# List comparison evidence v1

Synthetic inputs only. WorkToolMap and comparetwolists.net are both maintained by Jack.
This is a disclosed self-owned product test, not an independent ranking.

## Reproduce
1. Open https://comparetwolists.net/?analytics_test=1. Use the footer Analytics control to turn analytics off for maintenance.
2. Paste base-a.txt and base-b.txt. Run F01 (trim on, case-sensitive), F02 (trim on, ignore case), F03 (trim off, case-sensitive). Read expected.json, generated independently before any run.
3. Import the base CSVs with comma delimiter, header on, ID (second column). For base.xlsx select AInputs/BInputs and ID (column B), header on. Readme is deliberately the wrong worksheet. Expand the first result to inspect CSV record 2/column 2 or AInputs/B2.
4. Paste export-a.txt/export-b.txt with default rules. Each group must have 125 items. Copy every group; download each as TXT/CSV and all groups as XLSX. Compare all values and their order to expected.json, not only the first preview page.
5. Change input or Ignore case after a run. Old copying/save output must be unavailable until rerun.

TXT outputs have no header. CSV/XLSX outputs have a Value header. The XLSX input uses text cells; preserve text types in other software. Values and order are verified; different generated ZIP/XLSX bytes on a later run do not necessarily imply different values.

## File map
base-* and base.xlsx: small reference input. export-*: 250 items per side.
expected.json: independent specification, frozen before runs. observed.json: browser observations and independent export reread checks.
observed-onlyInA.*, observed-onlyInB.*, observed-inBoth.*: actual TXT/CSV downloads.
observed-all-groups.xlsx: actual complete export; observed-base.xlsx: actual imported-base result.
manifest.json: byte sizes and hashes, excluding itself.

## Scope and failures
One online product tested in Chrome 152 against production commit 01bccc32c0026d9542f9736e5300bcb5c2acec8f. Clipboard permission was granted to the isolated test browser. No real customer lists were used.
The first automation read transient empty rows before asynchronous rendering; visible-row/source waits passed unchanged expectations. This was a measurement timing error, not a demonstrated product defect. The existing Cloudflare beacon CSP warning did not prevent comparison.
Excel: official-document review only, NOT_RUN. Excel is installed, but native app control is disabled and no live workbook was verified. The guide gives an EXACT/SUMPRODUCT helper-column method for these same inputs. No Excel performance or export superiority claim is made.
Not tested: all Unicode rules, formulas, hidden/filtered rows, formatted numeric cells, multi-column joins, quantity reconciliation, lost-digit recovery, physical mobile devices, other online competitors.