In case you missed it: Cloudflare shipped its Precursor yesterday, and this is a big shift in how bot detection works. Instead of just looking at your initial landing request, it scores the behaviour of the entire session. This means mouse arcs, keystroke timing, even the micro-tremors of a human hand. This completely kills the oldest trick in data scraping: refreshing, rotating your identity, and starting over. The bot score now follows the session, meaning a reset doesn't actually reset anything. Tools like proxy rotation, header fuzzing, and pacing were all built for a world where you were judged one request at a time. The game has changed to five-minute behavioral coherence. p.s. Just the other day, I spent 5+ minutes on convincing the bot I'm not a bot... and ultimately failed ;]
Do you think as a result of this we will rely more on browser based scraping and we will see an increase in costs for scraping?
If you're still building scrapers, is your stack stuck thinking in single requests? Digging into the breakdown of how session fingerprinting changes things here: https://foura.ai/blog/cloudflare-precursor-session-fingerprint