the evolving website
Your information

Privacy and AI

Last reviewed

This website is an experiment owned by Andrew Marsland. Here’s what happens when you use it.

Reading and searching

  • Cloudflare hosts and protects the website. It processes network requests, including IP addresses and browser information.
  • Cloudflare Web Analytics measures visits and page performance. Cloudflare describes it as privacy-first analytics; hosting still processes network information.
  • Keyword search runs in your browser. The website does not save your search terms.
  • The optional homepage self-check runs in your browser. Its answers are neither sent nor saved.

Optional Google Analytics

Google Analytics 4 is available only when you choose Allow analytics. Declining leaves reading, Ask AI and the study available. We block the Google tag completely before consent; no cookieless measurement is sent while it is off.

When enabled and allowed, Google receives page visits and fixed action names for Ask AI and the study. It also processes network and device information, including your IP address when receiving requests. Cookies distinguish browsers; they do not prove a visitor is human. We exclude questions, answers, study responses, URL query strings and fragments, and referring URLs. Advertising features and Google signals are disabled.

Your choice is stored in this browser for up to 180 days. Use Analytics preferences at the foot of any page to change it. Declining removes this site's GA cookies and reloads the page to stop the loaded tag. It does not erase data already received by Google. GA cookies are configured for up to 180 days; provider retention and browser rules also apply. Blocking cookies or analytics means reports will be incomplete.

A study-complete event means someone chose “I’ve finished the study”. It does not prove task success or that a LinkedIn reply was sent. Read how Google uses information from partner sites. Processing may occur outside Australia. User and event retention is set to two months without activity-based resets; this does not set the retention of every aggregate report or provider record.

Asking AI

Choosing Ask sends your question to a Cloudflare Pages Function. Ordinary AI answers also send it and selected public website text to Workers AI. Latest-update and named journal-topic answers return published excerpts; dated comparisons return published summaries. Those paths use no model call. Don’t include private information.

We don’t save your questions or answers in an application database. A shared AI limit stores only a UTC date and the number of reserved requests in the existing D1 database. Rows older than seven days are removed on the next reservation; inactive rows may remain longer. Cloudflare’s Workers AI data policy explains its processing, operational metadata and retention exceptions. Cloudflare says it does not use these inputs to train models.

AI answers can be wrong. Check the linked sources; keyword search is available without asking AI.

Leaving feedback

  • Yes / No feedback is optional. We save totals by page and Sydney date, without names, comments or visitor IDs.
  • Roadmap preferences are optional. Submit preference saves a daily count for the selected idea in the existing Cloudflare D1 database. We do not save names, comments or visitor IDs. Test-marked submissions are counted separately. Public totals help review priorities; they do not automatically set the roadmap order.
  • Aggregate counts are public for the last 90 days. Reports can reuse a short cache and may take a minute to update. Older rows are removed on the next submission, so inactive stored rows may remain longer.
  • Browser session storage remembers a page vote or roadmap preference for that day. Closing the session normally clears it. Temporary IP-based throttling helps limit abuse; IP addresses are not written to the feedback database.

Votes can include repeats or bots. They do not identify unique readers or prove the website is useful.

Request verification

Choosing an AI or feedback action loads Cloudflare Turnstile. Cloudflare processes security signals, including IP and browser information, and returns a short-lived token. The server checks the token, hostname and action before backend work. The application does not store verification tokens or use them to identify visitors. Reading pages and keyword search do not require this check.

See Cloudflare’s Turnstile privacy information. Verification is not a unique-human count.

Providers and retention

Application storage and provider processing are different. Provider security records, operational metadata and backups follow Cloudflare’s policies; this page does not promise that all data is deleted within 90 days or stays in Australia.

Read Cloudflare’s privacy policy and Web Analytics details. Following an external link takes you to a service with its own privacy practices.

Questions, corrections or access problems

Contact Andrew through LinkedIn. A LinkedIn account may be needed. Describe the page and problem; don’t send sensitive information.

Owner and content reuse · How we use reader evidence explains the measurement plan.

Was this page useful?