Method & Assumptions
This page records the underlying assumptions and methods used throughout Coffee | One Roast at a Time.
It is not a set of rules, standards, or best practices.
It exists to make explicit how decisions are framed, how evidence is treated, and how conclusions are (or are not) drawn.
Nothing here is fixed permanently.
If an assumption stops holding, it will be revised or removed.
Why this page exists
Most of the work in this project happens in the Notes: individual roasts, observations, questions, and outcomes.
Over time, patterns emerge. Some of those patterns become stable enough to appear in the Guide.
This page sits between the two.
It captures:
- how signals are interpreted
- how evidence is weighed
- how uncertainty is handled
- what is deliberately not optimised for
It exists to reduce hindsight bias and prevent accidental certainty.
Cupping vs Drinking
Formal cupping and everyday brewing serve different purposes in this project.
They are intentionally kept separate.
Cupping
Cupping is treated as a diagnostic tool.
Its purpose is to:
- assess structural qualities (balance, cleanliness, development, defects)
- surface issues that may be masked in brewing
- allow side-by-side comparison under controlled conditions
Cupping is deliberately blunt.
It is not how the coffee is consumed, and it is not intended to be.
Brewing
Brewing reflects lived experience.
It answers different questions:
- Is the coffee enjoyable to drink?
- Does it hold up across methods?
- Does it require milk, dilution, or correction to be pleasant?
Positive brewing results are recorded when they occur, even if formal cupping has not yet taken place.
Relationship between the two
In this project:
- Cupping does not retroactively justify roast-day decisions
- Brewing outcomes do not override cupping results
- A coffee that drinks well but performs poorly in cupping remains a valid data point
Cupping informs future understanding.
It does not rewrite past intent.
Roast-day decision making
Roast-day decisions are made in real time, with incomplete information.
They are based on:
- observation
- system behaviour
- accumulated experience
- guardrails rather than targets
Later tasting or cupping outcomes are recorded separately and do not invalidate the conditions under which the roast was executed.
This separation is deliberate.
Lazy Mode
“Lazy Mode” refers to a deliberate reduction of intervention, not a lack of care.
It prioritises:
- repeatability over optimisation
- restraint over correction
- fewer variables rather than tighter control
Lazy Mode is used as a baseline to reduce noise and allow clearer signal detection over time.
It is not presented as a recommendation, only as a documented approach.
Evidence and confidence
Not all observations are treated equally.
Within the project:
- single outcomes remain open questions
- repeated outcomes become supporting evidence
- patterns only enter the Guide when they stop changing
Confidence is earned slowly and can be withdrawn.
Nothing is promoted prematurely.
What this page is not
This page does not:
- define success
- provide recipes
- guarantee outcomes
- claim general applicability
All observations are scoped to a single setup, a growing but limited set of coffees, and an evolving understanding.
How to read this page
This page is not required reading.
If you simply want to roast and drink coffee, you can ignore it entirely.
If you want to understand how conclusions are being formed — and where uncertainty is intentionally preserved — this page provides that context.
This page will change slowly, or not at all.
If it changes often, something upstream is unstable.