I noticed that in your question you is highlighted three times. The point of a good OSS project is I think that is not about you, how awesome your library is, or how it is better / faster / cleaner / smarter / nicer than what everybody else is doing. Most people don't care about that. It is about your (potential) users. Does it make their jobs easier?
If that is not the case, you have hard sell. If it is the case, finding some early adopters that use it in their (pet) projects and are really enthusiastic about it. Their enthusiasm is more worth than blogs, readme's or demo's. Directly approach people in the field to give your lib a try regularly works for this.
Selling to the first users is mainly a thing of getting your idea across. I see too many libraries that dump code examples in their readme, but don't make it clear to my why I would want to take that approach, or how they treat the problem domain different then before.
So my 2 cents:
That is what worked for MobX at least so far :)