krisrockwell.io  /  neuromancer-semiotics  /  data  /  Data

Data

██╹the coded corpus · nine flat files

Everything the study was computed from: the coded matrix, the instrument, the blind re-code, and the tests that refuted it.

Nine flat files. Each is rendered here as a readable view; each view links back to the raw file it was built from. The flat files are canonical — the whole matrix can be recomputed from them at any point, which is what makes the audit possible.

The MatrixEvery coded passage after the second pass: the chapter document, the passage id, and each code applied to it. Every number on the study page is computed from this file. MODE codes carry the validated triad; all other families are set in neutral ink.codes_matrix_pass2.txt · 217 lines · 16,998 bytes · raw
Co-occurrenceEvery co-occurrence matrix the study reports: cue against mode, target against mode, target against verdict, target against cue, verdict against inference.pass2_tables.txt · 81 lines · 6,340 bytes · raw
The InstrumentThe codebook itself: every code with its definition, what it includes and what it excludes. This is the instrument that was applied — and the only thing the blind coder was given besides the passages.codebook.txt · 69 lines · 16,421 bytes · raw
Blind InputThe 36 passages handed to an independent coder — chapter, passage id, and the text. No codes, no context, no argument.blind_input.txt · 36 lines · 3,735 bytes · raw
Blind OutputWhat the independent coder returned, against passage id. Compare with the matrix to see where the two readings part.blind_output.txt · 36 lines · 3,137 bytes · raw
AgreementAgreement statistics by family and every passage-level disagreement. The verdict family — which carries every headline — agreed 33% of the time.blind_test_results.txt · 43 lines · 2,677 bytes · raw
SelectionA blind selection test on three chapters, spans only. The independent reader marked passages this study never saw: 38% coverage overall.selection_test_results.txt · 66 lines · 6,592 bytes · raw
Blind SpansThe raw spans an independent reader selected across three chapters, before any codes were applied. The basis of the selection test.blind_spans.txt · 72 lines · 9,777 bytes · raw
Loop AAn early shakedown pass: eleven passages pulled to test whether the codebook could be applied at all before the full corpus was attempted.loopA_test.txt · 11 lines · 4,065 bytes · raw

Rendered from by scripts/build_data_pages.py. The flat file is canonical; this page is a view of it. · back to the study