Background

For those completely unfamiliar, YouTube is an advertiser-supported service that allows users to upload and download videos for free. In a world where personal upload bandwidth is not yet remotely sufficient for most individuals to self-publish even text and images, this serves a fairly valuable purpose. Although to start it seemed to be publishing primarily unauthorized copies of already published material, it seems to have matured to the point where it consists primarily of home clips of people recording their pets, their vacations, their silliness (or outright stupidity), and their personal accomplishments.

Okay, so perhaps that doesn't sound all that valuable — but it has also allowed for users exercising their fair use rights to post short clips catching politicians in the middle of lies, voter suppression, or other scandals, and provide those clips to anyone needing them as a reference. It's also provided a means for average citizens with average budgets to create and distribute their own political ads with at least a chance of getting more than just their family and neighbors to see them. Although the numbers for that ad project aren't particularly high (less than 5,000 viewers), it got citizens involved in a political process normally reserved for our gentry and major corporations, and short news clips documenting wrongdoing got a much larger audience.

The Problem

Perhaps it is an effect of the recent acquisition by Google, but the people administering YouTube seem to have lost track of exactly what it is that they provide. First it was YouTube sending a cease-and-desist over footage of Larry King Live where Bill Maher claimed that Republican Party chair Ken Mehlman is gay. CNN appears to be very unhappy about this footage, as not only did they get YouTube to send the cease-and-desist (which could hardly be for copyright infringement, as the entire posted scene was less than a minute and a half out of an hour long program, posted non-commercially, in classic fair use fashion), but they also altered their official transcripts so that the entire exchange disappeared. The ability to prevent evidence (in this case either of slander or of hypocrisy, depending on which political side you may wish to take) from disappearing is one of the primary societal benefits that YouTube provides, and yet they have jumped to negate that value here. It was a somewhat futile jump in this case, as well: the person who posted that clip is planning to cut it back to ten seconds (so that it is indisputably fair use) and repost it — and the entire clip is still visible on the Huffington Post. (I suspect he will still have problems: although he may be within his fair use rights to redistribute such a clip, he has no contract with YouTube that states that they must help him do so. CNN can't stop him from putting the clip up on his own website, if he can handle the bandwidth, but YouTube is well within their rights to decide what they want to transmit from their site, as long as they aren't claiming any sort of common carrier status.)

More recently, though, YouTube also sent a cease-and-desist to TechCrunch... for the crime of having placed on their website a simple form box that converts a link to a YouTube video display page to a link directly to the video (in FLV format). This is useful (though not my preferred method, as I'll get to below), because YouTube itself doesn't provide any easy way of doing this — deliberately. This isn't Google's fault; YouTube was built this way originally, trying to prevent anyone else from sharing the video that its users were openly sharing through YouTube (and both Lawrence Lessig of Creative Commons fame, who also wrote an amusing entry on the chutzpah of a company that made its fortune by riding a wave of copyright violations now attempting to restrict downloading, and Joichi Ito of Digital Garage and Infoseek Japan have written some interesting thoughts on "true sharing" sites vs. "fake sharing" sites). This also harms the societal value of a site such as YouTube, as the value of information that can be archived and built upon is far greater than the value of information that cannot, but that's well within YouTube's rights: nobody can make them provide a tool they don't want to provide (especially when it's to their financial benefit, e.g. more advertisement views, when downloading complete video is harder to do).

But there's a difference between failing to make something easy and trying to suppress the publication of any other tools to make it easy — and I am avoiding the word "possible" for a reason, because there is no way to make capture impossible (more on that below). In this case, it is of even more dubious legal standing than the cease-and-desist sent on the behalf of CNN: ostensibly, by their own terms of use, all of the users uploading videos to YouTube have already signed away all related rights:

YouTube's response once this was pointed out was even more telling:

... demonstrating that Mr. Levine, at least, seems to have a very shaky grasp of what "streaming" entails.

The Technology

A "streaming" video is nothing more than a video compressed and compacted in such a way that it can begin to play before it is completely downloaded. It still must be downloaded. The data must move from the server to the client, which can then pass it to the player application to display colored pixels on the screen. If you permit users to "stream" video from your site, you permit them to download video from your site.

YouTube makes a minimal attempt to maintain control of the data by first downloading to your computer (as a Flash applet) the player that it wants you to use to play the video. The only way to make such a transmission feasible, however, is if it's really just a thin layer over a player that's already installed on the computer, playing a standard format of video: in this case, the Macromedia Flash player, playing FLV files. If this were not the case, the size of the player could well account for more bandwidth than the size of the video with each transmission. All that the YouTube player really does is to draw the screen border and controls in the browser window, and then ask the (already installed) Macromedia Flash system to download and play the actual video from the same website. So if the Flash system can request the video to be played from the website, what's stopping anyone else from downloading the video from the website with another application?

