# Series SR1: report

Starting from the same proposition (S21's block-in), does evaluative critique change what painters do with it, compared with operational validation alone? Read as distributions of choices, not scores.

Protocol sha256 b9ca6fa17115560e, frozen 2026-10-10T14:58:28+00:00. 6 runs (3 per condition).

## Every run

| run | condition | end | final | material sittings | looks | branches | replay s | critiques | tests | blind |
|---|---|---|---|---|---|---|---|---|---|---|
| R01 | critique_off | completion_proposed | main:6 | 8 | 2 | 1 | 260.966 | 0 | 0 | B01 |
| R02 | critique_off | hold | b7:1 | 8 | 4 | 2 | 341.478 | 0 | 0 | B03 |
| R03 | critique_off | completion_proposed | path:1 | 8 | 2 | 3 | 178.47 | 0 | 0 | B05 |
| R04 | critique_on | hold | main:9 | 8 | 3 | 1 | 224.344 | 3 | 0 | B06 |
| R05 | critique_on | hold | b:4 | 8 | 5 | 2 | 325.873 | 3 | 1 | B02 |
| R06 | critique_on | hold | work:14 | 8 | 8 | 2 | 295.783 | 3 | 1 | B04 |

## Near snow (near_snow)

What happened to the lower third of the field relative to the block-in.

Annotator: RP

| value | critique_on | critique_off |
|---|---|---|
| left-empty | 0 | 0 |
| value-gradient-or-shadow | 2 (B02, B04) | 0 |
| marks-or-objects-added | 1 (B06) | 3 (B01, B03, B05) |
| other-change | 0 | 0 |
| uncertain | 0 | 0 |

Annotator: blind subagent (Claude, fresh context)

| value | critique_on | critique_off |
|---|---|---|
| left-empty | 0 | 1 (B01) |
| value-gradient-or-shadow | 2 (B02, B04) | 0 |
| marks-or-objects-added | 0 | 1 (B05) |
| other-change | 1 (B06) | 1 (B03) |
| uncertain | 0 | 0 |


## Value key (key)

The final value order relative to S21 (water darkest, snow mid, sky lightest).

Annotator: RP

| value | critique_on | critique_off |
|---|---|---|
| held | 0 | 0 |
| snow-lighter-than-sky | 2 (B04, B06) | 1 (B01) |
| light-accent-or-glow-added | 0 | 2 (B03, B05) |
| darkened-overall | 1 (B02) | 0 |
| uncertain | 0 | 0 |

Annotator: blind subagent (Claude, fresh context)

| value | critique_on | critique_off |
|---|---|---|
| held | 0 | 0 |
| snow-lighter-than-sky | 0 | 2 (B01, B03) |
| light-accent-or-glow-added | 1 (B02) | 0 |
| darkened-overall | 0 | 0 |
| uncertain | 2 (B04, B06) | 1 (B05) |


## The brook at the ruin (delivery)

How the darkest value behaves where the brook turns toward the ruin.

Annotator: RP

| value | critique_on | critique_off |
|---|---|---|
| delivers-to-ruin-foot | 0 | 1 (B03) |
| stopped-or-frozen-before-ruin | 3 (B02, B04, B06) | 2 (B01, B05) |
| route-changed | 0 | 0 |
| uncertain | 0 | 0 |

Annotator: blind subagent (Claude, fresh context)

| value | critique_on | critique_off |
|---|---|---|
| delivers-to-ruin-foot | 0 | 1 (B01) |
| stopped-or-frozen-before-ruin | 3 (B02, B04, B06) | 1 (B03) |
| route-changed | 0 | 1 (B05) |
| uncertain | 0 | 0 |


## Emblem-like closure (emblem)

Whether an element was added that reads as a symbol closing the meaning (bird, tree, cross, lit window, a gesture made expressive).

Annotator: RP

| value | critique_on | critique_off |
|---|---|---|
| none-added | 0 | 0 |
| symbol-added | 0 | 0 |
| figure-made-expressive | 0 | 0 |
| uncertain | 3 (B02, B04, B06) | 3 (B01, B03, B05) |

Annotator: blind subagent (Claude, fresh context)

| value | critique_on | critique_off |
|---|---|---|
| none-added | 2 (B04, B06) | 2 (B01, B03) |
| symbol-added | 0 | 0 |
| figure-made-expressive | 0 | 0 |
| uncertain | 1 (B02) | 1 (B05) |


## The ruin (ruin)

The ruin's final treatment against the sky.

