":"((current_window_avg_meeting_minutes - prior_window_avg) / prior_window_avg) * 100, averaged over users.","disclaimers":[],"scientific_anchor":null,"threshold_source":null},"after_hours_pct":{"claim":"After-hours intensity (% of cohort)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"A cohort-level signal of work bleeding past stated hours.","paper_url":null,"kind":"operational_definition","formula":"Share of users whose mean after_hours_minutes/day > 30 over the window.","disclaimers":["Counts meeting + (opt-in) Gmail/Slack metadata only. Never content.","30 min/day threshold is internal calibration; the long-hours category itself is anchored in WHO/ILO 2021."],"scientific_anchor":{"claim":"Long working hours are associated with elevated cardiovascular and stroke mortality.","citation":"Pega et al., Environ Int 2021 (WHO/ILO meta-analysis, 60+ studies)","grade":"A","effect_note":"Working ≥55h/week → 35% increased CVD mortality and 17% increased ischemic heart disease vs 35–40h/week.","paper_url":"https://doi.org/10.1016/j.envint.2021.106595","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"weekend_pct":{"claim":"Weekend work prevalence (% of cohort)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Above 20% sustained = norm has shifted; investigate.","paper_url":null,"kind":"operational_definition","formula":"Share of users with any weekend-flagged day where meeting_minutes > 0.","disclaimers":[],"scientific_anchor":{"claim":"Psychological detachment from work during off-time reduces emotional exhaustion and improves next-day affect.","citation":"Sonnentag & Fritz, J Occup Health Psychol 2007 (Recovery Experience Questionnaire)","grade":"B","effect_note":"REQ validates detachment as one of four recovery experiences; low detachment scores predict emotional exhaustion in subsequent diary studies (Sonnentag, Binnewies, Mojza 2010).","paper_url":"https://doi.org/10.1037/1076-8998.12.3.204","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"deep_work_gini":{"claim":"Deep-work equity (Gini)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"0 = perfectly equal across the team; 1 = one person absorbs all deep work.","paper_url":null,"kind":"operational_definition","formula":"Gini coefficient over total deep_completed_minutes per user across the window.","disclaimers":[],"scientific_anchor":null,"threshold_source":null},"meeting_culture_score":{"claim":"Meeting culture score","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Inverse of back-to-back rate. >= 60 is healthy.","paper_url":null,"kind":"operational_definition","formula":"max(0, min(100, round(100 - (back_to_back_count / total_meetings) * 120))).","disclaimers":[],"scientific_anchor":{"claim":"Micro-breaks restore vigilance and reduce fatigue accumulation.","citation":"Albulescu, Macsinga, Rusu, Sulea, Bodnaru, Tulbure, PLOS ONE 2022 (meta-analysis)","grade":"A","effect_note":"Meta-analysis of 22 micro-break studies: even very short breaks (≤10 min) restore vigilance and reduce fatigue. Back-to-back meeting schedules prevent micro-breaks; harm is inferred via this established mechanism rather than measured at the meeting level.","paper_url":"https://doi.org/10.1371/journal.pone.0272460","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"cannibalization_rate":{"claim":"Deep-block cannibalization rate","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Share of planned deep-work that didn't happen. > 15% = systemic.","paper_url":null,"kind":"operational_definition","formula":"1 - (sum(deep_completed_minutes) / sum(deep_planned_minutes)), over the window.","disclaimers":[],"scientific_anchor":{"claim":"Task-switching leaves attention residue that degrades the next task; interruption also makes work faster but more stressful.","citation":"Leroy & Glomb, Organization Science 2018 (attention residue); Mark, Gudith & Klocke, CHI 2008 (speed/stress)","grade":"A","effect_note":"Leroy & Glomb 2018 demonstrate attention residue across four experiments. Mark 2008 shows interrupted work is faster but more stressful. The popular '~23 min resumption' figure is HCI folklore, not cleanly sourced to a peer-reviewed result; Resett uses ~23 min only as an internal operational planning estimate, never as a cited result.","paper_url":"https://doi.org/10.1287/orsc.2017.1184","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"chronotype_spread":{"claim":"Chronotype spread","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Inferred from calendar; can be overridden by the user's MEQ result.","paper_url":null,"kind":"operational_definition","formula":"Count users by inferred chronotype: avg first-meeting time < 09:30 = lark, > 11:00 = owl, else intermediate.","disclaimers":["Calendar-time cutoffs are a proxy. The MEQ-derived User.chronotype takes precedence when available."],"scientific_anchor":{"claim":"Chronotype shifts sleep timing by several hours across the population, and the time-of-day cognitive peak shifts with it.","citation":"Roenneberg, Wirz-Justice & Merrow, J Biol Rhythms 2003 (MCTQ, sleep timing); Schmidt et al. 2007 + Adan et al. 2012 review (time-of-day cognition); MEQ: Horne & Östberg 1976","grade":"A","effect_note":"Roenneberg 2003 documents large interindividual variation in sleep timing; the cognitive-peak shift is anchored separately on Schmidt 2007 (synchrony effect) and the Adan 2012 review. Population-mean time-of-day advice is wrong for any specific person; personalize via MEQ/MCTQ.","paper_url":"https://doi.org/10.1177/0748730402239679","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"slack_after_hours_msg_mean":{"claim":"Slack after-hours message volume (cohort