Idea the First

Google has the technology, the resources, and the name-recognition to make the the Web work the way it was supposed to.

I refer, of course, to two-way linking. Right now, linking is a one-way affair. As I wrote four years ago, “The web consists of a huge number of people shouting and pointing at each other.” When I make a link from my page to yours, I’ve created a relationship between those pages. The problem is that there’s no good way for you to reciprocate and deepen that relationship, and no way for me to keep the relationship flourishing even if you neglect it.

Trackback, I have asserted, is a major step towards getting this stuff right, because it leaves a trace of the link on the target end. More generally, the rise of good programmatic APIs and of standardized formats for pushing structured web data around make it more feasible for cooperating sites to have a genuinely dynamic back-and-forth of content. But progress has been oddly stunted. As exciting as the uptake in browser-driven standards-based dynamic-data web applications has been, I feel like the big leap in the linking structure of the web itself has yet to take place.

Google could make it happen, though.

Step one would be a dead-link repair button in the Google toolbar and Google Web Accelerator. If I click on a link that takes me to a 404, a site not found, or some other page that isn’t what I though the link promised to deliver, I punch the “repair” button and ask Google to fix things for me. Google looks at the page I came from and the link I followed, and then looks into its caches to see when that link was added to the page. It then pings the target URL in its cache, as of that date, and serves me up the page that the link’s author was linking to back when the link was live. Bam. Far fewer dead links.

Step two is step one, automated. Collaboratively filter user requests for prior, working versions—and offer them to me with a “did you mean” notice when the secret-sauce AI sees that other people have stumbled over the same dead link. And while you’re at it, ping the link’s author; she may want to update the link. Indeed, ping the target, too—she might very well want to fix up the link, or to confirm that what Google thinks the new target ought to be really is the correct target.

Step three is when Google combines this feature with the inbound-links search feature it already offers. A little good visualization, a good interface for reporting URL changes … and now I can change my site around without breaking inbound links. The idea is to become a protocol-standardizing middleman who helps site authors and web browsers coordinate with each other. People have never been able to make URIs invariant the way that they were supposed to be; perhaps the solution is to give up on the invariance and to figure out how to repair the damage caused by the inevitable changes.

Do all that, and the web will be, well, better.