From cd458e2a9bf3f0798b62fc6ed6a726b65eb9b456 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ren=C3=A9=20Schultz=20Madsen?= Date: Fri, 28 Aug 2026 08:34:17 +0200 Subject: [PATCH] docs(overenskomst): consolidate the GLS-A overtime-basis findings MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Records the position that the overtime trigger under GLS-A / 3F Jordbrugsoverenskomsten is per day, on the strength of § 22 stk. 1's own wording ("efter normal daglig arbejdstids ophør"), corroborated by § 22 stk. 3's same-day notice deadline and by external practice (FV 2015.0097; Industriens Overenskomst § 13 stk. 2). States plainly what is NOT agreement text: § 22 stk. 1 never gives a figure for the daily boundary, so 26640 s is a derivation from § 8 stk. 1 (37 / 5), attested in the agreement only as the søgnehelligdag reduction unit. Records the counter-evidence rather than dropping it. For staldarbejde, § 8 stk. 2 sets a 37-hour average over up to 8 weeks, § 9 stk. 2 places hours on any day around the clock, and GLS-A's own praktikant circular and turnus guidance describe the norm and the overtime trigger in period terms. That tension is unresolved and is put to GLS-A as the third of three questions; it decides whether the engine models a daily boundary or a period budget. Also records what the engine implements correctly, six open items against it (per-day supplements modelled as hourly tiers, § 22 stk. 4 netting absent, planned shift data neither read nor written, dead OvertimeBasis enum, pay lines never persisted, the gross-vs-worked underpayment), four rejected readings with reasons, and the negative result on the payslip reconciliation: the figures were not generated by applying any rule to recorded work, so the remaining nine periods should not be fitted. Customer and worker identifiers are deliberately omitted throughout. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_01Cy9hsoDeyXWvQd6X8A3Pom --- ...2026-08-24-glsa-overtime-basis-findings.md | 347 ++++++++++++++++++ .../specs/overenskomst-research-findings.md | 8 + 2 files changed, 355 insertions(+) create mode 100644 docs/superpowers/specs/2026-08-24-glsa-overtime-basis-findings.md diff --git a/docs/superpowers/specs/2026-08-24-glsa-overtime-basis-findings.md b/docs/superpowers/specs/2026-08-24-glsa-overtime-basis-findings.md new file mode 100644 index 00000000..ed44c2fb --- /dev/null +++ b/docs/superpowers/specs/2026-08-24-glsa-overtime-basis-findings.md @@ -0,0 +1,347 @@ +# GLS-A / 3F Jordbrugsoverenskomsten — the overtime basis, and what the engine implements + +Findings consolidated 2026-08-24. Supersedes nothing; extends the daily-boundary section of +`overenskomst-research-findings.md` (decided 2026-08-07) with primary-source and external +evidence gathered 2026-08-23/24. + +**Position taken:** the overtime trigger under this agreement is **per day**, and the engine's +daily tier ladder is the correct shape. The daily boundary *value* is not stated in the +agreement; 7,4 t (26640 s) is a derivation from § 8 stk. 1 (37 ÷ 5), not agreement text. +The counter-evidence for a period-based reading of staldarbejde is recorded in section 4 below and +is not dismissed — it is the one point that would need GLS-A to close. + +Primary text throughout: +`/home/rene/Documents/workspace/microting/eform-angular-timeplanning-plugin/docs/superpowers/specs/sources/jordbrug-2026-2029.txt` +(Jordbrugsoverenskomsten 4010, GLS-A / 3F Den Grønne Gruppe, 2026-2029, 2. udgave 06.07.2026; +SHA-256 and source URL in `sources/SOURCES.md`). Line numbers refer to that extraction. + +--- + +## 1. The trigger is daily — the evidence + +**§ 22 stk. 1 (lines 731-734), verbatim:** + +> Stk. 1 Overarbejdsbetaling +> For overarbejde efter den normale arbejdstids ophør betales følgende: +> 1. og 2. time efter normal **daglig** arbejdstids ophør (+30 % af C-løn) +> For overarbejde herudover samt søn- og helligdage (+80 % af C-løn) + +The operative trigger contains the word *daglig*, and the rate table is structured per day — +"1. og 2. time" counted from that day's cessation point. This is the agreement's own wording, +not an interpretation. + +**§ 22 stk. 3 (lines 752-758)** corroborates it structurally: + +> Overarbejde er medarbejderne forpligtede til at udføre, når arbejdsgiveren anser det for +> nødvendigt af hensyn til driften. Om sådant overarbejde skal medarbejderne have besked +> **senest inden middag den pågældende dag**. +> I tilfælde hvor overarbejdet består i søn- og helligdagsarbejde, skal medarbejderne have +> besked senest dagen før. + +A same-day notice deadline presupposes a daily baseline to notify against. Reinforced at +line 449: *"**Beordret** overarbejde udløser overtidsbetaling efter overenskomstens regler herom."