Comment by johng
5 hours ago
I don't know how this works under the hood but it seems like no matter how it works, it could be gamed quite easily.
5 hours ago
I don't know how this works under the hood but it seems like no matter how it works, it could be gamed quite easily.
True, but there's probably many ways to do this and unless AI content starts falsifying tons of its metadata (which I'm sure would have other consequences), there's definitely a way.
Plus other sites that link to the content could also give away it's date of creation, which is out of the control of the AI content.
I have heard of a forum (I believe it was Physics Forums) which was very popular in the older days of the internet where some of the older posts were actually edited so that they were completely rewritten with new content. I forget what the reasoning behind it was, but it did feel shady and unethical. If I remember correctly, the impetus behind it was that the website probably went under new ownership and the new owners felt that it was okay to take over the accounts of people who hadn't logged on in several years and to completely rewrite the content of their posts.
I believe I learned about it through HN, and it was this blog post: https://hallofdreams.org/posts/physicsforums/
It kind of reminds me of why some people really covet older accounts when they are trying to do a social engineering attack.
> website probably went under new ownership
According to the article, it was the founder himself who was doing this.
If it's just using Google search "before <x date>" filtering I don't think there's a way to game it... but I guess that depends on whether Google uses the date that it indexed a page versus the date that a page itself declares.
Date displayed in Google Search results is often the self-described date from the document itself. Take a look at this "FOIA + before Jan 1, 1990" search: https://www.google.com/search?q=foia&tbs=cdr:1,cd_max:1/1/19...
None of these documents were actually published on the web by then, incl., a Watergate PDF bearing date of Nov 21, 1974 - almost 20 years before PDF format got released. Of course, WWW itself started in 1991.
Google Search's date filter is useful for finding documents about historical topics, but unreliable for proving when information actually became publicly available online.
Are you sure it works the same way for documents that Google indexed at the time of publication? (Because obviously for things that existed before Google, they had to accept the publication date at face value).
1 reply →
"Gamed quite easily" seems like a stretch, given that the target is definitionally not moving. The search engine is fundamentally searching an immutable dataset that "just" needs to be cleaned.
How? They have an index from a previous date and nothing new will be allowed since that date? A whole copy of the internet? I don't think so.... I'm guessing, like others, it's based on the date the user/website/blog lists in the post. Which they can change at any time.
Yes they do. It's called common crawl, and is available from your chosen hyperscaler vendor.