It is both a great source of information and a gargantuan pile of shit.
The information part is obvious with things like achive.org, project gutenberg, and various other sources of documents. And things like OCW and non-political articles on kikepedia.
The gargantuan pile of shit concerns both the content and the usability of which the content speaks for itself.
The usability aspect is most obvious when looking for simple things. For example
>At which time does store X in my town Y closes today?
You either go to google, join the botnet and hope that info is accurate, (and wade through countless javashit and images/maps in the process)
Or you go to the website of the store (likely a chain of stores)
>Go to X.com
>load piles of javashit and unneeded data for your issue
>pick a store
>Have to enable jewgle maps of other 3rd party map framework
>zoom in to town Y
>click store
>If lucky it shows the hours, else a few more clicks are required.
In a saner world you would solve that by typing something like
>Shop X in Y list hours
Which would take a fraction of effort, time, data and code to accomplish.