* +Overtime under this agreement is ordered against a daily norm, not accrued against a period budget. + +**External corroboration — the daily basis is standard Danish drafting.** + +- *Industriens Overenskomst § 13 stk. 2* defines overtime as work outside + "den i den enkelte uge fastlagte daglige arbejdstid for den enkelte medarbejder" — daily, + and explicitly plan-anchored. Different agreement; does not govern here; establishes the pattern. +- *Faglig voldgift FV 2015.0097* (2. november 2015, Kristelig Fagforening mod Kristelig + Arbejdsgiverforening, https://arbejdsretten.dk/media/16489/kendelse-kf-mod-ka-fv-2015-0097.pdf) + construes KA/KF § 10 stk. 1, which defines overarbejde as work "ud over den planlagte daglige + arbejdstid". Also daily. The same agreement handles short-notice *placement* changes through a + separate deeming clause — "betales der for de ændrede timer **som ved** overarbejde" — which sits + definitionally outside the overtime definition. Jordbrug has no such deeming clause. + +**Placement is a separate instrument, not overtime.** § 23 Forskudt arbejdstid pays a +forskydningstillæg for hours up to 2 h before 06:00 or after 18:00. GLS-A's own guidance states +the two are alternatives, not cumulative. Working a full planned duration displaced in time is +therefore *not* overtime under this agreement — it is at most forskudt tid. For staldarbejde it is +not even that; see section 4. + +--- + +## 2. The daily boundary value is NOT stated in the agreement + +§ 22 stk. 1 never states a figure for "normal daglig arbejdstid". Searching the 2026-2029 text: + +- `planlagt arbejdstid`, `arbejdsplan` — appear only inside the opt-in flexible regimes + (§ 9 stk. 4, § 9 stk. 5, and the kapitel 21 protokollat), never in the ordinary regime § 22 governs. +- `vagtplan`, `aftalt arbejdstid` — do not occur at all. +- `7,4` — occurs only as the **søgnehelligdag reduction unit** (§ 9 stk. 2; protokollat stk. 2), + never as a stated daily norm. + +So 26640 s is a derivation: § 8 stk. 1's *"Den normale effektive arbejdstid er indtil 37 timer +ugentlig"* (line 309), divided by 5. It is a reasonable and conventional derivation, and 7,4 is +demonstrably the agreement's own daily unit — but it is not quoted agreement text, and any claim +that it is should be corrected. + +**The ordinary regime imposes no planning duty at all.** § 9 stk. 1 (lines 377-378) states only a +permissible window: *"Den daglige arbejdstid lægges mandag til lørdag, mellem kl. 6.00 og kl. 18.00."* +A mandatory written plan and a 6-hour daily floor exist **only** in the opt-in protokollat +(kapitel 21, lines 3204-3327). The drafters knew how to require a plan and chose to require it only +in the alternative regime. Consequence: in the ordinary regime there may be no plan to anchor a +boundary to, which is a structural argument for keeping the derived 7,4 t as the default. + +--- + +## 3. What § 50 fixes for udenlandske praktikanter + +**§ 50 stk. 4 c (lines 2207-2212)** — the rates, and they match the engine exactly: + +> Overarbejde og arbejde på søn- og helligdage afregnes med et tillæg til praktikantens +> normaltimeløn på **50 % for de første 2 timer og herefter 80 %** eller tilsvarende frihed. +> Overarbejde afspadseres eller betales med overarbejdsbetaling efter praktikantens ønske. + +This is the `26640 / 33840 / null` ladder: normal time, then a 2-hour OVERTIME_50 band, then +OVERTIME_80 open-ended. Note the rates differ from ordinary staff (§ 22 stk. 1 gives 30 %/80 %), +and that overtime is a multiple of the *praktikant's own* rate, not the § 22 C-løn basis. + +**§ 50 stk. 4 d (lines 2213-2222)** — the staldarbejde supplements: + +> For arbejde **i normal arbejdstid** på lørdag eftermiddag