mean per active user)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Messaging-layer proxy for after-hours work intensity.","paper_url":null,"kind":"operational_definition","formula":"For users with any SlackDailyMetric rows in window, mean of messages_sent_after_hours per user-day. Users without Slack connected are excluded from the denominator.","disclaimers":["Counted from DM metadata only — no message contents persisted.","Users without Slack connected do not pull this number down to 0."],"scientific_anchor":{"claim":"Long working hours are associated with elevated cardiovascular and stroke mortality.","citation":"Pega et al., Environ Int 2021 (WHO/ILO meta-analysis, 60+ studies)","grade":"A","effect_note":"Working ≥55h/week → 35% increased CVD mortality and 17% increased ischemic heart disease vs 35–40h/week.","paper_url":"https://doi.org/10.1016/j.envint.2021.106595","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"slack_weekend_msg_mean":{"claim":"Slack weekend message volume (cohort mean per active user)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Detachment proxy at the messaging layer.","paper_url":null,"kind":"operational_definition","formula":"For users with any SlackDailyMetric rows in window, mean of messages_sent_weekend per user-day.","disclaimers":["Counted from DM metadata only — no message contents persisted."],"scientific_anchor":{"claim":"Psychological detachment from work during off-time reduces emotional exhaustion and improves next-day affect.","citation":"Sonnentag & Fritz, J Occup Health Psychol 2007 (Recovery Experience Questionnaire)","grade":"B","effect_note":"REQ validates detachment as one of four recovery experiences; low detachment scores predict emotional exhaustion in subsequent diary studies (Sonnentag, Binnewies, Mojza 2010).","paper_url":"https://doi.org/10.1037/1076-8998.12.3.204","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"slack_focus_block_interruption_mean":{"claim":"Mentions during focus blocks (cohort mean per active user)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Interruption-layer signal: how often messaging breaks scheduled deep work.","paper_url":null,"kind":"operational_definition","formula":"For users with any SlackDailyMetric rows in window, mean of mentions_received_during_focus_blocks per user-day.","disclaimers":[],"scientific_anchor":{"claim":"Task-switching leaves attention residue that degrades the next task; interruption also makes work faster but more stressful.","citation":"Leroy & Glomb, Organization Science 2018 (attention residue); Mark, Gudith & Klocke, CHI 2008 (speed/stress)","grade":"A","effect_note":"Leroy & Glomb 2018 demonstrate attention residue across four experiments. Mark 2008 shows interrupted work is faster but more stressful. The popular '~23 min resumption' figure is HCI folklore, not cleanly sourced to a peer-reviewed result; Resett uses ~23 min only as an internal operational planning estimate, never as a cited result.","paper_url":"https://doi.org/10.1287/orsc.2017.1184","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"slack_response_latency_p50_seconds":{"claim":"DM response latency p50 (cohort median of per-user p50)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"Lower latency ⇒ stronger felt urgency to respond. Behavioral proxy for telepressure, not the validated B&S 2015 scale.","paper_url":null,"kind":"operational_definition","formula":"Per-user p50 over inbound→outbound reply pairs in DMs (only computed when ≥5 pairs in a day); cohort median over those per-user p50 values across the window.","disclaimers":["Days without ≥5 reply pairs contribute nothing to this metric.","Behavioral proxy — does NOT substitute for the validated telepressure scale."],"scientific_anchor":{"claim":"Workplace telepressure — felt urgency to respond to ICT messages — predicts reduced recovery and elevated burnout symptoms.","citation":"Barber & Santuzzi, J Occup Health Psychol 2015","grade":"B","effect_note":"Validated 6-item telepressure scale; higher scores correlate with lower psychological detachment and higher exhaustion. Rapid DM response latency is a behavioral proxy, not the validated instrument.","paper_url":"https://doi.org/10.1037/a0038278","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"},"sharply_worsening_trajectory_count":{"claim":"Direct reports flagged sharply_worsening (cohort count)","citation":"Operational definition (Resett, internal)","grade":"op","effect_note":"A cohort-level count, not per-user labels. Manager view of named direct reports lives at /org/burnout-flags and is gated on user.burnout_proxy_visibility_opt_in.","paper_url":null,"kind":"operational_definition","formula":"Count of users in cohort whose 14d burnout-proxy trajectory (see app/services/burnout_proxy.py) evaluates to 'sharply_worsening'. Cohort gate (N≥min_cohort_size) applies.","disclaimers":["NOT a burnout count — sharply_worsening is a stressor-load trajectory, not a burnout diagnosis. See Maslach & Leiter World Psychiatry 2016 for the validated diagnostic instrument."],"scientific_anchor":{"claim":"Burnout is a clinical syndrome with three dimensions; diagnosis requires the Maslach Burnout Inventory or equivalent instrument.","citation":"Maslach & Leiter, World Psychiatry 2016","grade":"A","effect_note":"Calendar/messaging signals are stressor proxies, NOT a substitute for the MBI. Use them to flag patterns worth investigating, never to label a person.","paper_url":"https://doi.org/10.1002/wps.20311","kind":"scientific_claim","formula":null,"disclaimers":[],"scientific_anchor":null,"threshold_source":null},"threshold_source":"internal_calibration"}}