I find myself dreaming about "replace Wikipedia" again.
----
What does it take to replace Wikipedia?
- A base of 50k "decent" encyclopedia articles. ChatGPT or a dozen other LLMs can do this.
- A way to generate articles on current events. How will the machine know things? You collect primary sources, and write the article based on them. ⚙️ a "primary" source is a source with direct connections to an event. Despite the protests of many Wikipedia editors, contemporary news articles are generally primary sources. A "secondary" source analyzes the primary sources, with the benefit of hindsight. A "tertiary" source is an encyclopedia or the like, built on secondary sources. 💡 the need to prefer secondary sources on Wikipedia is for multiple reasons; one of them is that editors can easily mis-interpret primary sources. hopefully the machine can avoid this problem 💡 the problems of "who creates the primary sources we use" are more difficult
- A way to encourage people to use your site and not others. 💡 "advertising spend" can do this. Also, appealing to a vertical (or clique) that dislikes Wikipedia can generate word-of-mouth marketing 🔥 the cynical approach would be to have a dozen different encyclopedias, targeting a dozen different cliques
- A way to generate revenue. ⚙️ Wikipedia, famously, has no ads [other than the annual donation-drive banners]
- A way to remain up to date.
----
The new encyclopedia would use hypertext. Maybe. 🔥 my experience has demonstrated that most people are less enamored by hypertext than I am