Google has launched a web accelerator services (aka, a distributed caching, compressing proxy services). It seems to use a browser plug-in. I haven’t seen any dissection of whats going over the network, but I assume its adding some id tracking info along.

Which is kind of scary. Not so much because of the obvious privacy concerns (though, thats probably something to keep in mind), but because I think it might be closing the last chance for a competitor to compete with google. Why?

It seems the obvious way to gather a huge amount of interesting information about the web, and more importantly, how people use it. With a intelligent proxy setup, it’s very easy to track peoples paths though a web site, which link on a web page people actually follow, etc. And for google, thats very useful information to use when trying to return useful search results. And if nothing else, it gets them an easy way to see what links off of a google search result page people actually follow.

Since it’s not exactly easy to setup a large distributed proxy caching network, thats a pretty strong entry barrier to competitors. Not to mention it’s all but impossible for anyone to get a significant number of users to start using a system like that. OS and browser vendors and maybe AOL and google could do it. So something to possibly be worried about.

But theres also alot of interesting things you could do with it. Both for the proxy user and provider.

Throw in a browser plug-in, and you can easily add web site citation handling. Or Tivo style thumbs up/thumbs down. It’s a great place to plug-in a language translation layer. Or inserting “smarttag” style additional information. Or adding “whats related” info. And with a browser plug-in (or *cough*, a browser) you wouldn’t need to change the pages themselves like “smarttags”.

Maybe add in some bookmark sharing/syncing features for even more web meta info.

For the proxy user, you can provide a fair number of interesting value adds. Akamai style caching of large content. Pre-fetching of web sites based on likely paths though a web site. Site path based searching (find “cheese” in the pages people follow from this page).