Privacy

Who is responsible

Threadbaire is run by me, Lida Liberopoulou. I'm responsible for the personal data this site collects. Contact: lida@threadbaire.com.

Reading the site

When you visit, these services receive standard request data (such as the page, referrer, browser type and IP address) so the site can load and I can see which pages are read:

• Vercel hosts the site and keeps operational server logs.
• GoatCounter counts page views and referrers, without cookies. It also counts requests that don't run scripts, such as feed readers, AI tools and direct downloads of the markdown files.
• Cronitor measures page load performance, without cookies.
• Tailwind CSS is loaded from its content delivery network to style the pages.

For requests that don't run scripts, the site sends GoatCounter a short record of its own instead: which file was requested, and whether the request came from a declared crawler or reader, a browser, or something presenting itself as one. For browser-like requests this includes the browser's user-agent string, from which GoatCounter keeps only the browser and system name, and an approximate location (country, region and city), which Vercel works out from the IP address. For other requests it includes the country only. The IP address itself is never sent to GoatCounter.

I use this only in aggregate: which articles are read and where readers come from. I don't build profiles or try to identify visitors. The legal basis is my legitimate interest in running the site and understanding its readership.

The share buttons are plain links. Nothing is sent to LinkedIn or X unless you click one, and then their own policies apply. The same goes for Substack, Medium, GitHub and Hugging Face, where some of the work is also published. The Copy for AI button copies text to your clipboard; nothing is sent anywhere.

Writing to me

The contact form is handled by Formspree and delivered to my email, which is hosted by Google Workspace. Emails you send directly also arrive there.

• I use your name, email and message only to reply and, if we agree to work together, to scope the work. Compute Watch contributions are also used as described below.
• Enquiries are never put into AI tools.
• I keep correspondence only as long as it's useful for the conversation, and delete it sooner if you ask.

The legal basis is my legitimate interest in answering you, or taking steps you've asked for before an agreement.

Compute Watch contributions

If you send prices, a correction or a question for the Compute Watch:

• What I use it for. To check against the public indexes, to combine with other reports into broad ranges, and to reply to you. Once at least five people have reported on a chip, I send the combined ranges to everyone who contributed, and later publish them as ranges in a separate, labelled section of a note. Individual figures are never published.
• Counter-arguments. If yours is the strongest of the month, I may summarise or quote it in a note, without your name unless you ask to be credited.
• AI tools. Prices you report, and your name, email, employer or anything else that could identify you, never go into AI tools. Prices are logged by hand and combined into ranges by a script on my own computer. Arguments, corrections and questions may be discussed with AI tools while I draft an answer, with your name and identifying details removed, on accounts where training on my data is switched off. If you'd rather your message stayed out of AI tools entirely, say so and it will.
• How long I keep it. Reports, for as long as the watch runs, so ranges can be compared month to month. Your email, for as long as you want to receive ranges. When the watch ends, I delete both within 12 months.
• Deleting. You can ask me to delete your report and your email at any time. A range already published can't be taken back, but your report won't be used in future ones.
• Confidential material. Please don't send anything your contract doesn't allow you to share. Contributions are held in confidence.

The legal basis is my legitimate interest in the research, and in answering you.

Commissions

Commissions are confidential. Nothing about your question, exposure or decision is published or referenced without your agreement.

My research is AI-assisted. For commissioned work, how AI tools are used is agreed with you at scoping. By default, material that identifies you or your position goes into AI tools only in anonymised form, or on accounts where training on my data is switched off.

If you commission work, the engagement terms set out who contracts with you and how your material is handled from that point.

Where data goes

Some of the services above are based in, or process data in, the United States. They rely on recognised safeguards such as the European Commission's standard contractual clauses. Security in transit is handled by HTTPS throughout.

Your rights

Under the GDPR you can ask to see the data I hold about you, correct it, delete it, or object to how I use it. Email lida@threadbaire.com from the address you used. I'll reply within a month.

If you think I've handled your data wrongly, you can complain to the Hellenic Data Protection Authority or the authority in your own country. I'd appreciate the chance to put it right first.

Changes

If how I handle data changes, I'll update this page and the date at the top.

12 October 2026: Compute Watch arguments and questions may now be discussed with AI tools, without your name. Reported prices and identifying details still never are.