When you click 'fingerprint me', this site will test whether you've visited ~50,000-1,000,000 sites; it'll take one minute. At the end, you'll see a list of what sites were found, and top 10 users who are similar to you. It will show their name if and only if both you and they consent to have your names shown.
It may take a couple minutes to process your data. You'll see a live status report as it's working. It's first come first serve, so you may need to wait for someone else's info to get processed first. You'll see the queued jobs count going down as it does.
Please do not open a new tab/window on this site while the scraper is running, as it can disrupt your session and prevent you from seeing your results. Occasionally, a scraping will hang - you can tell because the job queue is at 0 but it's not calculating your results. If that happens, just restart it for now; I'm working on fixing the underlying issues.
Enjoy.
CSS Fingerprint is a research project inspired by the EFF's Panopticlick.
Its intent is to see how well the CSS history hack can be used with "fuzzy" artificial intelligence techniques to uniquely fingerprint users despite changes in their browsing history, even on new computers or new browsers, and to tell how socially/culturally similar any two users are.
The AI is now active, but in order to get better results, I need more data.
To help out, please visit this site from multiple different browsers / computers that you own, on multiple days, using the same input each time.
No personally identifiable data will be shared with anyone except other EFF-friendly researchers who agree to keep it confidential. Aggregate and anonymized data will probably be used in a future paper.
Thanks!
- Sai Emrys (saizai)
Email: saizai.com, cssfingerprint@
Browsers each have idiosyncracies and require different methods to test efficiently. Here are the results of yours.
A correct test should say true for cssfingerprint.com, false for the garbage, and true for google if you actually use this browser at all.
This test only tells the server your user-agent and the information in the table above. It tests 1000 URLs but doesn't report the results of that, only the timing. This will help me to create better scraping methods.