> Why would a youtube video with a DOI not be an acceptable form of publishing

I can, at least personally. For some stuff video might work, but text has a much higher level of precision and information density than video. When writing an article, I will often edit sections many times until it has the right amount of information, references, and flow. That is much harder in video, and scientists don't know much about video editing as is...

Also, when learning something, it typically takes multiple articles. I regularly have 5 articles open on my desk (yes, printed out), where I'm cross referencing equations, methods, etc, with one another. That is much harder to do on a computer, let alone with video.

Again, maybe for some fields it makes sense (maybe?). But it's hard to beat ease of use of text.

(Also, there's nothing wrong with conforming to a particular style. It's about consistency across authors, which helps the reader. Software projects have style guides, don't they? As well as traditional media outlets. Why shouldn't scientific publications?)

> Also, when learning something, it typically takes multiple articles

This is where I think video would be most helpful, actually. To relate it to software, when I'm learning a new language or a new tool, a video helps me navigate and learn my unknowns. But when I'm already an expert on a topic, video is intolerable because it takes too long to get to the point and often doesn't have the information I'm looking for.

Likewise for science, a video recording for materials and methods would probably be incredibly helpful for replication purposes. Also review papers might lend themselves better to video format because their purpose is to summarize existing research in a way that's digestible to someone relatively new to the field. It doesn't fit all purposes but it would make the science more accessible

About videos: I would argue it have a overlapping niche position with (either in-person or virtual) conferences. In optics community, SPIE and OSA do publish recordings of some (mainly hybrid) conferences. Those videos are indeed useful for learning about quasi-real-time works in progress from fellow colleagues (in slightly more depth than the PDF abstract) and save you from traveling.

Please not video. Video is a horrible way to communicate anything non-trivial: it's slow & difficult to navigate around; it's difficult to extract from (eg. I can't just copy a reference); it's difficult to skim; and it's very time-consuming to edit/revise.

Sure it's not easy to write good documents, but that's half the point - good documents take effort and time to create. The world is full of very low value papers. If the author isn't willing to devote the time and effort to write well, I'm unlikely to devote my time on reading something that is probably trash!

Many of video's advantages are actually disadvantages. Video is easy to produce, so the web is filled with low-value video. Video is easy for low-literacy people to produce, but scientific literacy is less about language and more about thought (I'm perfectly happy to read a idiosyncratic or ungrammatical paper, if it is clear and the contents are valuable).

Scientific writing should be all about communicating with the reader. So unless a video is better for me, the reader, stick with text. Please.

> I can't just copy a reference

You could definitely refer to a certain timestamp of a video.

A reference would be the video referring to another paper (or video). That's difficult to extract from the video unless the publisher is going to fully annotate it - and if that's the case, why not just publish the annotations and images? Much faster to read!

There's already a peer-reviewed video-oriented journal: the Journal of Visualized Experiments http://jove.com/ The downside is of course it's not open-access by default, and it can be expensive even if you don't pay for open-access since they'll send a freelance videographer if you don't have equipment/time/confidence to make the video yourself.