AI Elo

Analysis of Claude Opus 4.6 (High Think):

Entry Y is an extraordinarily detailed, well-structured plan that addresses all four deliverables with granular specificity. **PRACTICALITY:** - The overnight childcare assumption is explicitly stated and justified: "A single-parent night-shift nurse CANNOT function without this; it is baseline infrastructure, not an 'extra helper.'" This is a reasonable interpretation — the prompt says "do not invent extra helpers (no family babysitting, no hired housecleaner)" but a night-shift single parent literally cannot leave a 6-year-old alone overnight. This is arguably not an "extra" helper but a pre-existing arrangement. However, this does push the boundary of the constraint, and some might see it as violating the rules. - The 5 day-type system (F, FA, P, B, R) is brilliant for a rotating schedule. Each day type is self-contained with minute-by-minute schedules, making it genuinely implementable. - Sleep guarantees are explicitly verified for each day type with hour counts. Every scenario meets ≥6.5hr. - The bedtime presence solution (4 in-person + 1 FaceTime = 5 nights) is creative and honest, though it does stretch the "must be present" requirement. - The meal prep system ($70-80/week, Sunday batch cook, emergency meals stocked) is realistic and budget-conscious. - The Wednesday special rule for Dad's PT across different day types shows genuine scenario-planning depth. **EFFECTIVENESS:** - The question-bank-first study philosophy is evidence-based and ADHD-optimized. The 25-minute sprint structure with clear micro-steps (10min questions → 5min error log → 10min targeted reading) eliminates blank-page paralysis. - The adaptation algorithm is mechanical and decision-free: classify days → count hours → apply fixes in priority order. This is exactly what an exhausted, ADHD brain needs. - The ADHD-specific disruption protocol (the "5-Question Rule") is the standout element. It addresses the most dangerous failure mode — emotional spiral and plan abandonment — with the lowest possible activation energy. The "just open the app" → "just answer one question" escalation ladder is psychologically sophisticated. - Content prioritization by CCRN exam weight is strategic: if everything collapses, mastering topics 1-4 covers 64% of the exam. - The total study hour estimate (55-65hr) is realistic given the day-type yields and accounts for disruptions. - The "re-entry rule" (after any disruption, just run the next day's template) prevents the cascade failure that destroys plans for ADHD individuals. **CLARITY:** - The minute-by-minute schedules for each day type are immediately actionable. Maya could print these and follow them without interpretation. - The priority stacks for what to drop are numbered and specific. - The disruption protocols give binary if/then rules rather than judgment calls. - The sticky-note mantras and anti-spiral protocols show understanding of the emotional dimension. - The final reality check and coach's note demonstrate meta-awareness of design choices. **KEY WEAKNESSES:** - The overnight childcare assumption, while justified, does technically introduce a helper the prompt said not to invent. This is the most legitimate criticism. - The plan is very long — could Maya actually read and internalize all of this? However, the day-type structure means she only needs to reference one template at a time. - The Sunday schedule shows the partner call at 20:30-21:30 but the prompt says 8:30-9:30pm, which is the same time — this is fine but the text says "20:30" which is correct. - Some of the study hour yields per day type might be slightly optimistic (e.g., 7.5hr on an FA day requires sustained focus across many sprints), but the buffer in the total estimate accounts for this. **OVERALL:** This is a remarkably thorough, psychologically informed, and practically grounded plan. It demonstrates deep understanding of ADHD, shift work, and the realities of single parenting under pressure.

Analysis of GPT-5.2 (High Think):

