The problem is lack of semantics. Computers don't understand the content of a web page or article, and you have to manually parse it and make sense of it. You'll find 80% of what you need in 20% of relevant pages, and the remaining 20% in the other 80% of pages.
The semantic web is the solution. A system that understands content, understands what you know, and understands what you want. Then, all it has to do is show you what's relevant and necessary to the task you're trying to accomplish. In most cases, you won't have to actually do the task, as it will be automated.
I don't see any way around this. Changing how search engines work will still not let you find articles that complement each-other perfectly. And you'll still have to read more than necessary.
Yes, the problem won't be solved completely and you will still end up reading the same content more than once. But at least, by analysing the text the search engine could detect which articles are coming from the same source (e.g. inspired by the same press release or just telling the same news event over and over again) and group them accordingly. Currently, I don't feel like Google or other search engines are doing any of that (maybe they are, but if so, the results are not good enough).
The semantic web is the solution. A system that understands content, understands what you know, and understands what you want. Then, all it has to do is show you what's relevant and necessary to the task you're trying to accomplish. In most cases, you won't have to actually do the task, as it will be automated.
I don't see any way around this. Changing how search engines work will still not let you find articles that complement each-other perfectly. And you'll still have to read more than necessary.