RESEARCH QUESTION — THREE AGES OF THE SAME PAGE

Switch between them: the layout never changes, only how much of it is filled. A freshly promoted page is mostly provenance and that is its finished state, not a stub. The evidence spread is two structurally separate columns with a balance strip above them, so a one-sided literature is visible before you have to think about it.

{{ t.label }}
{{ winTitle }}
⌘K
⌂
?
▤
◈
⌖
◻
RESEARCH QUESTION {{ ageChip }} {{ revChip }} + attach source history resolve
{{ question }}
◆first wondered {{ originWhen }} · {{ originWhere }}
{{ originQuote }}
WORKING ANSWER {{ answerMeta }}
{{ answer }}
No position yet.
Write one when you have one — even a bad guess is a starting position, and the history will show where it went from there.
THE SHAPE OF IT {{ balanceMeta }}
{{ forLabel }}
{{ againstLabel }}
nothing against
! {{ lopsidedText }}
SUPPORTING {{ forCount }}
{{ s.cite }} {{ s.kind }}
{{ s.why }}
{{ s.anns }}
Nothing attached yet. Sources arrive here from the Reader, or with ⌘⇧A.
{{ forMore }}
OPPOSING {{ againstCount }}
{{ s.cite }} {{ s.kind }}
{{ s.why }}
{{ s.anns }}
{{ againstEmptyText }}
{{ againstMore }}
UNSORTED {{ unsortedText }} read then place · never forced
POSITION HISTORY
{{ h.when }} {{ h.text }}
{{ historyLink }}
OPEN THREADS
{{ t.glyph }} {{ t.text }}
None recorded. Threads usually appear once you have read enough to know what is missing.
RELATED
{{ r.text }} {{ r.rel }}
Nothing linked. Sub-questions spawned from this page will collect here on their own.
FEEDER CAPTURES
{{ feeders }}
{{ scout }}

ATTACHING A SOURCE — AND DECIDING WHICH SIDE IT IS ON

⌘⇧A from anywhere on the page, or drag a source in from the Reader. Side is a choice with three answers, not two: unsorted exists so that attaching never costs a judgement you have not made yet. The note and the annotation links are the part worth typing, and both are optional.

ATTACH TO — IS THE OVERNIGHT BENEFIT CONSOLIDATION OR ENCODING? esc
⌕ cordi rasch 2021 3 in vault · 1 stub
Cordi & Rasch (2021)
How robust are sleep-mediated memory benefits? · Current Opinion in Neurobiology · pdf in vault · 11 highlights
WHICH SIDE
{{ s.label }}
{{ s.hint }}
WHY IT IS HERE optional — but this is the field you will reread in six months
Meta-analysis: the overnight benefit shrinks to about a third of its reported size once preregistered studies are separated out.
LINK TO ANNOTATIONS
h·9 — “shrinks with preregistration” h·12 — “funnel plot asymmetry” + 9 more
attach ⏎ attach and open ⇧⏎ {{ attachHint }}
WHY TWO COLUMNS AND NOT ONE BIBLIOGRAPHY
A tagged list lets you collect thirty papers and never notice that twenty-nine of them agree with you. Two structurally separate columns make the ratio a visual fact you cannot scroll past, and the balance strip states it in words when one side is empty.
Filing a source on the wrong side is cheap to undo and is itself a recorded event — if a paper moves from supporting to opposing after a second reading, that move shows up in the position history alongside the answer revision it caused.
WHAT AN EMPTY SECTION SAYS
Empty sections are dashed outlines with a sentence explaining what would go there, not greyed-out headings. Nothing on the page is a placeholder waiting to be filled in — a page with one source and no answer is a page in a normal state, and the only thing that ever raises its voice is a one-sided literature.

EDITING THE ANSWER — AND THE OPTIONAL “WHY”

Left: editing is a plain text field and ⌘⏎ commits. The revision is recorded either way. Right: ⌥⏎ instead opens one line for what changed your mind, with the source, annotation or experiment responsible attachable in the same gesture. Nothing blocks the save; the note can also be added to any past revision later, from the history.

WORKING ANSWER — EDITING revision 6 · autosaves on blur
Probably both, but the dissociation designs that could separate them are underpowered. The strongest case for consolidation proper is TMR, where the cue is experimenter-controlled and orthogonal to encoding
⌘⏎ save ⌥⏎ save with a note on what changed esc reverts
No dialog appears on save. The previous text is kept, the timestamp is kept, and the page returns to reading. A revision with no note still counts as a revision — it just lands in the quiet trail.
⌥⏎ — WHAT CHANGED one line, then you are back on the page
Probably both, but the dissociation designs that could separate them are underpowered…
Cordi's funnel plot — the effect is mostly small-study bias
◈ Cordi & Rasch 2021 · h·12 + source + annotation + experiment
⏎ save bothesc saves the revision without the note
Escaping here is not a cancel — the position change is already recorded. You are only declining to explain it, which the history renders as a date rather than a story.

POSITION HISTORY — A TRAIN OF THOUGHT, NOT A DIFF LOG

Seven revisions over eleven months; three of them carry a why. Explained revisions are the narrative and get the full width. The unexplained ones are not hidden — they collapse into a dated rule that still shows the position moved, and any of them can be annotated later from here.

Is the overnight retention benefit attributable to consolidation, or to encoding strength?
7 revisions · 3 explained · 11 months
{{ h.label }}
 newest first
{{ h.date }}
{{ h.rev }}
{{ h.text }}
{{ h.why }}
{{ c.label }}
{{ h.from }}
{{ h.text }} {{ h.action }}
Read top to bottom this is a person changing their mind twice on evidence and once on a calculation they ran themselves. The four silent revisions between them are visible as movement without pretending to be reasoning.
Hovering any quiet revision offers + why — the note can be written months after the fact, which is usually when you know what actually changed.