UC Berkeley makes course video content unavailable to public
news.berkeley.edu
news.berkeley.edu
I don't understand their motivation, did they think Berkeley would spend the money to generate captions? How is the current situation a win for anyone, now nobody can use the videos.
EDIT: Ugh, they even have this in remedial measures: "Pay compensatory damages to aggrieved individuals for injuries caused by UC Berkeley’s failure to comply with title II."
This is absolutely ridiculous. A university was providing something, for free, to everyone who wanted it. And because someone complained that their personal circumstances made it impossible to enjoy this free thing, now no one is allowed to get this content.
It's as if I threatened to sue Costco because one of their free samples had nuts in it and I couldn't eat it, resulting in them canceling the program.
I imagine the reasoning is that if you allow the university to deprioritize the needs of the disabled on the basis of "Well, something is better than nothing", then it incentivizes the production of such content because of the lower costs, and results in more disenfranchisement of the disabled.
I have to object to your reducing the situation to "because someone complained that their personal circumstances", as if it were just one person who is potentially disenfranchised by the situation. Just because one person's name is on a legal complaint doesn't mean that the law impacts that one and only person.
The disenfranchisement of the disabled is a problem, but if the logical (for Berkeley) solution to this problem is the disenfranchisement of all, perhaps we shouldn't go through with "solving" this problem. Perhaps it is better to accept that some amount of unfairness is inevitable, and that it's preferable to have these videos publicly available than to not.
EDIT: Plus, as noted, the chilling effect of damages awards.
The Americans with Disabilities Act came in in 1990. The fact that even after twenty seven years organisations chose to ignore the law shows the contempt that organisations have for people with disabilities.
Accomodations are reasonable if they require a small adjustment to the product - like automated speech-to-text transcriptions would be, or making navigation so that it's possible to use with screen-readers for blind people. However, if the required adjustment effort overwhelms the original product, then I'd say that "don't look a gift horse in the mouth" applies - yes, the content isn't perfect, and it's not suitable for many uses, but it's a gift; if you don't want it, then don't take it and it shouldn't be anybody's duty to make that gift differently.
People tried asking nicely, and were ignored. Now people are taking action. We can't say we weren't warned. We were warned 30 years ago when the law was passed.
Really, you could turn it around and call it contempt on the disabled's part, for suing with no clear benefit to anyone
The ADA does not generally apply to individuals.
> from here on out
From 1990, which is when the law was introduced. We should have been doing this for nearly thirty years now.
It costs roughly $1 for a professor to prop his camera up at the back of a room and record himself speaking for an hour, then post it online (11GB @ 3¢/GB = 33¢; we'll add double that for bandwidth &c.); it costs $180-$600 to close-caption that video. The ADA turns lecture videos from something anyone can do to something basically no-one will do.
Yes, life sucks when one's deaf. Yes, it'd be awesome for every organisation to have the resources to close-caption every recorded lecture. But requiring every recorded lecture to be close-captioned merely ensures that to a first order of approximation no lecture will be close-captioned.
Huh? I do not follow
It's a public university. They could pay a grateful student a federally-subsidized wage to transcribe 5 hours of video per week, and have it count toward the University's community service quota.
No, both of your figures are wrong. I actually worked in UC Berkeley lecture recording and captioning, and the figures are on the order of magnitude of $10/hour for each. The problem lies in the massive backlog of older courses that would need to be captioned for only a few non-students per course.
And a 10x increase is still large, although perhaps not prohibitively large.
> To add it all up, an untrained captioner could easily spend eight to ten times the length of a video to create a timed caption file. Let’s call it 9X on average, meaning that each minute of video takes nine minutes for this beginner to caption.
$1 / min of video * 1/2 = $0.50 / min of video 1 min of video = 9 min of transcription time $0.50 / 9 min of transcription time = $3.33 / hour of transcription time
What Berkeley was doing afterward was saying "well, we've provided this class in compliance with the law, might as well open up access to the materials to anyone who could benefit from them".
The court's interpretation would be like saying, "If a store gives away unused, edible meat at the end of the day, they must provide an equal amount of vegetarian food."
I can understand "equal access" arguments for "if you sell meals, you must provide a vegetarian option" (which Berkeley did the analog of). But why would they be expected to do the same for "byproduct" giveaways?
To use a hyperbolic example: you can see that "Free meals (NO GAYS)" would be discriminatory.
