3 ms·

This sounds much more expensive than just serving the cached content you already have. I would hate for a future where I have to double-take the content I'm re

by czk 2y ago

This sounds much more expensive than just serving the cached content you already have. I would hate for a future where I have to double-take the content I'm reading to make sure CloudFlare didn't decide to confuse me with a bot and feed me AI generated slop. The war on crawlers thing is noble in some sense, but at the end of the day I wonder if this just hurts the regular users more than anyone else. Gatekeeping, captchas, false-flagging real people as bots, and getting to decide who gets to see what content on the internet. OpenAI can afford to work around this with their billions of dollars but the average guy will be fed AI generated slop from CF when he tries to script a requests one-liner in python.

theamk 2y ago
The AI slop is definitely less expensive for the author, as all the efforts are expended by CF's infrastructure, while the host sees zero requests. In the previous discussions, one of the big AI problems was that they follow every link [0], including expensive ones like "git blame" view.. as well as "every commit of every log", which can be easily in millions if you have mirrors of few repos with many branches. It's simply impossible to cache this all at CloudFlare, and solution like AI maze (or outright blocking) seem like the only way. I feel you about "requests one-liners in python" - I am still running a few of those in crontab on some server.. but their time might be past. The "average guy" does not care about this kind of scripting, so we will get lumped together with evil AI and blocked. [0] https://drewdevault.com/2025/03/17/2025-03-17-Stop-externalizing-your-costs-on-me.html https://drewdevault.com/2025/03/17/2025-03-17-Stop-externali...
- czk 2y ago
  Yeah you are right about the caching thing, I understand what CF is doing I just cant help but feel like there is something deeply wrong about this approach. But I don't have any better ideas to propose. Maybe I'm just holding on too tight to the 'old web' that I've known and loved.

by czk 2y ago

theamk 2y ago

The AI slop is definitely less expensive for the author, as all the efforts are expended by CF's infrastructure, while the host sees zero requests. In the previous discussions, one of the big AI problems was that they follow every link [0], including expensive ones like "git blame" view.. as well as "every commit of every log", which can be easily in millions if you have mirrors of few repos with many branches. It's simply impossible to cache this all at CloudFlare, and solution like AI maze (or outright blocking) seem like the only way. I feel you about "requests one-liners in python" - I am still running a few of those in crontab on some server.. but their time might be past. The "average guy" does not care about this kind of scripting, so we will get lumped together with evil AI and blocked. [0] https://drewdevault.com/2025/03/17/2025-03-17-Stop-externalizing-your-costs-on-me.html https://drewdevault.com/2025/03/17/2025-03-17-Stop-externali...

czk 2y ago
Yeah you are right about the caching thing, I understand what CF is doing I just cant help but feel like there is something deeply wrong about this approach. But I don't have any better ideas to propose. Maybe I'm just holding on too tight to the 'old web' that I've known and loved.