betales et tillæg **pr. dag** på: +> pr. 1. marts 2026 ... kr. 73,90 +> For arbejde **i normal arbejdstid** på søn- og helligdage betales et tillæg **pr. dag** på: +> pr. 1. marts 2026 ... kr. 177,60 + +Two things follow, both load-bearing: + +1. These are **flat per-day amounts**, not hourly rates. See § 5 for the engine discrepancy. +2. They are payable only for work **i normal arbejdstid**. A day treated as entirely overtime + cannot simultaneously claim the supplement. This is a hard consistency constraint on any + reading that makes whole days overtime. + +**§ 50 stk. 7 (lines 2255-2257):** *"Overenskomstens øvrige bestemmelser er gældende for +praktikanter, hvor andet ikke følger af § 50."* The rest of the agreement applies by default; +§ 50 contains no carve-out from the arbejdstid rules or from the kapitel 21 protokollat. + +**Salary basis.** § 50 stk. 4 a's monthly figures divide to 160,33 t/md (13.315,41 ÷ 83,05), +and the agreement's own conversion constant is attested at lines 3041-3042: +*"...omregnes timelønnen til månedsløn med det gældende timetal, p.t. 160,33."* +160,33 = 37 × 52 ÷ 12. A praktikant is a full-timer at 37 t/uge by definition. + +--- + +## 4. The counter-evidence, recorded — staldarbejde and the period framing + +This is the one point the daily reading does not fully dispose of, and it is recorded here +deliberately rather than omitted. + +**§ 8 stk. 2 (lines 311-313):** +> Ved arbejde med pasning af dyr er den ugentlige arbejdstid indtil 37 timer **i gennemsnit over +> en periode på op til 8 uger**. + +**§ 9 stk. 2 (line 385):** +> Ved arbejde med pasning af dyr kan den normale arbejdstid lægges **på alle ugens dage, hele døgnet**. + +So for animal care there is no placement window (hence no § 23 forskudt tid), and the weekly norm +is an 8-week average rather than a fixed weekly cap. + +**§ 22 stk. 4 (lines 760-763)** anchors the *reckoning* weekly, not daily: +> Ved opgørelse af overarbejde fradrages forsømt tid af **den normale ugentlige arbejdstid**, +> medmindre forsømmelsen skyldes en medarbejderen utilregnelig grund eller en grund, som er +> rettidigt anmeldt til arbejdsgiveren og godkendt af denne. + +**GLS-A's own circular for this exact worker category** ("Lønoversigt — udenlandske praktikanter – +landbrug") states: +> **Staldarbejde** Den normale arbejdstid er indtil 37 timer pr. uge eller **296 timer i en 8 ugers +> periode** og kan lægges på alle ugens dage, hele døgnet. + +and GLS-A's guidance page on arbejdstidens lægning ved pasning af dyr describes overtime as hours +**above the turnus allotment** — 74 t for a 2-week rotation, 111 t for a 3-week — with normal hours +reduced by 7,4 t per søgnehelligdag. SIRI's trainee folder likewise ties the permit condition to the +**weekly** figure: *"din ugentlige arbejdstid skal være 37 timer."* + +**Assessment.** § 22 stk. 1's daily trigger is agreement text and governs. § 8 stk. 2 governs the +*norm* (how many hours are normal over the period), not the *trigger* (when a given day's hours +become overtime); averaging a weekly norm does not license unpaid overrun of a daily one. That is +the reading this document adopts. But GLS-A's own guidance describes the trigger itself in turnus +terms, and GLS-A wrote the agreement — so their guidance carries real interpretive weight even +though it is not agreement text. This tension is unresolved and is the subject of section 8 below. + +--- + +## 5. What the engine implements correctly + +| Rule | Source | Engine | Status | +|---|---|---|---| +| Overtime triggered per day | § 22 stk. 1 | `CalculatePayLinesForDay`, invoked per row | correct | +| 2 h at 50 %, then 80 % | § 50 stk. 4 c | tiers `26640 / 33840 / null` | correct | +| Overtime on the praktikant's own rate | § 50 stk. 4 c | pay codes carry hours; rate applied downstream | correct | +| Daily boundary 37 ÷ 5 | derived from § 8 stk. 1 | `PayRuleSetLock.NormalTimeBoundarySeconds = 26640` | defensible derivation, not agreement text | +| Saturday / Sunday / holiday day codes | § 50 stk. 4 d | `SAT_NORMAL`, `ANIMAL_SUN_HOLIDAY`, `SAT_ANIMAL_AFTERNOON` | correct in routing (see defect 6.1 on units) | +| Saturday afternoon split at 12:00 | § 50 stk. 4 d / § 15 | `PayDayTypeRule` time