>The ADA’s nondiscrimination mandate states that no qualified individual with a disability shall, by reason of such disability, be excluded from participation in or be denied the benefits of services, programs, or activities of a public entity, or be subjected to discrimination by a public entity
The "or be subjected to discrimination" is possibly what they're getting at.
Yes, I can. And if that situation modeled the dynamics of the current one, it would change my mind. But I don't see how it does.
They're not denying the free meals/course materials to gays/vegetarians/the deaf. They're giving away byproducts that, by their nature, gays/veg's/the deaf cannot use, which came about as an incidental product of legal, non-discriminatory behavior.
For an analogy that does capture the dynamics: imagine if some factory's production process spewed out a waste product that happened to be perfectly usable as an STD prevention device for hetero sex but not homosexual sex. So the owners shrug and say,
"Hey, anyone who wants these hetero STD prevention devices, feel free to take them away!"
Would you say they're violating the spirit of such anti-discrimination law if they didn't also pay for the manufacture of gay-optimized condoms at a 1:1 ratio?
There is nothing in the nature of these videos that makes them unsubtitleable.
It also promotes a social harm. It says it's fine to discriminate against people with disabilities.
Do you seriously believe that no one should be able to give away uncaptioned videos, on the grounds that doing so "promotes the idea that it's fine to discriminate"? Khan Academy should have been shut down until their videos were accessible? Personal blogs?
Do you not draw any distinction here between a class for sale vs its free materials?
Accommodation is great, but you also have to appreciate the costs as well as the benefits.
I understand the intent of the regulation, but there should be other ways than do not disincentivize releasing content for free.
If I'm blind and I go to an art gallery should there be an audio track to somehow explain what I'm not seeing?
What about a play? Are plays specifically exempted or is it just that nobody has bothered to sue a theatre for a lack of subtitling just yet?
What if I go to a fancy restaurant but have no sense of taste or smell? How will that restaurant compensate me for the experience I'm missing out on?
The ADA also covers things like plays (classified as "entertainment"), and plays like Hamilton were sued for not providing audio descriptions of what is happening on stage. Movie theaters have to provide the same service to blind users and this also applies to Netflix movies and shows.
Restaurants probably wouldn't fall into the entertainment category, so they're not covered by the ADA.
For the "deaf and blind" scenario, most litigation is held off by the provision that says the establishment doesn't need to provide an accessible experience if providing such experience would be "unduly burdensome". For example, a blind and deaf man sued Cinemark for not providing tactile ASL interpreters for him (he communicates using sign language and by touching the hands of someone signing to him). The theater said it costs $60 per hour for an interpreter and a movie would require 2 interpreters, so $120 per hour is a bit nuts.
But audio descriptions and subtitles are not considered unduly burdensome - they're not that difficult and are basically seen as a requirement at this point.
Doesn't that mean that libraries must remove books that are not also available in braille?
In most cases, the resolution is to provide the material in the format required to the user. This is usually a trivial matter as books can be ordered in braille fairly easily.
The university decided that making their content available to all users was too difficult, so their resolution is to make it available to no one. A library could do the same and that resolution would be legal. It's just stupid. Which is why people are saying what the university is doing is stupid.
I was absolutely astounded when I found out the state of the art today. I was used to, "Computer, how do you wreck a nice beach" from a couple of decades ago.
Why can't a place like Stanford auto-caption things? Especially things like lectures which are super easy - a single speaker droning on in a silent room using high-quality speakers. without any editing the output should be absolutely useable.
The only time it really wouldn't work at all is in a very noisy environment where you can hardly understand it anyway.
I realize some of the captions might end up garbage, but I don't think enough would to make it into a complaint.
By the way, I find Google's autogenerated CC is useless on any video no matter the type or audio quality. (For example a lecture.) Another reason to think the link uses something else is that it takes several seconds (during pauses) to come up with the result - it is slower than real-time. I don't know if Google takes the time to slowly process every video in case someone clicks the autogenerated CC... The quality definitely strikes me as different. (And one of the reasons I was so surprised that the above link is actually useable.)
I'll grant that on previous occasions YouTube's auto captions have been hilariously wrong, but these days it really seems to be more fully baked.
I was curious to see how well the feature worked on older videos, and tested a random CS lecture from 2008.[2] Captions were perfectly fine.
The caption feature rolled out September 2009. [3]
[1] https://support.google.com/youtube/answer/6373554?hl=en
[2] https://www.youtube.com/watch?v=QMV45tHCYNI
[3] https://googleblog.blogspot.com/2009/11/automatic-captions-i...
