Client or server-side traffic trackers?

The argument for using client-side traffic trackers have always been that they are less cost expensive then server-side once. But why should the bandwidth load be on the user and not the content provider?

Client-side traffic trackers usually requires the user to send a request and download a JavaScript or an image from a third party provider. This will in some cases cause user's Web browsers to notify them that a third party provider might be listening to the connection.

Server-side traffic trackers can get all the same data as client-side trackers. The disadvantage is that the local server will have to process the data it self. However the Webpage will with most system load much faster than if it had to pull in an external script.

I have experimented a lot with different client-side trackers — Google Analytics, Sitemeter, and others — and have always concluded with that the reduced page loading speed, and reduced visitor's privacy is not worth it. Right now all my Websites are using server-side tracking only, something which offers the visitors better privacy as the information is not shared with a third party provider which makes it's profits on building personalised consumer profiles of your visitors.

Though I shall admit that FeedBurner's new client-side tracker is pretty good and loads almost instantaneously.