Absolutely nothing, as it turns out. The webserver is honoring a standard web request to fetch a file. YouTube can't even rely on the "through the functionality of the website" clause, because in fact the video is served up straight by the website. In fact, this process is so ordinary, that depending on your operating system, you may find a complete copy of the video file in a temporary directory or in your browser's cache. If you're running Firefox on Linux, just look in /tmp, and you'll probably find a copy there, ready for you to copy it to another directory at your leisure. (If you want more detailed instructions on doing just that, they can be found on the O'Reilly website.) If you pass your requests through a local proxy, you can just check the proxy logs and find the actual URL requested by the flash player, and then redownload that file. Or you can look at the information in the page itself: in Mozilla Firefox, if you right click on the background, select "View Page Info", click on the Media tab, and then look for an entry of type "Embed" with the .swf player in it (it will look something like "/player2.swf?video_id=r90z0PMnKwI&l=62&t=OEgsToPDskJynqmdr249uvEfdIzsoqjr"), you can simply replace "player2.swf?" with "get_video?", stick "http://www.youtube.com" on the front (so it reads "http://www.youtube.com/get_video?video_id=r90z0PMnKwI&t=OEgsToPDskJynqmdr249uvEfdIzsoqjr"), and then download that way. Or you can type "about:cache" into the URL bar, and then search through the page for entries ending in ".flv". Or you can use one of the many downloadable widgets to make this simpler. The TechCrunch tool isn't even a very good example of such a widget: it runs entirely on the TechCrunch website, letting them know which IPs are downloading which videos every time the tool gets used, which isn't my idea of a bright way to preserve your viewing privacy, but that is literally all it is doing: changing a few letters around in the URL to point you at the spot where the actual video is being downloaded to your computer by the viewer, so you can save it and view it with another application or at a different time. My own personal favorite utility for this is a bookmarklet that also handles Google Video links in a similar way (and it's a pity the Greasemonkey extensions on the same page no longer seem to work).

So is there anything YouTube could actually do to prevent this? Well, they could make it a little harder, but ultimately no. In the end, for it to be useful to the end-user, it has to display pixels on the screen, and if nothing else, the end-user can simply recapture those pixels as they are displayed frame-by-frame, recompress, and save, with some minimal loss of quality from recompression. If they want to do it remotely conveniently, they have to actually transmit video in a form standard enough for it to be useful to users for reasons other than their site alone — and that means that it's infeasible to prevent the actual displayed file from being capturable and replayable later. Even if they were to go to a protocol that didn't pass over HTTP, they would have to prevent anyone from reverse-engineering the video stream well enough to create a new player that could save the stream to disk. This is what happened to the RealPlayer mms and rtsp protocols, for instance: sure, you can pay them extra for a client that can save some streams, but you can also just download MPlayer for free — an open-source client that can dump the streams as well.

The Ethics

But even if they can't do so technologically, should they be able to do so? Is is ethically or morally or even legally the correct choice to support YouTube's right to try to prevent individuals from locally storing its videos? The answer is again fairly clearly no. What YouTube is attempting to do here is to curtail Fair Use rights. The right to store a local copy of a video transmission for time-shifting or other rights covered under Fair Use was pretty solidly confirmed by the Betamax case, Sony v. Universal Studios, in 1984. This doesn't apply to redistribution, mind you: if you are deliberately retransmitting those downloaded videos in their entirety, that's definitely not covered by Fair Use — but even here, users are in the clear, because all of the uploaders have already granted all of the downloaders rights "to use, reproduce, distribute, prepare derivative works of, display and perform". YouTube's users thus have no grounds to prevent exactly that, and YouTube, not having been assigned copyright, has no standing to do so for them. If they change their terms of use, as they have indicated they intend to, they can withdraw that additional authorization, but they cannot withdraw the individual's fundamental Fair Use rights.

So if YouTube, as their general counsel wrote, is actually encouraging content be submitted on the expectation that nobody will be able to store the files on their own computers for later viewing, they are encouraging content submission based on a lie (and particularly given their fine print, a rather egregious lie). Not only that, but it is a promise that they have no right to make, regardless.

But I don't believe the letter: users that read the fine print know that they've agreed to unlimited redistribution, and users that haven't obviously don't care that much. And none of them should expect to curtail Fair Use.