The Department found that [...] many had automatic captioning
generated by YouTube’s speech recognition technology. In March 2015, the Department
selected 30 videos. [..] for review. Automatically generated captions were inaccurate and incomplete, making the
content inaccessible to individuals with hearing disabilities.Sigh. No good deed goes unpunished.
Reminds me of this Louis CK bit: https://vimeo.com/69662330#t=4m39s
"How does the world owe you something you didn't even know existed 30 seconds ago?"
Mandatory disability accommodation is a huge part of our society. There can certainly be reasonable limits (for example, noncompliant buildings that predate the ADA were not torn down).
And yet, here we are, with a building build after the ADA, that is noncompliant. It's too expensive to make it compliant, so the only legal option is to tear it down.
While I understand the strive for equality, the destruction of information is a price that should not have to be paid in today's world.
ADA imposes much more expensive mandates to much smaller businesses, and I'm under the impression that most people consider that to be acceptable.
"Automatically generated captions were inaccurate and incomplete, making the content inaccessible to individuals with hearing disabilities."
So it seems like they need to be quite accurate.
That's really not a lot. Probably blame a dumb and inflexible budgetary process.
Have you ever tried transcribing a video in a level of detail such that someone who can't see the video has a full understanding of what's going on? Give it a shot and report back to us.
Source: I've actually transcribed instructional videos for disabled students.
Update: Google suggests budgeting 4-9 hours of transcription time per hour, which is closer to my experience.
You might get by with 4 hours if it was just someone talking (with no slides, etc.). If the person performing an activity (say, a lab course) 9 is going to be a lot nearer the mark.
Remember, this has to be done so that both deaf and blind students can use the material.
Suppose it costs $100 to make a video accessible for those with hearing impairments. Furthermore, suppose this video is really important - the university is willing to spend up to $20,000 on making it, but fortunately only has to spend $15,000. Without being a dick about "deaf accessible or you can't publish", the university would get $5000 of producer surplus and the hearing impaired lose out. With this rule, the university gets $4900 worth of producer surplus, and the hearing impaired can actually watch this very important video.
It's really easy to forget about the cases that aren't getting fought over but are still subject to the rule. If you've got an idea for a better rule that lets Berkeley publish these videos, but still keeps universities from pretending that they can't afford to make videos accessible when they actually can, I'm all ears.
They are releasing high quality content online for free to a large segment of the world's population. Just let them do it!
Here's another idea: The people who are so concerned about accessibility could join their efforts and provide a transcription. In multiple languages, by the way. We don't want to exclude anyone.
It's pretty clear that such arguments lost with the passage of the ADA.
Furthermore I have never heard of a building being closed until accessibility accommodations are built.
If you think about it critically for just a second you realize there is no line. Accessibility is an absolute in the United States.
If you disagree then I wish you the best of luck in your campaign to repeal the ADA.
Should they provide transcriptions to satisfy the transcription validation committee, I'm sure you'd be demanding the videos to be taken down because they are not also transcripted in Spanish or dubbed in French, or some other excuse that would prevent the majority of people to have access to the content (including those who could be interested in working on transcriptions instead of complaining online).
The non-obvious thing is that this compliance-by-force is non-local. If you only look at Berkeley's videos, you're giving up these videos pointlessly. You have to also look at the videos that are going to be published anyways, and whether they'd include access for the deaf if people weren't dicks about enforcement.
Would it be better if this was approached from the other side? Creators could create without this concern. Disabled consumers would then pay for services that make specific creations accessible to them. The US government could then explicitly subsidize the disabled people.
If you insist on looking at this incident in isolation, yes. But if you look at it like that, there's no reason to imprison murderers, because it won't bring the victim back to life.
If there were no requirement, very few things would be accessible. As a minority, the deaf will always be irrelevant to the majorities' utility calculus.
I think ThrustVectoring said it best in a post above yours: we're requiring creators to subsidize the disabled.
An alternative way to subsidize the disabled might be to give them some amount of money every year which they can use to buy transcription, parking, etc. services. This might result in only certain content being transcribed. This might result in less parking spaces that are continuously empty. Maybe that would be a better use of everyone's time. Maybe it would create unnecessary decision burden on disabled people.
Also, if your goal was to prevent murderers from murdering again, why would you ever let them out? To generalize this to other crimes: why would you let any criminal out if your goal was to prevent them from committing more crimes?
http://www.businessinsider.com/why-norways-prison-system-is-...
You would let murderers out because you think there is some compromise between their freedom and the crime they committed.
