Forgive me for saying so, but the way we write scientific papers today is hardly consistent with our level of technological sophistication. A lot seems to be done "just because people always did this way". Tradition weights far too heavily on the Science Halls, and that leaves little space for true innovation. 

One problem that always plagued me is reading scientific papers that are riddled with citation acknowledgements. It is archaic, and your eyes need to skip text all the time. Plus it takes up valuable space, and whomever had to shorten that paper to fit the conference rules know what I'm talking about. Who reads hard copies of scientific papers today, anyway? 

The idea is to create a Scientific Text Editor, one that is specific tailored to view and edit scientific papers. A special format of file would be designed to hold both the textual content, the references, the images, tables, mathematical equations, and the citation acknowledgements. Get the best tools available to handle each format of data and place them working together. 

This text editor, would work best as a website service rather than a desktop application, but I'm not entirely sure. Personally, nothing in this world would make me happier than if it was freely done by scientists, and for scientists, to put an end to the evil domination of the printing industry.

When in View Mode, this editor would highlight a sentence if the user hovers to it, and pop up a balloon showing author, year and the first three words of the paper's title. Just so that the reader can know the source. If double clicked it opens up the free paper (preferably) or just the corresponding web page so that the paper can be purchased or accessed. 

We have fairly decent search tools for scientific papers already, but they could be incorporated into this scientific text editor, just to make it a complete toolkit. A few spell checking tools too would come in handy. LATEX support could be nice too. 

Real gain would come when these scientific paper files started talking to each other. I'm talking about seemingly creating webs of references,  collaboration networks, measure Research productivity, assist in the so necessary peer-review process. It could even lead to AI parsing better scientific papers and performing all sorts of time demanding tasks.

That is the idea: Science is much too important to be done the same fashion as it was in the previous century.