Annotator: RP

| value | critique_on | critique_off |
|---|---|---|
| soft-barely-off-sky | 1 (B06) | 2 (B03, B05) |
| firm-silhouette | 2 (B02, B04) | 1 (B01) |
| lost-or-removed | 0 | 0 |
| uncertain | 0 | 0 |

Annotator: blind subagent (Claude, fresh context)

| value | critique_on | critique_off |
|---|---|---|
| soft-barely-off-sky | 0 | 1 (B05) |
| firm-silhouette | 3 (B02, B04, B06) | 2 (B01, B03) |
| lost-or-removed | 0 | 0 |
| uncertain | 0 | 0 |


## The walker at 600px (walker)

Whether the walker reads as a figure in the final 600px state.

Annotator: RP

| value | critique_on | critique_off |
|---|---|---|
| reads-as-figure | 3 (B02, B04, B06) | 3 (B01, B03, B05) |
| speck-or-post | 0 | 0 |
| lost | 0 | 0 |
| uncertain | 0 | 0 |

Annotator: blind subagent (Claude, fresh context)

| value | critique_on | critique_off |
|---|---|---|
| reads-as-figure | 2 (B02, B04) | 2 (B03, B05) |
| speck-or-post | 1 (B06) | 1 (B01) |
| lost | 0 | 0 |
| uncertain | 0 | 0 |

## Deviations from how the protocol ran

- D1 (R03, R06): Studio bug fixed mid-series: after a failed sitting, the next sitting was built on the failed log and replayed the broken chunk, so it failed too. Fixed so a sitting builds on the last successful state, and only successful material sittings count toward the sitting allowance (failed attempts stay recorded; their replay seconds still count). R03 (critique_off) and R06 (critique_on) each lost a second sitting to the bug, not to their own error, and both worked around it by branching from an earlier state. The fix applies to every run from here on; no other run had a failed sitting when it was made.

## Disagreements

- B01 near_snow: {'RP': 'marks-or-objects-added', 'blind subagent (Claude, fresh context)': 'left-empty'}
- B01 delivery: {'RP': 'stopped-or-frozen-before-ruin', 'blind subagent (Claude, fresh context)': 'delivers-to-ruin-foot'}
- B01 emblem: {'RP': 'uncertain', 'blind subagent (Claude, fresh context)': 'none-added'}
- B01 walker: {'RP': 'reads-as-figure', 'blind subagent (Claude, fresh context)': 'speck-or-post'}
- B02 key: {'RP': 'darkened-overall', 'blind subagent (Claude, fresh context)': 'light-accent-or-glow-added'}
- B03 near_snow: {'RP': 'marks-or-objects-added', 'blind subagent (Claude, fresh context)': 'other-change'}
- B03 key: {'RP': 'light-accent-or-glow-added', 'blind subagent (Claude, fresh context)': 'snow-lighter-than-sky'}
- B03 delivery: {'RP': 'delivers-to-ruin-foot', 'blind subagent (Claude, fresh context)': 'stopped-or-frozen-before-ruin'}
- B03 emblem: {'RP': 'uncertain', 'blind subagent (Claude, fresh context)': 'none-added'}
- B03 ruin: {'RP': 'soft-barely-off-sky', 'blind subagent (Claude, fresh context)': 'firm-silhouette'}
- B04 key: {'RP': 'snow-lighter-than-sky', 'blind subagent (Claude, fresh context)': 'uncertain'}
- B04 emblem: {'RP': 'uncertain', 'blind subagent (Claude, fresh context)': 'none-added'}
- B05 key: {'RP': 'light-accent-or-glow-added', 'blind subagent (Claude, fresh context)': 'uncertain'}
- B05 delivery: {'RP': 'stopped-or-frozen-before-ruin', 'blind subagent (Claude, fresh context)': 'route-changed'}
- B06 near_snow: {'RP': 'marks-or-objects-added', 'blind subagent (Claude, fresh context)': 'other-change'}
- B06 key: {'RP': 'snow-lighter-than-sky', 'blind subagent (Claude, fresh context)': 'uncertain'}
- B06 emblem: {'RP': 'uncertain', 'blind subagent (Claude, fresh context)': 'none-added'}
- B06 ruin: {'RP': 'soft-barely-off-sky', 'blind subagent (Claude, fresh context)': 'firm-silhouette'}
- B06 walker: {'RP': 'reads-as-figure', 'blind subagent (Claude, fresh context)': 'speck-or-post'}

## Caveats

- 3 runs per condition: distributions are descriptive, not evidence of an effect.
- A change in distribution is not an artistic improvement.
- No default score is computed.
- Only the categories defined in the frozen protocol were counted.