What is your suggestion?
Shutting down videos to millions of potential viewers because the recorders did not have the time nor money to produce transcripts, gains nothing / takes away everything.
If things were really fair, we should be able to now sue the complainants for taking away from us everything that could have resulted from viewing the course material.
This is, uhh, kind of false. In liberal democracies like the US, we have some extraordinarily protections for the needs of the few at the expense of the many. For example, we've ruled as unconstitutional anti-miscegenation laws. The founding fathers were explicitly worried about the tyranny of the majority, too.
If Berkeley isn't going to educate me to the point where I can take advantage of these free videos, it's not right that anyone else should get them either.
Unintended consequences of a well intentioned law - everyone loses access to information, even many disabled people who still could have used it.
A better approach would have been to build a tool that allows crowdsourcing of captions with a way for students with special needs to tag videos for prioritisation.
URLs of the videos won't be removed: "Individual video URLs (links) will remain unchanged."
http://news.berkeley.edu/2017/02/24/faq-on-legacy-public-cou...
How could one get all URLs and put them in one place for all 20k+ videos?
edit: If Netflix can be sued and forced to provide captions, so can any other entity providing uncaptioned video.
As pmoriarty mentioned above, youtube-dl works very well for downloading entire playlists:
http://news.berkeley.edu/2017/03/01/course-capture/
The university is removing public access to all of the YouTube and iTunesU video content that is currently available. Apparently this is 20,000 videos. For example, here are the lecture courses in Computer Science that are available but will soon disappear:
http://webcast.berkeley.edu/series.html#c,d,Computer_Science
The reasons for doing this seem to stem largely from this letter from the Department of Justice
https://news.berkeley.edu/wp-content/uploads/2016/09/2016-08...
which informed the university that the free online video content was in violation of the Americans with Disabilities Act.
Fortunately the videos were being shared via torrents so I could finish the course, and eventually reuploaded by others on their own personal youtube channels.
Oh right! Thanks!
Q: Why now? Is this related to the DOJ letter?
A: ...The Department of Justice letter indicates [...] Course Capture content [...] is in violation of the Americans with Disabilities Act.
Later it is mentioned specifically this is due to the videos are not captioned. Thus the solution to this is to remove all older recordings.
Then, have online students sign up to write sub-titles or verify subtitles. For every hour of subtitling you get 20 hours of hearing. Verification via a CAPTCHA like system that interrupts video playback to verify snippets of audio. Sell this corpus to deep learning people.
With hundreds of thousands of people doing MOOCs it should be very doable.
There are many types of disability and to produce content that is fully inclusive is almost impossible.
You have hearing problems, various types of vision problems including color blindness... you can have extreme cases where someone is deaf-blind and also has no hands therefore braille doesn't work. Can that person deny content to everyone else by suing and force everyone to adapt to their needs? Can you possibly cover each individual disability?
This sets a precedent that invites other universities to also take down all their content, basically wasting a lot of excellent, valuable hard work.
Berkeley was compelled to remove the content, they chose to in order to avoid the costs of resolving the dispute, to focus on unquestionably compliant content moving forward.
"And then they came for me ..."
The Department found that [...] many had automatic captioning
generated by YouTube’s speech recognition technology. In March 2015, the Department
selected 30 videos. [..] for review. Automatically generated captions were inaccurate and incomplete, making the
content inaccessible to individuals with hearing disabilities.Rush - 'The Trees' [1]
The cochlear implant issue itself is a rather complex and interesting one. Most hearing people are immediately outraged when they learn that some deaf people want to prevent their children from hearing, but I think this is a knee-jerk reaction that stems mostly from ignorance of Deaf culture[1] and its value in the Deaf community.
There's a great documentary called Sound and Fury[2] that explores this issue by interviewing deaf and hearing families as they struggled to decide whether to have implant surgery performed on their children. I strongly encourage everyone to watch this documentary before making up their minds on this issue.
Also, talking to some Deaf people, getting some exposure to and knowledge of Deaf culture, learning about the long and ugly history of discrimination and paternalistic interventions by hearing people against deaf people, and the struggle of deaf people for self-determination would probably be a good idea too.
With this knowledge I think it becomes difficult to dismiss the issue as ridiculous or say that there's any easy, obvious answer.
[1] - https://en.wikipedia.org/wiki/Deaf_culture
[2] - https://en.wikipedia.org/wiki/Sound_and_Fury_%28film%29
The libertarian response would be that you have the right to be how you are, but the ADA should be torn down.
