On this page
Deterministic result
| Calculation over code rows | Result | Interpretation |
|---|---|---|
| Records in the approved dataset | 2 | Two rows match dataType = code |
| Training records | 2 | Both rows use the training deal type |
| Live-access-only records | 0 | No code row is classified only as live access |
| Acquisition or sale records | 0 | No code row has those deal types in this snapshot |
| Rows with a disclosed USD value | 0 | No valueUsd is present for either row |
| Rows with undisclosed value state | 1 | One row is explicitly marked undisclosed |
| Rows with a reported non-cash value text | 1 | One row has a reported usage benefit but no USD figure |
The counts above are deterministic filters over the approved dataset. They are not a market-size estimate, demand score, or performance result. The dataset’s own rules keep undisclosed and non-comparable values out of numeric totals.
Record A: source-code training invitation
The approved tracker row carries source date 3 Jun 2026. The row records a training classification, an undisclosed value, and an invitation for developers to provide live and archived application source code 404 Media, (opens in a new tab). The report describes an arrangement in which developers retain intellectual-property ownership and grant a nonexclusive license for the reported use. Read the linked article and the underlying invitation before repeating any rights statement. The public report does not supply a comparable USD figure for this dataset row.
General information, not legal advice.
Record B: opt-in anonymized coding sessions
The public opt-in coding-session program report IT Brief, (opens in a new tab) is dated 17 Sep 2026. The approved row records training, no USD value, and a reported non-cash usage benefit associated with opt-in anonymized coding sessions used to train an open-weight model. The source describes an ability to leave the program and a data-processing agreement context. Those details should be read from the source, not generalized to every coding session or every license.
General information, not legal advice.
What the records support
- Public reports can describe different rights structures: a source-code license invitation and an opt-in session-data program.
- A training label does not prove that live access, acquisition, or ownership transfer occurred.
- An undisclosed or non-cash value is not a zero and cannot be compared with a disclosed USD amount.
- The source date, record date, value state, and rights wording must remain adjacent in any citation.
- A record can support a research question without establishing that a particular owner, buyer, or recipient will transact through Rightmark.
Reproducibility worksheet
| Step | Input or rule | Recorded result |
|---|---|---|
| 1 | Select rows where dataType = code | 2 rows |
| 2 | Group by dealType | training: 2 |
| 3 | Count non-null valueUsd | 0 |
| 4 | Classify valueType and value text | 1 undisclosed, 1 reported non-cash text |
| 5 | Open each dated source link | 2 source reviews required |
| 6 | Write caveats beside each assertion | No price or relationship inference |
Clearly fictional worked example
SYN-CODE-ROW is a fictional third row with dealType = training, valueType = undisclosed, and no source URL. The correct deterministic result is to exclude it from a published analysis until a dated public source is supplied. It cannot be used to inflate the count or to imply a market amount. This is a data-quality example, not a Rightmark record.