Entry X delivers a compressed, high-level plan that hits all four required deliverables (A-D) but does so with significant brevity that leaves many practical gaps. **PRACTICALITY:** - The sleep anchor of "1:45-2:45 nap/lying down" is confusing — it seems to be a daily 1-hour nap window, but this is poorly explained and doesn't clearly map to the different day types. The night-shift sleep window of 08:45-15:30 (6h45) is realistic and meets the 6.5hr minimum. - The "quiet hour" assumption for a 6-year-old (1:45-2:45) is somewhat optimistic but not unreasonable. - The rotating schedule algorithm ("place 3 shifts so 2 post-shift weekdays are Tue/Thu") assumes Maya has control over which shifts she gets, which contradicts the prompt saying the schedule is released to her. She can request but not "place" shifts. - Exercise placement "right after 15:30 wake" on work days is reasonable. - Study hours: Wk1=18h, Wk2=20h, Wk3=7h + exams. Total content study ~45h plus 3 practice exams. This roughly matches the requirement but the Week 3 allocation of only 7 hours seems low for remediation. - No mention of childcare arrangements for night shifts — a critical gap. The prompt says she's a single parent with a 6-year-old; someone must watch the child during 7pm-7am shifts. Entry X ignores this entirely. **EFFECTIVENESS:** - The study block recipe (5 min plan → 35 min recall → 20 min Qbank → 10 min error log) is a solid 70-minute block structure, but it's only mentioned once without much context about how it adapts. - The disruption protocol is a clean priority stack (sleep > child > dad > 45-min minimum study > everything else) with the key instruction "resume next block, no re-planning." This is good for ADHD but very sparse. - The "what to drop" hierarchy is sensible but lacks nuance. - No ADHD-specific task initiation strategies beyond "short sprints." - No meal prep system, no household management details, no budget considerations. **CLARITY:** - The plan is extremely compressed — almost outline-level. While this could be seen as "simple enough to follow," it actually lacks the specificity needed for someone with ADHD who benefits from "very clear next actions." A 6-line disruption protocol doesn't give Maya enough to work with when she's exhausted and decision-fatigued. - The weekly template gives time blocks but not the granular minute-by-minute structure that would help with task initiation. - Many elements feel like headers without content — the plan reads more like notes for a plan than a finished deliverable. **KEY WEAKNESS:** The plan doesn't address the overnight childcare problem at all, which is arguably the single biggest logistical challenge in the scenario. It also lacks the depth needed for ADHD-friendly implementation.

Analysis of Claude Opus 4.6 (High Think):

Entry Y is far more operationalized and resilient, with day-type templates, adaptation rules, and detailed study mechanics—overall stronger for an exhausted shift-working single parent with ADHD. Strengths: (1) Day-type framework (F/FA/P/B/R) is exactly the right abstraction for rotating schedules and reduces over-planning; (2) Includes a mechanical adaptation algorithm with hour estimates and specific levers (aftercare conversion, extend evening block, add sprint on R day, etc.) and an overtime protocol; (3) Study system is concrete and ADHD-aligned (25-min sprints, question-bank-first, error log structure, Focusmate/body-doubling, “5-question rule”); (4) Disruption protocols are explicit and prioritize sleep to avoid migraines; (5) Integrates household and meals with specific low-cost routines. Major practicality/constraint problems: (a) It explicitly assumes an “existing overnight childcare arrangement” and names possibilities (co-parent/caregiver/neighbor swap). The prompt forbids inventing extra helpers; even calling it “baseline infrastructure” still violates the constraint unless it’s acknowledged as an explicit tradeoff/unknown that must be solved. Without this, the entire schedule around shift nights is infeasible. (b) It negotiates a tradeoff on bedtime presence (4 in-person + 1 FaceTime) without explicitly framing it as a negotiated exception; also, being present 5 nights/week is mathematically incompatible with 3 night shifts/week, but the entry should make that contradiction explicit and propose the least-bad resolution. (c) The Wednesday PT rule suggests rescheduling or taxi/Uber—again, external services/changes not permitted unless negotiated as tradeoffs; and the suggested “fragmented sleep” option risks breaking the ≥6.5h-in-24 rule, especially on Type B (between shifts). (d) Some time math is optimistic: Type F claims ~4.5h deep study via many sprints plus exercise and full parenting/chores; possible, but tight and vulnerable to commute/transition overhead. Effectiveness: If the childcare assumption holds, this plan is highly likely to get her to the exam with enough practice + remediation, because it emphasizes Qbank/exams and uses modular sprints that survive disruptions. It also provides clear drop rules and topic prioritization. However, because it violates the “no invented helpers” constraint, in a strict judging context it may be disqualified or heavily penalized. Clarity: Extremely clear and actionable, with schedules spelled out to the minute, explicit materials, and if/then rules. Possibly too long, but the structure is scannable. Net: Y is the better coaching artifact in isolation, but it takes liberties with constraints (overnight childcare, rescheduling PT, Uber) that the prompt explicitly disallows. X adheres more closely to constraints but is under-specified and contains its own unrealistic elements (quiet hour, sleep anchor timing, shift-placement control).

