In 1998, if a search engine offered results that were related to the query and were somewhat useful, that was a success - and perhaps even a revolutionary win for the search engine. AI has changed the world since then in two ways. First, these kinds of results are no longer needed. LLMs do a fine job answering simple questions with generic answers that, for the most part, suffice. Generic results that don't have really original ideas or extraordinarily high quality writing are somewhat superfluous. Second, the results that are displayed today are often themselves generated by LLMs in order for publishers to monetize cheaply produced content. In a world where pages cost nothing to produce, Page Rank makes no sense. It’s 2025, and we are so used to getting low quality results that we celebrate when search results are passable and aren’t blogspam. The aim that we’re proposing is both simple and revolutionary: High quality search results only. There are three different components to this: 1. Relevance 2. Quality 3. Personalization. People Rank solves all three. Relevance is special because it doesn’t look at the quality on its own - it looks at the content as it relates to the search query. Quality is an understanding of how valuable the content on the page is in terms of originality, clarity, and depth. Personalization is how likely it is that a particular user will enjoy the content. Each of these measures is different, and we aim to understand all three without looking at the content of the page at all. The only data we’re interested in is how people who have viewed the content enjoy it. How we go about discovering what people enjoy is a bit counterintuitive as well. Traditionally, search engines spy on users to do “behavior analysis” - looking to see how much time they spend on pages. As you might imagine, this is behavior that’s easy to game (write a suspenseful cliffhanger at the beginning, and then change the topic and go on and on with some irrelevant ads). Searching for recipes online is largely unusable because of this. This approach is also quite disrespectful of users' privacy since most people aren’t aware of how much data Google has on them. People Rank is interested in which pages people express a desire to stash for themselves (stashing is like bookmarking - where people save stuff for later - but it allows people to find stuff with AI so that there isn’t a need for folders). It’s an altruistic form of selfishness since while it does benefit the user themself, primarily, this benefits the community since there’s a strong signal of quality. Of course, we care about who stashes which pages. For example, if a farmer stashes something about gardening, that would connote value, whereas if a computer scientist did, it might not. Understanding this graph, and how each user and each page fit into it, is what we call People Rank.