Circle Thrice · internal · not the member portal

Alpha Review Dashboard — three layout options

One screen per day of the 2027 alpha, 58 fields, no scrolling at 1440×900. All three panels show 2027-01-01. The comparison page below is as tall as it needs to be; the panels inside it are not.

1 · Auspiciousness scale

Two designs for the range column T carries. Both are drawn on the real 2027 distribution and at the real 3.6px cell width, because contrast at cell width is the whole question.

Ten points · 1 to 10 · sequential

The bet: more steps give a reweighting somewhere to land, so a day can move without jumping a whole category.

7 of 10
2027-01-01

The year at 3.6px per day

The 365 days by step

Two costs. T is a sum of nineteen ±1 flags, so it lands on nine values and step 5 stays empty for the whole year — filling ten steps needs fractional weights or an input that does not exist yet. And a sequential ramp forfeits divergence: at cell width a bad day and a good day differ only in how dark they are, which is a legend lookup rather than a glance, and the negative lean reads as a light band with no direction in it.

Five points · 1 to 5 · divergent

The bet: fewer steps, each one legible at cell width and anchored to a true centre, so the strip can be read without a legend.

4 of 5
2027-01-01

The year at 3.6px per day

The 365 days by step

One cost, and it is the important one. Five steps hold contrast at cell width with nothing merged, and 3 is a real centre rather than the midpoint of nothing — but 153 of the 365 days land in step 2, so the lean the strip exists to show flattens into a single band. Ivy would see that the year is negative and lose the shape of how negative.

On the scale

Keep nine, drawn as eight

Neither, on this data. Ten steps cannot be filled by nineteen ±1 flags and cost divergence to get the resolution; five steps are legible but put 42 per cent of the year in one bucket. The nine-value range already in the sheet, drawn as eight with +3 and +4 merged, is the only version that both survives the cell width and keeps the skew visible — so the reweighting question is about the nineteen input weights, not about the width of the output. What would change this: if Ivy wants fractional weights, ten becomes reachable and worth redrawing.

2 · Layout comparison

Option	Centre of the screen	Demoted to pay for it	Colour Auspice Year	Empty fields	Optimised for judging
AScore-led	Auspiciousness at 116px with all 19 inputs — nine planetary conditions, seven matches, two penalties, composite — inside the same rule.	Event values truncate to one line each. The weekly expanded TODO (Q) drops to 11px. Content column is 448px wide.	Foot of the screen, full width, 26px cells — the tallest of the three, with the legend inline. Read as the score's own footnote.	Labels persist, values collapse. Empty today = hairline. Never populated = dotted slot + NOT YET TRACKED	Whether the scale is weighted right.
BContent-ledRecommended	K, P, Q and the assembled L at reading size — P at 21px, Q at 15px, L set against its inputs so the join can be heard.	Score and sky compress into one 296px instrument rail; T shows at 38px and its inputs at 9px glyph density. Fine reweighting needs a lean-in.	Directly under the spine, full width, 16px cells, no legend — the scale lives in the rail instead. Read before the day, not after it.	Same rule. The events rail is the region that yields when notes grow, which is why the sparse-day pass is drawn on this option.	Whether the writing is good.
CBoard	Nothing. Five regions at their own natural density, sky / score / events across the top, content and notes across the foot.	Nothing is demoted, so nothing reads first — the eye has to be aimed. Content sits in two columns and never reaches P at 20px.	Middle band, full width, 20px cells — the seam between the instrument row and the reading row, and the only element that spans both.	Same rule, and the cheapest here: the events region has room for all fifteen slots in two columns without truncation.	A row against the sheet it came from.

The density rule, named

Labels persist, values collapse. Every one of the 58 fields keeps a labelled slot at all times. A populated field shows its value; a field empty today shows a hairline in the value slot; a field never populated anywhere in the 365 rows shows a dotted slot and the words NOT YET TRACKED. Only AK and BI carry the third state.

Two substitutions to confirm

Glyphs render in Noto Sans Symbols 2 here because Physis is not in this file — swap the stack at build. The nine-step scale is drawn as eight: +3 and +4 are merged, because nine steps could not hold contrast at 3.6px cell width. F and H render as their stored characters with a query mark, pending Ivy naming them.

Score-led — the bet is that the pass is about the scale, so the arithmetic must be legible in one glance.

Bands: spine → sky → score / content / events → notes → year. 48 of 58 fields populated on this day. Band break: the year strip sits at the foot here and at the head in B — deliberate, and the one place the eye cannot travel straight down the page.