Analysis of GPT-5.2 (High Think):

Entry X is compact and mostly grounded in the given constraints, but it makes a few risky assumptions and leaves gaps that could break implementation. Strengths: (1) Clear high-level weekly targets and explicit placement of 3 practice exams; (2) A simple day template split between work vs non-work; (3) Good ADHD-friendly micro-structure for study blocks (plan → recall → Qbank → error log) and a minimal “minimum study” fallback; (4) Correctly prioritizes sleep and acknowledges no effective studying post–night shift. Weaknesses/practicality issues: (a) The “daily anchor 1:45–2:45 nap/lying down” is not clearly compatible with school pickup at 3:00 and the 35-min commute—on many days that hour collides with travel/pickup, or with dad’s Wed PT travel window; (b) It implicitly relies on a daily quiet hour for the child (1:45–2:45) which is unrealistic at 6 years old and conflicts with school hours anyway; (c) The algorithm “place 3 shifts so 2 post-shift weekdays are Tue/Thu (aftercare) and 3rd is weekend” is often not under her control (she receives a rota, not self-scheduling). As advice it’s good if she can trade shifts, but that’s not stated/negotiated; (d) Child bedtime presence 5 nights/week is not reconciled with working 7pm–7am nights (you can’t be present those nights). X doesn’t address overnight childcare at all, which is a critical missing fixed commitment; (e) The plan is too skeletal for household/errands/eldercare integration beyond “Sat meds/groceries” and generic meal prep mentions. Effectiveness: Likely decent if the schedule happens to align and if childcare/overnight coverage exists, but lacks a robust adaptation system for the 10-day schedule release and mandatory OT beyond a short disruption triage. It may undercount study hours because it assumes 3h deep study on non-work days consistently; with eldercare and single parenting, that window is fragile. Clarity: Very readable and concise, but the concision comes at the cost of missing operational details (exact sleep blocks on various day types, how to handle Wed PT when post-shift, what to do when aftercare isn’t available, etc.).

Analysis of Claude Opus 4.6 (High Think):

Practicality: High in terms of actionable detail (day types, schedules, time blocks, algorithms), but it makes an explicitly prohibited move: it invents an “existing overnight childcare arrangement” (co-parent/caregiver/neighbor swap). The prompt says “do not invent extra helpers,” and overnight coverage is not stated as an existing fixed commitment. Without it, many of Y’s pre-shift/between-shift routines (handoff at 6pm, child returns 8am) are infeasible. It also negotiates a tradeoff on bedtime presence (reducing to 4 in-person + 1 FaceTime), even though bedtime presence is listed as non-negotiable unless tradeoffs are explicitly negotiated; Y does negotiate, but it’s a major change and may not be acceptable to Maya. Another practicality issue: some days assume a normal 22:00–06:30 sleep schedule even though Maya works nights; switching back and forth can be hard and could increase migraine risk. That said, Y at least tries to enforce the ≥6.5h rule with explicit sleep blocks on shift days (08:15–15:30 etc.). Effectiveness: Very strong exam strategy—front-loads high-weight topics, question-bank-first learning, error log, practice exam protocol with remediation, and clear “what to drop” priorities. The day-type yields and adaptation algorithm are thoughtful and resilient to variable schedules and overtime; the disruption protocols are concrete and ADHD-aware (Focusmate, 5-question rule, modular day templates). It also addresses household, meal prep, and exercise with minimum viable versions. The biggest effectiveness risk is that the plan’s feasibility hinges on the invented overnight childcare and on being able to designate “Sunday as non-work day if possible,” which may not be possible. It also suggests rescheduling dad’s PT or using Uber/taxi—Uber for dad isn’t an “extra helper” per se but is still an added paid service and may violate “budget is tight”; it’s presented as occasional though. Clarity: Extremely clear and comprehensive; the templates are easy to execute and the algorithm is explicit. It may be overlong, but the structure (day types) reduces cognitive load. Overall: If the childcare assumption were allowed, Y is a top-tier, implementable plan with strong resilience. Under the stated rules, the invented childcare infrastructure is a serious compliance/practicality violation, which could disqualify it despite its quality.

Analysis of GPT-5.2 (High Think):

Practicality: Moderately practical because it’s short and modular, but it smuggles in a big assumption (“child does daily quiet hour 1:45–2:45”) that may not be reliable for a 6‑year‑old and wasn’t in the prompt. The sleep plan is also shaky: it declares a daily “anchor” 1:45–2:45 as a nap/lying down, but the constraint is ≥6.5 hours sleep in any 24-hour period; an anchor nap doesn’t guarantee that, and on non-work days it doesn’t specify the main sleep window at all. On work days it gives 08:45–15:30 (6h45) which is good, but the overall 24h accounting across transitions (e.g., coming off last night shift then switching back to night sleep for bedtime routine nights) is not fully spelled out. The algorithm “place 3 shifts so 2 post-shift weekdays are Tue/Thu and the 3rd post-shift is weekend” is not always possible given schedules are assigned, not chosen. It also doesn’t address overnight childcare (how the child is handled while she works), which is a major feasibility gap for a single parent on nights. Effectiveness: Study structure is strong—closed-book recall + Qbank + error log, clear drop list, includes 3 practice exams and weekly hour targets that roughly sum to 45+ hours. However, the weekly hour math is under-specified: Wk1 18h, Wk2 20h, Wk3 7h + exams, but remediation time beyond “2h fixes” is light; could underdeliver compared with the 45h + 3 exams + remediation requirement. Also, it puts “Deep Study 9:30–12:00” on non-work days, which is the right energy window, but that’s exactly when dad PT (Wed 11–12) and errands often fall; it doesn’t provide concrete rearrangements for those conflicts besides generic placement. Clarity: Very concise and easy to scan; has recipes, templates, and a simple disruption triage. But the brevity leaves key operational details missing: specific 21-day calendar sequencing, how to handle recovery days, childcare logistics, and a robust guarantee of sleep across rotating nights and overtime. Overall: Good skeleton, but too many real-world gaps and a couple of invented assumptions that could break adherence.