bands | correct | +| Grundlovsdag from 12:00, 24 Dec all day | § 29 stk. 1 | `CalculateGrundlovsdagPayLines` | correct | +| Placement is not overtime | § 23 vs § 22; § 9 stk. 2 for dyr | no forskudt-tid pay code; none needed for staldarbejde | correct for this preset | + +Two structural guards are worth recording as deliberate design, not accident: +`PayRuleSetLock.IsNormalTimeSplitPresetName` gates the § 50 split by preset *identity* (exactly two +normalized names), and `HasNormalTimeBoundaryShape` re-checks the *stored tier data* — because +presets are copy-at-create-time snapshots and a customer row may predate the +`20260807174115_CorrectPraktikantSection50Tiers` data migration. + +--- + +## 6. What the engine does not implement — open items + +**6.1 § 50 stk. 4 d supplements are per-day, modelled as hourly tiers.** +The agreement pays a flat kr. 73,90 / kr. 177,60 *pr. dag*. The engine encodes `SAT_NORMAL` and +`ANIMAL_SUN_HOLIDAY` as tier-1 pay codes accumulating **hours** up to 26640 s. If payroll multiplies +those hours by an hourly rate, the amount is wrong. Needs confirmation of how the receiving payroll system consumes the +export before it can be called a defect or a convention. + +**6.2 § 22 stk. 4 weekly netting is not modelled.** +The engine has no cross-day context at all. Culpable absence within a week should be netted off the +normal weekly hours before overtime is reckoned, with two exempt categories (utilregnelig grund; +rettidigt anmeldt og godkendt). The data model *can* carry this — `PlanRegistration` has `Sick`, +`OnVacation`, `OtherAllowedAbsence`, `AbsenceWithoutPermission` — but all four are **zero across all +14.108 rows**, so the classification does not exist in practice. Implementing netting without that +classification would wrongly erode overtime for sick and approved-absence workers. + +**6.3 Planned shift data is never read, and never written.** +`PlannedStartOfShift1..5` / `PlannedEndOfShift1..5` / `PlannedBreakOfShift1..5` are omitted from the +`Index()` projection (`TimePlanningWorkingHoursService.cs:128-192`), so they cannot reach +`CalculatePayLinesForDay`, whose DTO has no such fields. They are also non-zero in **0 of 14.108 +rows** — only `ServiceTimePlanningPlugin` (gRPC / flutter-time) ever writes them, and they are +minutes-since-midnight, not the 5-minute tick grid used by `Shift{N}Start/Stop`. Any future +plan-anchored work must fix the projection first and convert units deliberately. + +**6.4 `OvertimeBasis` is dead code.** +`enum OvertimeBasis { Weekly, Daily, DailyThenWeekly }` exists and nothing reads it. If the § 8 stk. 2 +question in § 4 is ever resolved in favour of a period basis, this is where it would live. + +**6.5 Pay lines are never persisted.** +`CalculatePayLinesForDay` has exactly two call sites, both Excel-export paths, and its results are +discarded. No production code writes `PlanRegistrationPayLines`; `PayrollExportService` reads a table +nothing populates. Any recalculation/audit feature starts by building that persistence. + +**6.6 Gross-vs-worked truncation.** +Pre-existing, previously reported: the `Index()` projection also drops pause stamps, so segment +truncation measures gross rather than worked time. Workers are **underpaid** roughly 11 minutes per +affected shift. Same root cause as 6.3 — a projection silently omitting columns the calculation needs. + +--- + +## 7. Readings considered and rejected + +**Overtime triggered by working outside the planned clock window.** Rejected. § 22 stk. 1 is a +duration test; placement is § 23's separate forskydningstillæg, and GLS-A's guidance treats the two +as alternatives. For staldarbejde § 9 stk. 2 removes the window entirely, so there is nothing to fall +outside of. + +**No plan ⇒ no overtime at all.** Rejected. § 22 stk. 1 is unconditional; it does not depend on a +plan existing, and the ordinary regime requires none. + +**Any unplanned day is entirely overtime.