While Einstein relied upon journals to share his research we technically do not. Today, I can share my thoughts with the world instantly through a variety of social networks and channels. Indeed, you're likely reading this piece through some social network. Most researchers, however, don't produce documents in a way that is compatible with the web, such as PDFs or Word. Yes, you can share Word documents and PDFs online just like you can print a tweet-- both are possible, but neither really make sense. Thus, for the most part, researchers are still reliant upon publishers to share their work online and therefore they are reliant upon publisher's models. To move beyond this, research writing needs to be designed for research and the web. That is, research writing should be collaborative in real-time, semantically structured, data-driven, in HTML or XML, and version controlled. It is finalfinalFINAL_2018.docx time to move to a new system.
By utilizing a system designed for research and the web, like Authorea, writing effectively becomes publishing. This is powerful as it gives researchers the tools of publishers, eliminating the need for publishers. Thousands have taken advantage of this already; compare an Authorea article to any traditional scientific publication and you will find that it is not only more accessible but also more advanced than many publications from the world's largest publishers. Of course, this is even true of some blogs and other tools outside of traditional academic publishing, like Github. So then why has research publishing not been revolutionized? Why do we continue to buy into antiquated models and methods? Because, although it is necessary to give researchers the tools to do it themselves, it is not enough to give researchers this power. Indeed, preprinting, which is the closest thing we have to a democratized model of publishing still encourages researchers to buy into the traditional system because that is what is tied to career advancement. 

"Where did you publish?"

Publications are, in effect, the currency of scientists. They are used to "buy" grants and careers and like money, come in different denominations of worth. A publication in one prestigious journal will advance your career more than ten open preprints or even open access articles--regardless of the actual content of the articles. Researchers often ask "where did you publish?" instead of the more apt, "what did you find?" And that is the crux of the problem, incentives for advancing our careers align with traditional publishing and traditional publishing does not align with good research practices. Again, this sounds like hyperbole, but it's not. Researchers hack their results and statistics to fit a preconceived narrative in order to get published (see Interactive Fig \ref{958229}).