jdq

writer and composer

I have come back to you as I left: a fool.


ckolderup
@ckolderup

[now mirrored on my actual blog: https://motd.co/2023/05/archive-stumbler/]

I like to browse the Internet Archive late at night. A lot of the internet is really boring now but the Archive is super refreshing but it lacks a key discovery feature: serendipity.

the archive.org collection listing sort bar, featuring the ability to sort by: views, title, date reviewed, creator

These are the only options you have in how you sort a collection when viewing it, so if you're faced with, say, the full archive of Western Hills Access Television because you want to know what was going on in Western Maine 1-4 years ago, the only way to embrace any kind of serendipity is to kinda scroll for a little while while unfocusing your eyes and then click somewhere on the page. Very unsatisfying!

Luckily the Archive can return data on a few of its main endpoints in JSON format, so you can chain a few HTTP calls together and make it possible to make a computer use a PRNG to pick for you. So I did!

screenshot of the interface for Archive Stumbler

Archive Stumbler will accept any Internet Archive collection URL and provide a hyperlink that you can click on (or, if you're settling in for some real crate digging, ctrl-click a bunch to open in new tabs) to send you to a random item in that collection. If you don't know where to start, there's a button you can click that'll pre-fill the field with some sample collections.

Hope you enjoy! It's pretty slapdash, so if you do anything wrong it will probably fall over in bad and confusing ways. EDIT: I think I know what the main failure is and might try to put a workaround in place! In the meantime, if you keep refreshing the tab it'll keep trying again with the same collection.

The one enhancement I'll probably add to it soon is encoding the value of the URL input box into the page URL so that you can make bookmarks/share links that will pre-fill with a specific collection. EDIT: I did do this one! The URL should auto-update any time you have a valid a.o collection URL in the box, you can copy that and share it with someone or make a bookmark/shortcut/etc.


You must log in to comment.

in reply to @ckolderup's post:

I don't know that there's a total lack of serendipity in browsing the Internet Archive. The crux is you're not likely to find it in the search bar. To give you an idea of what I mean, I went to check out a Cutie Honey: The Live archive last night, and found out through the related archives at the bottom of the page that the 1973 Cutie Honey is also archived.

I know it's not a glamorous discovery, but I have found shit through this feature I didn't otherwise think I would find.

I mean, sure, serendipity is a broad concept. Some would say that the top item of a sorted-by-title collection being actually of interest to you is serendipity, or that noticing a given uploader and browsing all their uploads only to find that you actually share two niche interests and the OTHER one is a goldmine is great. And it is! It's why I love digging through that site. I just think more websites that offer listings of large collections of things should embrace the beauty of selecting an index at random

I like this! I remember way back someone coded a way to randomly access something from everything technically publicly accessible in JSTOR and that was fascinating. I think JSTOR shut that down pretty quickly, if I remember right due to them being far less cool than IA is

thanks! I actually asked Jason Scott whether this already existed as some kind of undocumented URL param or API endpoint or something and got a very in-character "make it yourself" from him, so I assume it'll be tolerated at the very least!

this is cool! i'd be interested to see if even more randomness could be injected into the process - maybe by searching for items/collections using random words, or trawling the most recently updated stuff

that'd be a cool way to expand on this for sure-- my use of the search endpoints here is pretty naive, mostly just for collection traversal. At the very least I think I'm going to go through this weekend and add like 100 collections to the autosuggest button to make it a little more useful to people who DON'T already have a bookmarks folder full of a.o links like me!