A working philosophy
The Resett Method.
Six principles for matching work to your actual cognitive capacity. Each one names a claim, the evidence behind it, and the concrete practice we ship in the product.
Each principle below carries its evidence basis — peer-reviewed where graded A/B/C; industry case studies labeled as such. We don't ship copy that isn't backed by its stated source, and we don't make percentage productivity claims we can't substantiate. See Methodology for the live, API-backed metric definitions.
Principle 1
Cognition is finite. Calendar consumes it.
Working memory is a fixed resource. Every meeting and decision draws from the same pool.
Sweller's Cognitive Load Theory shows that working memory has a hard ceiling. When the day's intrinsic load exceeds it, performance drops on every subsequent task — not just the next one. Today the engine ships the simplest honest version of this: a meeting-load cap (default 60% of the workday) above which it refuses to place deep work. The full load-budget model — where two 30-minute decision meetings cost more than five status updates — is the direction we're building toward, not a shipped feature.
Practice
Shipped: a hard meeting-load cap the planner will not schedule past. Direction: a load-point budget that weighs decision density, not just minutes.Principle 2
Chronotype is a posterior, not a prior.
Your peak window is personal. The MEQ at signup is a starting estimate; your accepted patterns are the truth.
Roenneberg's MCTQ data on tens of thousands of chronotypes shows a 4+ hour spread between extreme owls and larks. Universal claims like "morning is for analytical work" are population means that don't apply to most individuals. We start with your MEQ score as a prior, then update the curve every time you accept, reject, or move a deep-work block. After ~30 days of activity, the learned curve dominates.
Practice
Take the MEQ short-form once. Let the engine refine your peak window from your actual behavior, not a textbook claim.Principle 3
Recovery is a budget, not a reward.
Detached hours after work are infrastructure. Skipping them compounds.
Sonnentag's Recovery Experience Questionnaire and Meijman & Mulder's Effort-Recovery Model both predict next-day vigor from psychological detachment, not from sleep duration alone. After a high-load day without recovery, the engine should under-fill the next day — not optimize it. Recovery debt accumulates silently until the user feels burnt out; by then, the cause is weeks behind.
Practice
Track detachment hours like sprint capacity. When recovery debt rises above threshold, the next day gets visibly emptier — restraint is the message.Principle 4
Meetings have cost vectors, not just durations.
A 30-minute decision review with 12 attendees costs more than a 30-minute 1:1. Same time, different cognition.
Rogelberg's meeting-effectiveness research (a decade-plus of work) shows that meeting cost scales with attendee count, decision density, and novelty — not just duration. Today Resett uses attendee counts and recurrence patterns in move eligibility and in the meeting-archaeology report. The full cost vector — type classification, async-replaceability, decision-debt per recurring series — is the roadmap this principle points at, and we'll label each piece as it ships.
Practice
Shipped: attendee-aware move eligibility + recurring-meeting archaeology. Direction: attendee-minutes-per-decision as a first-class sort.Principle 5
Calendar hygiene is the highest-leverage intervention.
Cancelling the wrong recurring meeting beats optimizing inside the wrong calendar.
Shopify's January 2023 calendar reset removed a large block of recurring meeting-hours (company-reported) in the first quarter by cancelling recurring meetings of 3+ people that hadn't produced decisions. The single biggest unlock for engineering teams is not better placement — it's removing meetings that should never have existed. Resett's meeting-archaeology report surfaces your longest-running recurring series and what they cost; acting on them is deliberately left to you (see Principle 6).
Practice
Run a recurring-meeting audit quarterly. Any series without a tracked decision in 30 days is a candidate for cancellation, not optimization.Evidence
Shopify (2023) company-wide calendar reset — widely reported industry case study (not peer-reviewed)Principle 6
Autonomy beats automation.
Every move is explained, predicted, and reversible. Surprises burn trust.
Self-Determination Theory (Deci & Ryan, 40+ years of replication) identifies autonomy as a basic psychological need; perceived loss of control predicts disengagement. Microsoft Productivity Score's 2020 backlash showed what happens when measurement crosses into surveillance. Resett surfaces decisions; users make them. Every recommendation cites a paper. Every move shows its predicted gain. Every action is reversible in one tap. Aggregate-only team rollups never expose individuals without explicit per-user opt-in.
Practice
Make the user the decision-maker on every recommendation. The product surfaces; never decides.
The full stack
Rigor
The enforcement machinery — CI jobs that fail the build on banned claims, and the list itself.
Methodology
The math — every metric's formula, threshold, and anchor, pulled live from the same API the product uses.
How it works
The product tour — what your first ten minutes look like.
Changelog
What we shipped — dated, specific, no vaporware.