Solving CP and spam in anonymous discussion boards
We asked 2 questions:
>1. How do we stop spam without JavaScript and adding friction to the anonymous posting process
>2. How do we stop CP spam without relying on an external dependency (like a big corporation's API)?
I believe I have conquered the top problems that have plagued anonymous discussion boards since their inception.
I solved the (Child) porn problem by deploying a lightweight vision model that routinely scans all newly uploaded attachments and rates them on a NSFW score from 0 to 1.0.
If the NSFW score is equal to or above 0.80, the attachment is flagged and blocked. Moderators are able to manually mark attachments as exempt if the model made a mistake.
This effectively prevents bad actors from posting child porn and immediately reporting it to law enforcement/hosting providers while you or your moderators are asleep. You can argue in court or to your hosting provider that you have taken steps to fight child porn distribution on your site by employing a system like this.
I solved the spam problem by implementing several layers of dynamic rate limits. Most websites rely on IP bans, CAPTCHAs, and JavaScript to keep spammers out. That adds friction for real users and doesn't work against distributed botnets. Instead, I have implemented a system that would let anonymous posting stay friction‑free 99% of the time, but when a flood of garbage appears, it slows to a trickle automatically.
Are there other websites that have implemented anything like this? This only took me 3-4 days to implement and I'm confident in its effectiveness.
>1. How do we stop spam without JavaScript and adding friction to the anonymous posting process
>2. How do we stop CP spam without relying on an external dependency (like a big corporation's API)?
I believe I have conquered the top problems that have plagued anonymous discussion boards since their inception.
I solved the (Child) porn problem by deploying a lightweight vision model that routinely scans all newly uploaded attachments and rates them on a NSFW score from 0 to 1.0.
If the NSFW score is equal to or above 0.80, the attachment is flagged and blocked. Moderators are able to manually mark attachments as exempt if the model made a mistake.
This effectively prevents bad actors from posting child porn and immediately reporting it to law enforcement/hosting providers while you or your moderators are asleep. You can argue in court or to your hosting provider that you have taken steps to fight child porn distribution on your site by employing a system like this.
I solved the spam problem by implementing several layers of dynamic rate limits. Most websites rely on IP bans, CAPTCHAs, and JavaScript to keep spammers out. That adds friction for real users and doesn't work against distributed botnets. Instead, I have implemented a system that would let anonymous posting stay friction‑free 99% of the time, but when a flood of garbage appears, it slows to a trickle automatically.
Are there other websites that have implemented anything like this? This only took me 3-4 days to implement and I'm confident in its effectiveness.