Research safeguards & help

Our operating objective: Make it substantially easier for a team to reach, inspect and revise a defensible conclusion.

Open record formats, portability & stewardship

Recover work, understand access, and get a question or dispute into the owner's queue.

Start with one question

Start a private research draft and inspect the original sources.
Save it to your account, then share a decision only when ready for invited-library members to read it.
Open the decision record to inspect evidence, invite challenges and record responses.
Export important work and verify recovery before relying on it for an essential project.

Graphs have text alternatives. Forms use labeled native controls. Report an accessibility barrier below; a full WCAG conformance audit has not been completed.

Checking account access…

Saving & tested recovery

Account saves reject an outdated version rather than silently overwriting another device's work. The last five successful account versions are retained automatically, including the version being replaced. Browser-only edits are not server backups.

Version history uses the same database. Keep an export in a separate trusted location for protection against a service-wide loss. Automatic off-site backups and a production recovery-time guarantee have not been established.

Verify and recover an invited-library archive offline

Download the recovery script and schema. With a Node.js version supporting node:sqlite, run:

node restore-library.mjs backup.json NEW_RECOVERY_FOLDER recovery-schema.sql

The tool checks archive and attachment hashes, restores records and relationships into a new SQLite database, and refuses an existing output folder. Hashes detect accidental changes; they do not authenticate the archive's author. This is an offline recovery drill, not a production restore or automatic reinstatement of access.

Access & ownership

Current access rules
Role	Access and responsibility
Signed-in researcher	Own account draft and its revisions; own support requests. Personal drafts are not readable through another researcher's account.
Invited member	Read and download the shared library; append attributed challenges and reviews. Sharing is to all invited members, not a private subgroup.
Decision author	Own protocols, results, revised conclusions and source reassessments. A response does not erase another reviewer's objection.
Original reviewer	Resolve or reopen their own challenge with reasons and evidence. This does not certify the science.
Site owner	Manage membership, export library membership/data, and triage support. Administration does not confer scientific review authority.

Attribution is preserved in exports. Uploaders must have permission to share material; uploading does not establish ownership or grant an open license. Removing membership stops future account access but cannot recall downloaded copies. The separate isolated-project workflow provides owner-only drafts and selected, named-reviewer packets; the invited library still has no private subgroup boundary. Ownership transfer is not implemented. See current security and processing boundaries.

Support & disputes

The founder, Jae DeSouza, is the initial support contact: Jae@astrasciencenetwork.org. Requests below are stored for the owner; no automatic email or guaranteed response time is provided. Do not include passwords, credentials or sensitive participant data.

Support process and escalation

Support is founder-operated and best effort. A stored request is not confirmation that a person has read it; no staffed on-call service or contractual response deadline is active.

Submit a request and retain its request ID. Include the project or packet ID, time, error text and requested action; keep private research and credentials out of the message.
For suspected unauthorized access or altered review history, pause sharing, retain a local receipt and use the support email as well as this form. For save failures, export pending work before closing the page.
The owner records triage, questions, actions and resolution in the ticket history. Requesters can reply or reopen a ticket.
Close a recovery or integrity incident only after recording what was restored or checked, the verification result and remaining uncertainty. A support resolution never resolves a scientific objection.

Process defined; response-time performance and independent incident handling have not been validated.

Who handles a scientific disagreement?

Record the exact disputed claim, source passage and requested correction.
The decision author responds with evidence. The original reviewer controls their own objection's resolution.
For an appeal or misconduct concern, submit a dispute request. The owner coordinates triage and must disclose conflicts.
If independent adjudication is needed, a qualified, non-conflicted reviewer must agree to the role. No independent panel is currently appointed.
Record the outcome and reasoning without deleting the original disagreement. Closing a support ticket does not resolve or alter a scientific claim.

The founder must not present their own disputed work as independently adjudicated. If no suitable reviewer is available, the dispute remains unresolved.

