Thank you for reporting this.
“Other traffic” is calculated as:
other = page_view_anon + page_view_logged_in
- page_view_anon_browser - page_view_logged_in_browser
These counts are recorded separately. If browser views exceed the total recorded anonymous and logged-in page views, “Other traffic” becomes negative.
One possible explanation is that someone discovered the browser-tracking endpoint and sent repeated requests directly, increasing browser views without corresponding page loads.
That would explain increased “Anonymous” traffic. I think that if you enable our recent crawler detection (Improving how we detect and flag likely crawlers in your traffic data), that traffic would be classified as likely crawlers.
In addition, I have a PR that clamps “Other traffic” to zero, preventing negative values from appearing.