Arbital: Google Maps for knowledge

While Ar­bital can do many things, cur­rently it’s most geared to­wards pro­vid­ing the reader with a tai­lored ex­pla­na­tion on any given sub­ject. Ar­bital can find out what you know by ask­ing a few ques­tions, and then dy­nam­i­cally gen­er­ate a tai­lored se­quence of pages for you to read. Take a look at Bayes’ rule: Guide to see how that works.

Ad­van­tages of Ar­bital’s al­gorithm:

  • Ar­bital can cut out whole chunks of ex­pla­na­tion if the reader already knows it, sav­ing them lots of time.

  • Ar­bital helps you learn piece­meal. Most pages are pretty short and teach you one or two things, and each page brings you a step closer to learn­ing what you are af­ter.

  • If there are mul­ti­ple pages ex­plain­ing the same con­cept Ar­bital can pick the best one us­ing a mul­ti­tude of sig­nals: page com­plex­ity, pop­u­lar­ity, and au­thor speci­fied meta­data.

  • Ar­bital dy­nam­i­cally mod­ifies each page to tai­lor it to the reader’s cur­rent knowl­edge and prefer­ences. For ex­am­ple, if you don’t read logic, the al­gorithm can re­place the math for­mu­las with text.

  • There is a sin­gle URL to learn a sub­ject that works for ev­ery­one, even peo­ple with vastly differ­ent lev­els of prior knowl­edge.

A few ex­am­ples:

  • If you want to learn Bayes’s The­o­rem, Ar­bital can find a tai­lored ex­pla­na­tion for you based on how good you are at math.

  • If you are learn­ing a new pro­gram­ming lan­guage, it will lev­er­age your ex­ist­ing knowl­edge pro­gram­ming in gen­eral and in­stead of teach­ing you func­tions from scratch, it’ll skip di­rectly to teach­ing you func­tion syn­tax for the par­tic­u­lar lan­guage.

  • If you are learn­ing about AI safety re­search, it will show you the tech­ni­cal de­tails if you can read math no­ta­tions.

  • If you want to un­der­stand a cer­tain his­tor­i­cal or mod­ern event, it can figure out what pre­vi­ous and re­lated events and figures you need to know about, so you have the full con­text.

Tools for authors

Every­thing you see on any Ar­bital page was cre­ated by its au­thor(s) us­ing Ar­bital’s mark­down syn­tax. There is no page-spe­cific code any­where. The magic in­gre­di­ents are Ar­bital’s net­work of req­ui­sites and the re­la­tion­ships that the au­thors cre­ate. Each au­thor can lev­er­age all the ex­ist­ing pages in the Ar­bital sys­tem and their ex­ist­ing req­ui­site con­nec­tions. For ex­am­ple, if you cre­ate a blog post about “So­cial me­dia graph anal­y­sis”, you can set “Graph the­ory” as a pre­req­ui­site. A reader who wants to un­der­stand your blog post can eas­ily go through the se­quence teach­ing them “Graph the­ory” af­ter which they’ll be ready to read your post.


  • More about Arbital

    Lots more in­for­ma­tion about Ar­bital vi­sion.

    • Arbital

      Ar­bital is the place for crowd­sourced, in­tu­itive math ex­pla­na­tions.