We prefer sources of different sorts for different reasons. Journals and conference proceedings that are printed on high-quality paper and bound professionally are considered archival. They have lasted for hundreds of years at least and may be found (even with difficulty) for generations. Online publishing of traditional-style articles provides easy and quick access, and properly converted to text, annotated, or indexed is highly searchable. Both of these modes are typically bound up with peer review in order to also provide some measures of quality, correctness, and authority, though no system of peer review is perfect.
Blogs, online videos, and social media appear to lack basically all of these features, but that doesn't make them unciteable. Given the choice between citing a video lecture of a professor demonstrating the proof of a theorem or a journal article where the same theorem is proved, cite the journal. But given only the video, the culture of academic publishing and writing is that you must cite it. If Terry Tao proves something new on his blog that relates to a paper you are working on, you should cite to it if your article is to be published before he can turn it into a traditional publication.
There's a hierarchy of sources with some disagreement to the ordering. But if there's something that you can point to which backs your point, proves your theorem, or demonstrates your idea, you have an obligation to consider it for citation. Different fields will have different standards for how compelling this obligation is. Pointing to a random blog to bolster your theory of the American Civil War probably won't pass muster in most History circles, but primary sources testifying to the fact you want to demonstrate usually will. In math or CS, the standards are probably a bit different.