** Assessed as **doubtful** and not adopted. The syllogism +(no planned hours ⇒ cessation at hour 0 ⇒ everything is "efter" it) is formally valid but cuts against +§ 22 stk. 3's ordered-and-notified framing, § 9 stk. 2's "any day, any hour" placement for animal care, +§ 22 stk. 4's weekly netting, and § 50 stk. 4 d's requirement that supplement days be worked +*i normal arbejdstid*. The single point in its favour is the ordinary regime's total absence of a +planning duty (section 2 above) — a gap, not an affirmative rule. + +**Protokollat om alternativ arbejdstidsplanlægning as a route to plan-anchored boundaries.** +Not applicable to this customer. The protokollat (lines 3204-3327) replaces kapitel 3 wholesale, +confines placement to 06:00-18:00, caps days at 9,25 t with a 6 t floor, allows at most 5 working days +a week with two consecutive days off, and adds a fleksibilitetstillæg of kr. 4,10 per hour plus +weekend honorering. It is employer-elected with 14 days' notice and needs no union agreement, and +§ 50 stk. 7 does not exclude praktikanter from it. But it **never uses the word "overarbejde"** and +never states when overtime begins; and the recorded October 2025 data breaches it four ways at once +(7 working days in week 2025-W42, a 10,42 t day, no consecutive days off, second shifts 18:55-19:55 +outside the window). Electing it would also remove § 9 stk. 2's around-the-clock allowance, which is +what makes this worker's actual pattern lawful. + +--- + +## 8. The open question for GLS-A + +One question remains genuinely unresolved by the text, and it is the only one that would change the +engine's shape rather than its parameters: + +> 1. For en udenlandsk praktikant på månedsløn beskæftiget med staldarbejde (§ 50, jf. § 8 stk. 2): +> hvad udgør "normal daglig arbejdstid" i § 22 stk. 1's forstand, når overenskomsten ikke angiver +> et tal? +> 2. Hvis der foreligger en skriftlig arbejdsplan uden for protokollatet om alternativ +> arbejdstidsplanlægning — fastsætter den planlagte arbejdstids ophør da overarbejdsgrænsen for +> den enkelte dag, eller gælder 7,4 timer uanset? +> 3. Hvordan forholder § 22 stk. 1's daglige udløser sig til § 8 stk. 2's gennemsnit over op til +> 8 uger? Beregnes overarbejde dagligt, eller mod periodens normtimetal, jf. GLS-A's egen +> vejledning om turnus (74 timer / 2 uger, 111 timer / 3 uger)? + +Question 3 decides whether the engine models a daily **boundary** or a period **budget**. Until it is +answered, the daily reading in section 1 stands as this document's position, on the strength of § 22 stk. 1's +own wording. + +--- + +## 9. Statutory backdrop (all regimes) + +- **Arbejdstidsloven § 4** — average ≤ 48 t/uge including overtime, over a 4-month reference period. + Does not define overarbejde; folds it into the ceiling. +- **Arbejdsmiljøloven § 50 stk. 1** — at least 11 consecutive hours' rest per 24 h. Stk. 2 permits + reduction to 8 h for agriculture up to 30 days per calendar year. +- **Arbejdsmiljøloven § 51** — one rest day per 7-day period; agriculture exempt from the Sunday + requirement; for care of animals the rest day may be postponed against equivalent later time off. +- **Direktiv 2003/88/EF art. 2** — defines "working time" and "rest period" only; **silent on + overtime**, which is left to national law and collective agreements. +- **Ansættelsesbevisloven § 8** — attaches consequences to work outside referencetimer/-dage, but the + remedy is a right to refuse, not overtime pay; §§ 6-11 are displaced where a qualifying CBA applies. + +Statutory sourcing caveat: retsinformation.dk returned HTTP 403 to every fetch attempt; the Danish +statutory wording above came from danskelove.dk / elov.dk mirrors. The EU directive and the +collective agreements were fetched from primary sources directly. + +--- + +## 10. Consequence for the payslip-reconciliation case + +Recorded here because it is what prompted this research, and because the conclusion is negative. +Customer and worker identifiers are deliberately omitted; see the SDD ledger for the specific case. + +The payslip overtime figures examined in that case — one udenlandsk praktikant on staldarbejde, +ten monthly periods 2025-10 through 2026-07 — were **not produced by applying any rule to recorded +work**: + +- In every period whose totals reconcile, `OT1 = Arbejdstimer − Løn − OT2` exactly, to the øre — + OT1 is an arithmetic remainder, not a rule output. +- 9 of 10 OT2 values are exact integers (20, 20, 25, 23, 22, 20, 25, 25, 18); 12 of the 20 OT figures + land