3 ms·
We need strong association of content with their creators. For example by using digital signatures. It's not hard. We've had this technology for ages. Yet the v
by jillesvangurp 4y ago
We need strong association of content with their creators. For example by using digital signatures. It's not hard. We've had this technology for ages. Yet the vast majority of content on the web is unsigned (with the exception of ssl certificates that merely proves content came from facebook.com or wherever instead of from a particular person on facebook). Bots, troll farms, spammers, scammers, etc. use this to suggest they are more reputable than they really are. Users can't tell the difference.
With verified creators signing their content, there is zero confusion about what source the content came from. That source might still use AI to produce content of course. Or it might be an AI.
The web might fill up with content from all sorts of sources but the only content you should care about should come from sources that are reputable as evidenced by their body of signed work over time. Doesn't matter if it's a bot or a human. Reputation is hard to fake for a bot. And people, AIs moderating content can just flag content sources by their keys. So now you can do things like figuring out what the reputation of a source is relative to other sources you trust.
Not that hard technically. We've had public/private key signatures for ages. Never caught on for email. Some chat networks use end to end encryption. But most public information on the web is effectively unsigned.
- GalenErso 4y agoI don't see how that would be effective without fundamentally changing the structure of the Internet. For example, I have access to your HN comment history. I could easily start my own blog with insights Ctrl+C Ctrl+P'd from HN'ers comments histories, and sign it as if it were my own. Unless we ditch graphical user interfaces and the HTTP/S protocol and revert to 80s computing, with a command line interface for everything.
- __MatrixMan__ 4y ago> Unless we ditch graphical user interfaces and the HTTP/S protocol and revert to 80s computing, with a command line interface for everything. I think we should do precisely that. SSL is all about explicitly trusting server names and implicitly trusting the data they serve. The whiz-bang UI's of the modern web are predicated on blindly running whatever code those servers give you. That's why we have all of these asinine trainings on how not to click the malicious link. It's time we started explicitly trusting people and implicitly trusting the data that they sign and let which server we're talking to fade into an implementation detail. If we trust the data it served for other reasons, it doesn't matter if we trust the server. We can just ignore whatever malware showed up because it's not signed by someone we trust. Besides, imagine what we could get done if we didn't have to stop to rebuild the UI for maximal engagement every few months. We could, I dunno, compete on merit.