LinkBack: adding a dimension to the web

LinkBack is a browser extension for Chrome that will automatically display incoming links for any web page you visit. Incoming links show up in an overlay window. For example, here's the Wikipedia entry for Hypertext with LinkBack operating in the lower-right corner.

Sorry, but for now LinkBack only works on Chrome.
Argh, Chrome has changed their policies so extensions can now only be installed from their store, or from the Chrome extensions page. So for now to install Linkback:
  1. Click on the button above, which will download the extension file (linback.crx) and make it visible in the bottom bar
  2. From the Chrome tools menu, go to the Tools > Extensions page, and drag the file from the bottom bar into the page.

Use

Just browse. The popup will appear if and when any incoming links are found. If the popup is intrusive it can be minimized with the handle at the lower-right. To disable it entirely use the Chrome extensions page.

Note that LinkBack makes use of a free service from seomoz.org which is rate-limited, so if any large number of people start using this it will slow down. If that becomes a problem, some other source of data may be needed.

LinkBack is turned off for a few websites where it is generally a nuisance, such as Facebook and Google.

Background: The Bidirectional Web

Ted Nelson coined the term Hypertext over 40 years ago, and articulated its principles, most of which have been ignored in the triumph of the Web (see below). One of these ignored ideas is the inherent bidirectionality of hyperlinks. This just means that any hypertext link from A to B should be visible and traversable from B to A as well. A simple idea, but suprisingly difficult to implement if you start with the sort of document-centric model used by the web.

Another related Nelsonian principle is that hyperlinks are inherently annotations to documents rather than (or in addition to) embedded parts of documents. In this model, it is easy and natural for a third party to be able to add commentary on somebody else's text in the form of links. This idea too is entirely absent from the web, and difficult to graft on.

There have been efforts over the years to add the ability to have backlinks and annotation to the web (Third Voice, CritLink, Annotea, and more recently Marginize). I myself briefly tried to get a web startup going sometime around 2002 that would have done something similar. Other related efforts are emerging as part of blogging software and protocols, but these have some flaws: they are limited in scope, and require the cooperation of the owner of the source document.

LinkBack is an attempt to do a little of this on the cheap. Of course, this isn't really bidirectional linking in the classic sense. Links are still one-way, all this does is make links that go either to or from a page visible, whereas before all you saw were the outgoing links. Still, it's better than nothing; it adds an interesting new dimension to web browsing, while offering a glimpse of where we could be someday.

Worse is better

The lame version of hypertext we are saddled with is merely one instance of the general "worse is better" principle, which essentially says that doing the wrong thing in a way that lowers adoption barriers will beat doing the right thing every time. See the triumph of C over Lisp for another example.

It is extremely unlikely that the right thing for hypertext will ever happen. We are stuck with simple but wrong model that took over the world. But: the computational infrastructure available to us (search engines, APIs, client-side addons like this one, microformats and emerging standards for semantics) are powerful enough to slowly and incrementally move us toward better approximations. That is better than nothing, given that we have to live in the real world.

Credits

Author: Mike Travers

Powered by:

Source and issue tracker on github.

Caveats

The extension sends information about the pages you visit to the backend service provider. Every page you browse generates a request to the LinkBack server on Heroku which in turn queries an API at Seomoz.org. The former does not keep logs, and the latter does not get enough information to identify you, but you should still be aware of this.

Also, do you really need to see more hyperlinks during your browsing? The web as it is isn't distracting enough?