on exactly .00. Clock-derived hours do not behave that way. +- Per-period, per-week and per-turnus readings of the 50/80 split are each falsified across all + periods; the per-day reading is structurally impossible for 2026-03 (OT1 7,74 permits at most 3 days + to reach OT80, which would each need 6,67 t of it — a 16,07 t day). +- Recorded hours diverge from payslip totals by up to ±54 t per month in alternating directions, + while the annual totals are internally consistent. +- 2026-02 and 2026-04 totals exceed `Løn + OT1 + OT2` by exactly their own stated retro lines + (5,83 and 27,74, the latter being 2026-03's own OT1+OT2). Taking the totals at face value puts two + 4-month windows over Arbejdstidsloven § 4's 48 t/uge ceiling (48,38); removing the duplication + brings them to 46,43. The law independently corroborates the double-counting reading. +- 2025-12 and 2026-06 fail even the bookkeeping identity, by −46,35 and −48,59 t. Both carry + "Afholdt ferie −5,00 dage", and neither gap is 5 × 7,4. + +A lawful roster reproducing all ten periods **can** be constructed (4 planned days/week at 9,25 t, +overtime from planned-day overrun plus a small number of unplanned days), but it depends on the +unresolved § 8 question and on the doubtful unplanned-day reading, and June 2026 requires 92 % of +available days worked. It would be a fabricated roster reproducing hand-entered figures. + +**Conclusion: the remaining nine periods should not be fitted.** The engine is not reproducing these +numbers because these numbers were not generated by an engine. + +--- + +## 11. Sources + +Primary agreement texts, with SHA-256 and URLs, in +`/home/rene/Documents/workspace/microting/eform-angular-timeplanning-plugin/docs/superpowers/specs/sources/SOURCES.md`. + +- Jordbrugsoverenskomsten 4010, 2026-2029 (2. udgave 06.07.2026) — `sources/jordbrug-2026-2029.txt` +- Jordbrugsoverenskomsten 4010, 2024-2026 (2. udgave 17.05.24) — fetched from gls-a.dk, structurally + identical protokollat; praktikanter are § 48 in that edition, § 50 in 2026-2029 +- FV 2015.0097 — https://arbejdsretten.dk/media/16489/kendelse-kf-mod-ka-fv-2015-0097.pdf +- GLS-A Lønoversigt, udenlandske praktikanter – landbrug (staldarbejde / andet arbejde) +- GLS-A, Arbejdstidens lægning ved pasning af dyr — https://www.gls-a.dk/nyheder/arbejdstidens-laegning-ved-pasning-af-dyr/ +- SIRI, Information til praktikanter inden for det grønne område (aug. 2020), nyidanmark.dk +- Direktiv 2003/88/EF — https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX%3A32003L0088 + +GLS-A's members-only pages under `gls-a.dk/services/arbejdstid/jordbrugsoverenskomsten/` — including +its own "Alternativ arbejdstidsplanlægning" page — are paywalled and were not accessible. If member +credentials exist, that is the most likely place an explicit answer to § 8's question already lives. diff --git a/docs/superpowers/specs/overenskomst-research-findings.md b/docs/superpowers/specs/overenskomst-research-findings.md index 32249200..ce96a149 100644 --- a/docs/superpowers/specs/overenskomst-research-findings.md +++ b/docs/superpowers/specs/overenskomst-research-findings.md @@ -435,6 +435,14 @@ Both praktikant presets are named `… 2026-2029` in the frontend catalogue whil # "Normal daglig arbejdstid" — what sets the overtime boundary (decided 2026-08-07) +> **Extended 2026-08-24.** This section's decision still stands, but the evidence base has +> since widened considerably — external sources (FV 2015.0097, Industriens Overenskomst +> § 13 stk. 2, GLS-A's own praktikant circular and turnus guidance, SIRI), the full text of +> the kapitel 21 protokollat, and a negative result on a customer payslip-reconciliation case. +> See `2026-08-24-glsa-overtime-basis-findings.md` in this directory. In particular, the open +> question is now three questions, and question 3 (daily boundary vs 8-week period budget for +> staldarbejde) is the one that would change the engine's shape rather than its parameters. + **Decision: the boundary stays 7,4 t (26640 s) by default. A planned day SHORTER than 7,4 t does NOT lower it by default.** Two of the decision's original justifications were overclaims [corrected 2026-08-08, see ledger G-065]: § 22 stk. 1 states no figure,