English working draft
Progress before a lawyer assesses the case
Readers should be able to see the documentary basis growing, preliminary analyses being completed and fundraising approaching an agreed goal. These counters do not require a lawyer to have assessed every submitted decision beforehand. They do not calculate the probability of unlawful conduct.
Three stages with two parallel tracks
- Collection and preliminary analysis. Decisions are received, duplicates checked and analysis checkpoints recorded. Crowdfunding can proceed in parallel under an agreed account arrangement with the lawyer responsible for holding the funds.
- Ready to invite counsel. Enough unique decisions must have been provisionally analysed, and the latest account report must confirm that available funds meet the funding target. A lawyer can then be invited to assess the material.
- Independent legal assessment. The lawyer assesses the basis, cost risk and whether to accept an instruction. Later spending requires a separate mandate, agreement and budget. Holding the account is not an agreement to conduct litigation.
Completed analysis or a fully funded target never constitutes a decision to sue. The two tracks are displayed separately, and the lawyer’s later assessment has its own status.
What the counters show
| Measure | What is counted |
|---|---|
| People with decisions | Unique participants with at least one accepted decision. Several documents from the same person do not count as several people. |
| Decisions received | Unique accepted decisions after duplicate checking. Repeated uploads do not inflate the count. |
| Machine-analysed decisions | Decisions with a preliminary machine-analysis record for every planned checkpoint. This work precedes the lawyer stage; gaps in documentation may remain unresolved. |
| Machine review flags | Points where a preliminary analysis has identified a possible need for checking. A flag is not a percentage, proof of intent or an established legal violation. |
| Legal provisions considered | Different provisions considered within the defined analysis. The same provision appearing in many decisions counts once here. |
| Checkpoints assessed | One recorded assessment of one decision against one particular provision. This count differs from the number of people or decisions. |
| Decisions reviewed by a person or lawyer | Separate later review stages. Machine-analysis provenance is retained so completed work does not disappear when a person later reviews it. These counts may overlap and must not be added together. |
| Reconciled available funds | Confirmed receipts minus completed spending and refunds. Pledges and gross receipts alone do not establish available funding. |
How possible issues are described
- Machine flag — possible need for checking: A preliminary analysis signal needs closer examination.
- Machine assessed — no flag: The limited analysis produced no such signal. This does not guarantee that the decision is correct.
- Priority legal review: A question needs human follow-up; priority is not a calculated probability of a violation.
- Possible issue noted by a lawyer: A later legal review has recorded a possible issue. This label is not a court finding.
- Documentation missing: Necessary material still needs to be obtained.
- No issue identified in this review: The limited review did not establish a reason for a note.
Dated reports and limited demonstration targets
Reports show counts and changes since the previous checked release. Small groups and small changes are withheld when they might reveal participants. Names, prisoner numbers, document excerpts, donor names and bank details do not belong in the public overview.
The analysis demonstration uses three fictional periods with 40, 80 and 120 machine-analysed decisions. It shows progress before lawyer review. The separate account demonstration uses targets of 100 provisionally analysed decisions and NOK 5,000 in reconciled, available test funds. These are demonstration values, not agreed production terms.
A separate field retrieves only a confirmed account report from the server. If there is no report, it shows that no confirmed report is available; it never invents an amount or assumes a zero balance. Synthetic analysis counts are not combined with actual account figures to portray the project as ready.
What remains before accepting real contributions
Secure submission, checks for unique participants and decisions, analysis provenance, data-controller responsibility, method review, retention rules and responsible publication must be established. This does not require a lawyer to review every decision in advance. Local MAITE documents and working notes are never automatically submitted to the study.
The current private demonstration does not receive real study submissions. A separate source-bound analysis workbench is available for reviewing defined inputs and outputs; matching quotations to supplied sources is not a legal conclusion.
Try the private demonstration · View the separate account report · Read about account custody and later legal work.