Link Analysis is Dead

Link analysis is dead. Okay - well maybe it is not dead, but it's effectiveness is definitely on the decline. A significant percentage of the blogs that I have browsed on blogger.com are obviously there to influence search engines because they are the same text and links repeated over and over again. These pages were obviously not intended to ever be viewed by a human.

Can the big three search engines effectively filter these out?

AJAX Sucks!

The latest buzz in web application development is AJAX, which stand for asynchronous JavaScript and XML (or something like that). AJAX can help make web applications much more interactive and much more use friendly by not forcing web pages to reload for every user interaction. Unfortunately, AJAX sucks...here's why...

AJAX is still the same old browser technology we have had around for a number of years, with the addition of the XMLHttpRequest object. The problem is when it comes to good old JavaScript and DOM, browsers are still very different and very quirky. You can write code to handle the differences in the browsers, but you end up putting in lots of 'IF' statements to do the same thing in a slightly different way. This takes a lot of experience, a lot of trial and error, and a lot of patience.

Having to write browser dependent code is not even the worst part. The worst part is that all of this is very difficult to test. I don't know of any automated testing tool that would allow you to test Internet Explorer, Mozilla, Firefox, Opera, and Safari. What usually ends up happening is you start abandoning the less popular browsers. Opera and Safari are usually the first to go, then Mozilla and Firefox. Those development teams that do support multiple browsers end up spending a lot of time and money either on multiple test tools or on additional testers.

Don't get me wrong - I really like the latest features that Yahoo and Google have added to their email and other services. I also think we still have a long way to go to make it simple to write robust web applications that can run on anywhere.

Google AutoLink

There has been a lot of commentary regarding the AutoLink feature of the latest Google toolbar. Some love it and some hate it. Here is some thoughtful commentary from Tim Bray's Ongoing.

I can understand both sides of the argument. Perhaps there is a happy medium. Let me propose 2 solutions...

1. The Google toolbar, instead of modifying the actual links in the page could provide links to AutoLink-able items in a sidebar or some other location outside of the actual web page. This way there should be much less confusion about where the links are coming from. This would still give Google the ability to derive revenue by driving traffic through their 'helper' links. It also could not be a conflict of interest because Google would not be modifying any web pages, but only supplying additional links which the end user can easily distinguish.

2. Use helper application protocols (like mailto). We already have several of these that are handled by the web browser and are very common. It allows a link to refer to an email address and a browser plug-in can be configured to handle the 'mailto' protocol so that end users can choose which service should handle that particular protocol. We could have a 'map' protocol for addresses, a 'book' protocol which could support both ISBN numbers or searching by title, author, etc. Perhaps page author could even be given the ability to provide a default protocol handler or URL for the case there a browser does not have a handler already configured. Google could then allow users to choose optional protocol handlers during install of their toolbar. This would allow Google to derive revenue by driving users to their services, while also removing the conflict of interest.