Memento: Adding time capabilities to the Web

This summer I'm working on a research project adding Memento support to the Android platform. I'll talk more about my project at a later date, but first I want to provide a quick overview of Memento.

Memento is an architecture which allows a web browser to seamlessly access older versions of web pages. It allows you to "time-travel" on the Web.

The best way to explain this is with an example. If you were to access cnn.com, you would be presented with today's version of the page. But what if you wanted to see how it looked one year ago? You would need to visit the Internet Archive's Wayback Machine to find a list of old copies of the page they had archived, and you would need to click on one of the links. And if IA didn't have the page archived, you would have to search other web archives for the archived version. This is potentially a lot of work.

Memento makes access to archived versions of a page transparent to the user. Using a web browser that supports Memento, you would only need to visit the URL as you normally would and supply a desired date... the browser would automatically locate the archived page from that date and display it to you without the need to manually search through multiple archives.

You can see this in action right now by using the Memento Firefox add-on. Below is a screen shot using the add-on to browse cnn.com as it appeared on July 9, 2009. I actually told the add-on to show the June 21, 2009, version, but the July page (from the European National Archives) is the closest page that was found in any archive. This is not a failing of Memento... it's a limitation of web archiving in general.


Memento uses HTTP content negotiation to add this time dimension to the Web. Instead of discussing the technical details here, I'll instead point you to the Memento Guide Intro if you're interested. Ideally, all web browsers and web servers in the future will support the Memento HTTP headers, and no special add-on will be necessary.

Memento is the brain-child of Michael Nelson (Old Dominion Univ) and Herbert Van de Sompel (LANL). It's made a quite a stir in the past year with a write-up in New Scientist, a paper at the Linked Data on the Web workshop (LDOW2010), and some significant funding from the Library of Congress. Tim Berners-Lee said this about Memento: "This is neat; there is a real need for this."