Friday, January 1st, 2027

Week {{ cd.A }}1

Lunar day {{ cd.J }}24

Month {{ cd.BL }}1

♀

Day ruler {{ cd.D }}Venus

← →
keys

Sky position

Sun block {{ cd.E }}{{ cd.F }}

♑︎n?

Moon block {{ cd.G }}{{ cd.H }}{{ cd.I }}

♍︎t?02:16ingress

Auspiciousness

+1{{ cd.T }}

Composite {{ cd.O }}+

Range in 2027−4 … +4

Planetary condition {{ cd.ABAJ }}

Content

Events

Review note

State

⏎ saves and advances

Colour auspice year {{ cd.T }}

Content-led — the bet is that what the generation did badly is in the writing, so the writing gets reading size.

Bands: spine → year → sky+score rail / content / events → notes. 48 of 58 fields populated. Band break: sky and score leave the horizontal band they hold in A and C and stand vertically in the left rail — the deliberate cost of the centre.

Friday, January 1st, 2027

wk 1 lunar 24 month 1 ruler♀Venus

← →
keys

Auspiciousness {{ cd.T }}composite + {{ cd.O }}

Condition {{ cd.ABAJ }}

Sky position {{ cd.ANAV }}

Sun {{ cd.E }}{{ cd.F }}♑︎n?

Moon {{ cd.G }}{{ cd.H }}{{ cd.I }}♍︎t?02:16

Content

Hygromanteia lunar-day suggestion {{ cd.K }} {{ k1 }}

Short TODO {{ cd.P }} {{ p1 }}

Weekly expanded TODO {{ cd.Q }} {{ q1 }}

Magical Todo, assembled from P + events {{ cd.L }} {{ l1 }}

Events

Review note

State

⏎ saves and advances

Board — the bet is that Ivy already knows where everything is, and the tool's job is to put a row of the sheet on one screen.

Bands: spine → sky / score / events → year → content / notes. 48 of 58 fields populated. Bands hold: sky and score sit at the same height as in A, and the year band is the only element crossing the instrument row and the reading row.

Friday, January 1st, 2027

wk {{ cd.A }} 1 lunar {{ cd.J }} 24 month {{ cd.BL }} 1 ruler {{ cd.D }}♀Venus

← →
keys

Sky position

Sun block {{ cd.E }}{{ cd.F }}

♑︎n?

Moon block {{ cd.G }}{{ cd.H }}{{ cd.I }}

♍︎t?02:16ingress

Auspiciousness

Score {{ cd.T }}

+composite {{ cd.O }}

Events

Colour auspice year {{ cd.T }}

Content

Review note

B′

Sparse-day pass on the recommendation — 2027-01-04, 43 of 58 fields, no events beyond the spine, and the note grown to four lines.

Thirteen of the fifteen event slots are empty today and two have never been populated at all. The note is the region permitted to grow; the events rail is the region that yields, because on a day like this it is the one with nothing to lose. Band break: notes takes 210px instead of 130px and the body row shortens to match.

Monday, January 4th, 2027

wk 1 lunar 27 month 1 ruler☽Moon

← →
keys

−2

Auspiciousness {{ cd.T }}composite − − {{ cd.O }}

Condition {{ cd.ABAJ }}

Sky position {{ cd.ANAV }}

Sun {{ cd.E }}{{ cd.F }}♑︎n?

Moon {{ cd.G }}{{ cd.H }}{{ cd.I }}♎︎q?

Content

Hygromanteia lunar-day suggestion {{ cd.K }} {{ k4 }}

Short TODO {{ cd.P }} {{ p4 }}

Weekly expanded TODO {{ cd.Q }} {{ q4 }}

Magical Todo, assembled from P + events {{ cd.L }} {{ l4 }}

Events

Review note

State

⏎ saves and advances

Recommendation

Option B

Content-led

Take B: the thing Ivy said she is hunting is what the generation did badly, and that lives in K, P, Q and the assembled L, which are the only fields in this set that have to be read rather than checked. It costs her the score at instrument density — nineteen inputs at 9px in a 296px rail, legible but not comfortable, which is a real tax if a session turns into an hour of weighing. What would change my mind is the answer to open question 3: if the reweighting is to be live, with sliders on the eleven inputs recomputing T on screen, the score needs the room and A becomes the right layout instead.

Still open for Ivy

1 · What F and H hold — drawn as raw glyph + query mark.

2 · Whether O rounds T or judges separately — drawn as a separate field.

3 · Live reweighting or recorded only — drawn as recorded only.

4 · Whether Jamie and Bruce read the notes — drawn as single-author.