here's an old mail of mine to -devel with additional context:

Just a heads up to this technical project which I've just discovered
(via their integration with Internet Archive):

http://timetravel.mementoweb.org/
http://timetravel.mementoweb.org/guide/api/
https://en.wikipedia.org/wiki/Memento_Project

They have added a new request header Accept-Datetime to HTTP content
negotiation (and described it in RFC 7089), allowing clients to ask to a
specific website a version of the website content as it was in the past.

So for instance with URL like (or the equivalent HTTP request to GitHub
directly, by setting the appropriate request header):

http://timetravel.mementoweb.org/memento/20150101/https://github.com/zacchiro/debsources

you will ask GitHub not to show you
https://github.com/zacchiro/debsources as it looks now, but rather how
it looked in January 1st, 2015. The specific implementation of what that
means is left to the content provider, and GitHub has implemented it
pointing to URLs like:

https://github.com/zacchiro/debsources/tree/e78ae3ba7583c42cbf9542b838760c882573cd88

i.e., the Git tree that was pointed to by master at the time.

This does not add much, in fact nothing at all, in terms of preservation
because if, say, GitHub disappears from the net, they will no longer be
able to resolve. Also, content providers might have all reasons to
retroactively delete old content copies.

But it is interesting technology, and also allows to have proxies for
the content resolution (e.g., Internet Archive itself) in case website
disappears all together. Assuming those proxies did crawl old copies of
some website.

I'm mentioning it here in case people could think of synergies between
SWH and Memento.