Keith Devens .com

Backlinks

I'm about to implement a very experimental version of backlinking on my website. I've gotten code that I think will be accurate enough. I haven't been happy with any previous implementation. All other implementations I've seen would do one or more of the following:

• "Let in" root links. In other words, if "root" links are let in, rather than having http://www.keithdevens.com/weblog/?id2215 be the backlink, just http://www.keithdevens.com/ or http://www.keithdevens.com/weblog/ might be the backlink, depending on where the referrer came from. This isn't helpful at all long term, because links typically scroll off the main page of a weblog, for instance.
• Backlinks "expire". Mark Pilgrim's backlinks expire, Chris Langreiter's expire, etc. I don't like this either. Backlinks should be more permanent than transient. They should only be as transient as pages on the web are, and because pages on the web sometimes are transient, you will need to make a backlink expire occasionally if either the page moves or a link to you isn't on that page anymore for whatever reason.
• Take only a limited number of backlinks. Disenchanted.com does this. Backlinks should let everyone in on the discussion. Not only the most recent X number or the most popular X number of backlinks.
• Have to be manually approved. No way I'm manually approving backlinks, they should be automatic.

I think I've been able to get around all of these problems. Once the code is implemented on my site I'll explain more and expand on the commentary here.

Also, there are two other options to consider. Should backlinks show up in the order in which they hit your site, or should backlinks show up in order of popularity (those that sent the most hits to your site show up first)? And how the heck did Xanadu implement backlinking?

Yikes, my whole scheme pretty much got shot to hell when I realized that some blogs have their permalinks above the post instead of below. Phooey on me.

You know what? You just wait. My goal isn't 100% accuracy, it's 90-95%. I'll get it yet, I have some ideas. Basically, I have to put a lot of intelligence into this thing, but it'll work, you'll see. I have in the back of my mind a few cases where there's no way to make it work in the general case, but hopefully that only makes up the 5 percent or so.

Permalink | Comments? | id 2215

02:03 (utc-4)
edited: 23:00 (utc-4)