If something goes wrong

Responsible owner: Jae DeSouza, Jae@astrasciencenetwork.org. No alternate responder or staffed on-call coverage is established. Response is best effort; this is a procedure, not a response-time promise.

Record and classify. Record discovery time, affected project IDs, symptoms, reporter and known scope in a support ticket. Treat suspected unauthorized access as urgent; distinguish outages from lost-record incidents. Do not put research secrets or credentials in the ticket.
Contain. For suspected disclosure, pause sharing, revoke affected reviewer/feed access, and rotate compromised credentials through their provider. Preserve relevant logs and receipts. For an outage, avoid repeated uploads and notify the hosting provider. For lost work, stop editing the affected copy and export pending work before recovery.
Inform affected users. The owner contacts known affected users directly using the established support contact, stating what is known, what remains unknown, what users should do, and when the next update will be given. Do not disclose private research in a public notice. Follow applicable contractual and legal notification obligations; no automated notification is claimed.
Recover and verify. Restore into a separate test project where possible. Compare question, evidence, objection, decision, next test and outcomes. Recheck revoked access. Record actual time and any missing data. Do not overwrite the last good backup.
Close with evidence. Record cause, actions, verification results, unresolved risks and prevention work. Inform affected users and allow reopening. Closing a ticket never resolves a scientific objection.

Practice exercise: use synthetic data to rehearse a leaked reviewer link, a hosting outage and a missing record. Record who responded, elapsed time and failures. No completed exercise or independent incident assessment is claimed.

Phone and separate-device recovery check

Complete recovery procedure and evidence checklist

Use a clearly labeled test project with non-sensitive content. This checklist is not a claim that a physical-phone or disaster-recovery test has passed.

Complete all six steps. Save privately and download both the decision record and full project backup. Open the actual files and inspect their contents.
Keep a copy outside ASTRA in a trusted location you already control. Record filename, date, project ID and backup checksum where available. External source files need their own backups.
On another physical device, sign in to the same account and open the saved project. Check question, evidence, objection, decision, next test and outcomes. Close and reopen the browser and repeat.
Separately test backup recovery: in Save options, select the private backup and choose Recover backup as a new private project. This must not overwrite the original project or reactivate reviewer/feed access.
Record device model, OS, browser, date, completed download, content comparison, elapsed time, help needed and failures. A desktop-sized browser window is not a phone test. Account reopening alone is not proof of recovery after provider-wide data loss.

Operating costs & pricing

No subscription or payment is started here. Sustainable paid pricing has not been validated. Any future paid offer needs explicit terms, usage limits and customer agreement.

Paid pilot and subscription readiness

Pricing remains unpublished while payment collection is unavailable; pilot enquiries are for scope and availability; there is no active checkout or validated subscription price.

Offer	Current status	Required before payment
Research pilot	Scope discussion available; no automatic purchase	Written deliverables, duration, limits, price, support contact, export arrangements and cancellation terms accepted by both parties
Team subscription	Not launched	Measured operating/support costs, tested willingness to pay, reliable account recovery and defined service terms
Institutional support	No SLA offered	Named support coverage, agreed escalation and response commitments, demonstrated recovery and separate contract

Track actual pilot support hours, infrastructure cost, completion and renewal decisions before setting recurring prices. Forecasts and calculator results are not validated demand.

Discuss a pilot scope

Plan a price using your actual costs

This estimate is not a quote. Enter invoices, support labor and other costs; unknown costs must not be treated as zero.

Price floor = total monthly cost ÷ paying teams ÷ (1 − target margin). Demand, taxes, churn and cost variability still need evaluation; this does not establish profitability.

Current interactive limits include 10 MB per upload, 1.4 MB per account draft, five retained draft versions, and a 16 MB attachment limit for an invited-library backup. These are product bounds, not a monthly spending cap. Actual infrastructure invoices and a billing-level budget alert remain an operator responsibility.