How children should be treated is a tricky subject since it is a balance between parent's rights to raise their children and society's interests and I think there is a lot of conflict that falls into this bucket.
As technology develops to the point that being deaf is a choice, I am much more sympathetic to the libertarian view that if you choose to remain deaf you should not be entitled to special support.
I have a congenital hearing defect in both ears, and have hearing aids since I was a child. (I am not deaf - only hearing impaired).
If I could undo all that, and fix my hearing, would I? Heck yeah. Without a doubt.
I also have a daughter - and I most certainly would not wish a hearing impairment on her. She doesn't need to go through the c*ap I went through.
If I was profoundly deaf and used sign language - well, I still don't think I'd wish deafness on my enemy - let alone my own child.
You don't see people with blindness or bad eyesight going - gee, I'm glad I'm blind or have bad eyesight.
I'd say some of the issues involved are somewhat analogous to the Europeans and Western people coming to "civilize" indigenous populations, which were viewed as primitive and inferior, and towards whom the Europeans held the attitude that they know best what's good for the "primitives" -- but without actually knowing much of the indigenous peoples at all, surely not understanding them or their culture, often misunderstanding them, not appreciating them, and at the same time in many ways it was the Europeans who were inferior. But they were blind to it.
Still, these Europeans managed to virtually wipe out the cultures and languages of the native civilizations and languages they encountered and dominated.
Today sign language and Deaf culture both face a very real possibility of extinction in the near future. Those are yet more reasons that many in the Deaf community are concerned.
At the risk of making this issue even more political, but with the aim of clarifying what's at stake for some in the Deaf community: imagine if what was at stake was the extinction of the Hebrew language and the Jewish identity.
Once again, I'd recommend watching the documentary I linked to for a clearer understanding of some of the issues, and to hear some of the reasons for the anti-implant position from the mouths of the affected people themselves, parents and children.
On the other hand, some deaf people and hearing people with deaf children, especially those who are not part of the Deaf community and who don't know sign language, feel like you do. The choices even in these cases are still not easy, as the film makes clear. Check it out.
I highly recommend watching the short followup, "Sound and Fury: 6 years later": https://www.youtube.com/watch?v=Nb-zKebd5us
Sign-language as a living language might suffer.
But would you consider this a net positive?
An analogy might be if blindness and sight problems became eradicated - and braille suffered a decline.
I would consider this a net positive for human quality of life.
I don't see the analogy between Hebrew/Jewish identity, and braille/sign language and blindness/deafness.
To me, the first is a cultural/religious identity - whilst the latter is an adaption to communicate as best as one can, with a more limited toolset. Of course, I see that it's not that black and white - there is a thriving deaf community, and they do have their own "language" (as in, more than just sign language itself), cultural norms etc.
However, I do really hope for one day where we might eradicate all of disabilities - deafness, blindness, mobility impairments etc.
We detached this subthread from https://news.ycombinator.com/item?id=13769321 and marked it off-topic.
That sounds like the real reason for the move.
https://news.berkeley.edu/wp-content/uploads/2016/09/2016-08...
It seems that the reason is genuinely that they were told that their free video content was in violation of the Americans with Disabilities Act.
EDIT:
Hmm, I think I was unclear. I am not referring to making a "professional grade" captioned video. I am simply referring to having literally what the speaker says included with the video as a transcription.
I do admit it would probably take more than $20. Even if it was $200 a video that would still be pretty cheap. It's probably what, less than a cent extra per student per semester? Unless of course you do this for every new iteration/instance of the class.
Have you ever transcribed content? The length of the recording is the lower bound on time it takes to transcribe, not the upper bound.
Furthermore, transcribing _just_ the audio is not sufficient to meet the ADA requirements. You also have to create an accessible way to describe all graphs and pictures for the visually impaired (generally, that means you have to describe each and every photo using words).
Furthermore, I can't imagine any public university administrators wants to get questioned by legislators about why they're outsourcing work nowadays. Nor do I think the administrators want to admit that they're paying (by first world standards) sweatshop wages.
If you find yourself in a US or UK hospital, it's very likely that your medical reports and prescriptions dictated by your doctor are sent off to India, Pakistan, Sri Lanka, Philippines, or elsewhere to be converted into text.
So yeah, English isn't a problem, and there is already a huge pool of talent in the 3rd world trained in turning audio into text.
If a picture is worth a thousand words, and the video is 30fps. Then a one hour video would require well over 100 million words. I agree, too much to do in real time.
It's harder than it sounds.