When I first started learning Javascript, in the Netscape Navigator version 3 days, all the programming for a page was done in the web browser. Javascript could read the query parameters to a page and construct the page with information according to the request. In those days, I did everything in the browser because I couldn’t afford my own web server, and the free servers of those days (Geocities, Tripod) only really supported static pages.
Finally I got access to a server that ran PHP, which was awesome because I could do all the programming on the web server, reducing the frustration of dealing with various versions of web browsers that supported different feature sets.
But recently, web browsers conform to standards better, support a richer feature set, and are supported by many Javascript client libraries. I’m finding that more and more, I’m ditching the server side program in favor of more client side programming. This is especially true with my embedded system work, where offloading the User Interface to the web browser provides a better experience.