> Whenever you watch a video you are downloading it.

Why is this comment downvoted? It's highlighting one of the most common misunderstandings that laypersons have regarding video download/streaming. Most people think that you can "view" content on the internet without downloading it. In this context, a tool which purports to "download" content, you know... sounds like it's nefariously doing something that the "viewing" tool (like a web browser) doesn't do.

This may be completely true in a technical sense, but that's not how the law works (see https://ansuz.sooke.bc.ca/entry/23). And while the same bits pass through your connection, this equivalence already breaks down right away: There is clearly a difference between persisting a media file to disk vs having it ephemeral in browser memory.

>There is clearly a difference between persisting a media file to disk vs having it ephemeral in browser memory.

Is there? When "streaming" video, there most certainly is a copy of the bits being stored on a disk to ensure that the video "stream" plays cleanly and without interruption.

Are you making the claim that "streamed" video is never buffered/stored on disk? That's an odd claim to make. I'm no expert on video streaming, but I would be very surprised to find that all video streams are only stored in RAM and not on disk.

I may well be wrong about that. Perhaps someone more knowledgeable could chime in.

That's again exactly the technical detail fallacy the comment you replied to is arguing against. No, they are not making the "odd claim" you suggest. They suggest that laws make the difference when deciding about infringement. E.g. by explicitly excluding temporary copies created while watching as intended, where it doesn't matter how the OS and the browser handle the memory internally, but a thing that results in a file on disk the user keeps is clearly different. (Similarly to how software being copied into swap-backed memory while you run it is not an illegal copy, whereas copying the file elsewhere might)

Certainly with DRMed video it is common for the video to never be buffered/stored on disk. Sometimes a few seconds is, but even that is uncommon, and more likely it would simply be retained in RAM now.

With more secure DRM systems the OS literally never gets access to the video buffer, protected by hardware, in order to even send it to disk.

An interesting question along these lines arose recently in relation to an Australian password disclosure law that related to accessing “computers,” which was used to compel disclosure of a smartphone passcode. To HN readers and the digital forensics people who pull data off smartphones, they’re obviously computers. But the judge was not convinced that a law written to allow access to “computers” in the early 2000s was intended to allow access to smartphones today, which contain far more personal information than the typical personal computer of 20 years ago. After all, if you asked someone “do you have a computer?” they would be unlikely to say yes based on their possession of a smartphone. And if you ask someone who streamed a YouTube video whether they “downloaded” it, I think in most cases the answer would be no. That’s why the tool is called “youtube-dl,” even though it is now used for streaming as well.

> There is clearly a difference between persisting a media file to disk vs having it ephemeral in browser memory.

yes, at some point actual human intentions must come into play. you can't defend stuff like CP by saying "it's just some EM pulses, what's the big deal?". or "no I'm not invading your privacy with my IR camera, you are broadcasting in the IR spectrum!".

in this case the implementation does blur the line a little bit. what if the browser's memory gets swapped out to a page file on a (spinning) hard drive? even if the cache gets "deleted" after closing the tab, it might be quite a while before the sectors containing that protected sequence of bits get overwritten. is this infringement?

I agree there is a legal/practical/moral difference between streaming and downloading something. But there's no need to obscure the technical difference by downvoting people when they point it out.

The point is that youtube-dl does more than just download videos. It can also be used for downloading metadata. I use -J to download metadata formatted as JSON.

Is that metadata protected by some sort of mechanism? Or is it just not queried by default using one's browser? I.e., is youtube-dl calling a public, unsecured API, or is it circumventing some sort of copy protection?

Because if it's just querying for metadata that anyone can already query for...your point seems immaterial as to the legality of the tool?

They're saying that sure, the name youtube-dl might well imply it's specifically for downloading things from YouTube, but that doesn't mean it's specifically for downloading video from YouTube.

How much of that metadata is not normally downloaded alongside the video?

Plenty of it? I regularly download the metadata and subtitles of entire channels or playlists so that I can search for specific words or phrases in thousands of hours of video. I know of no other way to accomplish this.

Subtitles are a perfect example of data that any normal browser downloads if you click CC, and can even be ^F'd if you click ‘Show Transcript’ on YouTube, but just happen to be orders of magnitude more useful if you control where they download to. I think you’re proving globular-toast’s point.

Subtitles are not video, so no. And I'd like to see you visit a thousand video pages with an RIAA approved browser and ^F on each of them. Nice joke!

Why should subtitles and video be considered so discretely? Are subtitles not copyrighted the same way as the audio and video portions of the work?

Beyond subtitles, there is certainly video metadata that youtube might have a claim on but the RIAA does not.

The principles are essentially the same, but they are discrete copyrights which could be owned by different people.

Thank you so much for writing this comment!

I've been in several situations where this would have been incredibly handy, but never realized it was possible.

`youtube-dl --write-sub --write-auto-sub --sub-lang en --skip-download [URL]` (Then just use grep)

There's all kinds of cool stuff you can do with youtube-dl. For example 'ytsearch20:kittens' will get a playlist of the first 20 search results for 'kittens'.

This sounds incredibly powerful. Wondering if c-span is on youtube and properly close captioned.

According to their FAQ, c-span.org's search uses closed captioning to facilitate search (but they don't provide copies of those transcripts.) Perhaps that might suite your needs though.