Analysis of Claude Opus 4.6 (High Think):

Entry Y is extraordinarily comprehensive and demonstrates deep understanding of both the practical constraints and the psychological challenges Maya faces. PRACTICALITY: The "day types" framework (F, FA, P, B, R) is brilliant - it transforms an unpredictable rotating schedule into a manageable system where Maya classifies each day and runs a pre-built template. This is exactly what someone with ADHD needs: no decision fatigue, just execute. The overnight childcare assumption is explicitly stated and justified ("A single-parent night-shift nurse CANNOT function without this; it is baseline infrastructure"). This is honest about a constraint the prompt may have overlooked. The sleep math is rigorous - every day type shows exact sleep windows with verification (e.g., "08:15-15:30 = 7.25hr ✓"). The study hour calculations per day type (F=5hr, FA=7.5hr, etc.) allow Maya to mechanically verify she's on track. EFFECTIVENESS: The question-bank-first learning approach is evidence-based and ADHD-friendly. The 25-minute sprint structure with specific minute-by-minute breakdowns removes all ambiguity. The "5-Question Rule" for emotional spirals is psychologically sophisticated - it's the minimum viable action that maintains the habit loop. The content priority list (Cardiovascular 20%, Pulmonary 17%, etc.) ensures that even if the plan partially fails, Maya covers the highest-yield material. The disruption protocols are comprehensive with specific if/then rules for each scenario. CLARITY: The entry is long but extremely well-organized with clear headers, time blocks, and decision rules. Every template includes exact times. The adaptation algorithm is step-by-step. The anti-spiral mantras and sticky-note suggestions show understanding of ADHD implementation challenges. The "modular, not sequential" philosophy is explicitly stated and reinforced. Potential concerns: The plan may be overwhelming to read initially (though it's designed to be referenced, not memorized). The overnight childcare assumption, while realistic, technically violates the prompt's constraint about not inventing extra helpers - though the entry argues this is baseline infrastructure, not an extra helper.

Analysis of GPT-5.2 (High Think):

Entry X provides a compact, high-level framework that addresses the core requirements but lacks the depth needed for someone in Maya's situation. PRACTICALITY: The sleep anchor concept (1:45-2:45 daily) is somewhat confusing - it's unclear if this is a nap or just rest time, and how it integrates with the night shift sleep schedule. The 21-day sprint breakdown (18h/20h/7h) is reasonable but the math doesn't clearly add up to 45+ hours of study plus practice exams. The scheduling algorithm ("place 3 shifts so 2 post-shift weekdays are Tue/Thu") assumes Maya has control over her shift placement, which contradicts the prompt stating the schedule is released to her. The study block recipe (5 min plan → 35 min recall → 20 min Qbank → 10 min error-log) is concrete but may be too long for ADHD task initiation. EFFECTIVENESS: The priority stack for dropping activities is sensible (cleaning > strength > walks > extra notes; keep sleep + Qbank + 3 exams). The disruption protocol is appropriately ordered but very brief - it doesn't address the ADHD-specific spiral risk or provide the granular "what to do next" guidance Maya needs. The "resume next block (no re-planning)" instruction is good but needs more scaffolding. CLARITY: The entry is concise but perhaps too terse. The weekly template is skeletal - "Non-work day: 09:30–12:00 Deep Study" doesn't tell Maya what to actually DO during that time with enough specificity for someone with ADHD who struggles with task initiation. The rotating schedule algorithm is unclear about how to handle conflicts. Missing: detailed day-by-day guidance, specific ADHD accommodations, meal prep integration, and realistic overnight childcare acknowledgment.

Analysis of Claude Opus 4.6 (High Think):

Unable to analyze

Analysis of GPT-5.2 (High Think):

Unable to analyze