 Hi and we're live. I'm Dazza Greenwood, a scientist at the MIT Media Lab and instructor for the upcoming MIT Computational Law course. And this video is an intro to Wright's Expressions Languages with Bill Rosenblatt, who is joining us right now by Hangout. And Bill, thanks for taking the time to come and go through these slides and then also to be available to the students at MIT next week. Much obliged. So perhaps you can give a little more background about yourself and then let's dig right in. All right. Well, thank you very much, Dazza, for inviting me to do this. I'm very excited about the class that you're teaching and it's a pleasure to be involved with it. As you mentioned, my name is Bill Rosenblatt and I am a consultant. I run a consulting firm called Giant Steps Media Technology Strategies. I'll just give you a little bit of my background here. So, yes. So I have a firm, as I said, which I've run since the year 2000. My background is sort of multifarious. I've been a software engineer. I have been a media industry IT executive. While I was a software engineer, I contributed to the GNU project, the open source project back in the mid-1980s. I worked as a tech market strategist for Sun Microsystems in the publishing and media industry. I have been a book author and editor and a radio producer. And I more specifically wrote a book on digital rights management back in 2001 or so. I run a conference called Cupwright and Technology Conference. And I've spoken at various events worldwide and I've guest lectured at other places. This is my first time at MIT. I've been across the river at the Berkeley College of Music. I've been in Columbia, Carnegie Mellon, and a bunch of other places. And I have served as an expert witness in various litigations that have to do with digital media and copyright technology. And I've served as a public policy advisor on digital copyright technology in the U.S., Europe, and Asia. So that's a little bit about me. Essentially, I'm not a lawyer, but I play one on TV sometimes. And what we're going to talk about today is really at the intersection of tech and law, which is a perfect fit for the course that you are teaching. So I want to start out by mentioning a book that I'm sure that many people who are taking this class are familiar with, which is Larry Lessig's first, at least his first book for a general audience, Code and Other Laws of Cyberspace, published in 1999, a seminal work. And he shows in that book something called or explains something called the Four Forces of Regulation, which has come to be known as the Pathetic Dot Theory. And the dot is in the middle of this diagram, and it represents people who are controlled or affected by four forces. And the four forces apply regardless of where you are, whether you're online, whether you're in the real world. These are the four forces that regulate your activities and your behavior. And they are law, which is just laws as they're written and executed. The market, which is the way businesses and economics behave. Architecture, which is technology, physics, things of that nature that are natural phenomena in whatever world you occupy. And then norms, which are behaviors, human behaviors, which are perhaps those that arise out of convention or what people, the way people are expected to behave out of convention. So those are the four forces. And when we're talking about what we're going to talk about today with rights expression languages, we're really talking about how these forces interrelate in the online world. So the other thing that I think it's important to establish, even though some of you might be familiar with this already is the basic copyright, the basics of copyright law and particularly the rights that a creator of a work such as a piece of music, photograph, sculpture, painting, novel or what have you are granted under law. This is what we call sometimes the copyright bundle of rights. And the first of these rights is something that is so fundamental that a lot of people ignore it when they're looking at this stuff. And that's in section one on one of the copyright act. It is the fact that copyright applies to material objects, which means CDs, DVDs, books, films, etc. It does not apply to digital files independently of the medium on which they are stored. So hold that thought. We're going to come back to it later. Other rights, the ones that are more typically discussed when we talk about the basics of copyright law are the rights of, in the second bullet point, reproduction, distribution, creation of derivative works, public display and public performance. So reproduction simply means making copies. Distribution simply means sending or giving your work to somebody else. Derivative works are things like film adaptations of novels or edited versions of photos or anything anything that you do to make something else out of a work that still has some characteristics of that work is called a derivative work. Public display, public performance, pretty self-explanatory. And then there are certain types of visual works such as paintings and sculptures that have additional rights under copyright law such as the right to be attributed properly to the creator of the work and the right for the author of the work to prevent harm to his reputation by damage or mutilation of the work. That's something known as a moral right. And there is a limited set of moral rights in US law. And by the way, I should mention all of this is US specific. If you are in another country, your mileage may vary, although some of the concepts may translate to other copyright laws. There is also something called the first sale doctrine, which is more typically described as a limitation on rights. And that is when you, if you're the owner of a copyrighted work and you give it to somebody else, you sell it to them, you give it to them, you lend it to them. Well, mainly you actually let's express it differently. If you obtain a copyrighted work lawfully, you buy it, someone gives it to you, what have you, then you can do whatever you want with it without any control from the entity that gave it to you. So the most typical example of this, you buy a book, you can resell that book, you can give the book away, you can use it as a doorstop. You buy a CD, you can resell it, lend it, give it away, use it as a Christmas tree ornament. This applies to most types of physical content, except certain types of software, and we won't get into the minutiae of that. But that's an important thing to recognize that if you, if you have a copyrighted work, then you have the right to dispose of it as you see fit. Okay, so now let's talk about, we've talked about the exclusive rights that a creator has to his or her work. Now let's talk about what others can do with works under copyright law. So if you get a copyrighted work, you can use it yourself. That's the normal case. You can also license rights from the copyright owner in addition to the rights that you already have. And you do that by contract. Then there is something called fair use, which is something that people have discussed for hours and days and months and years. We're going to just talk about it very generally. Section 107 of the copyright law. Fair use is really a defense to an allegation of infringement. If I claim that you've infringed my copyright, you can, in a court proceeding, say, no, I've made a fair use of it. And here's why I think I've made a fair use of it. This is something that a court decides. And it's based on a number of factors that are codified in the law. It's also based on a number of ideas that have come up through case law history, such as this idea of transformative use. If you transform the content in some way, then it may be a fair use of that content. So that, again, a lot can be said about that. The one thing that I want to make sure we understand here is that fair use is not something that can be decided by an automated process or a machine. It's something that a court gets to decide. And in the footnotes, you'll see a quote from a seminal law review article that was written in, I believe, 1989 by Pierre Laval, who is a federal judge, now an appeals court judge. And the money quote from that article is, we should not adopt a bright line standard for fair use unless it were a good one, and we do not have a good one. And that is essentially the reason why there is no machine-decidable fair use. So let's keep that in mind as we go on as well. So we're here to talk about rights expression languages. These are machine-readable rights descriptions. And the purpose of them is to make rights and licenses understandable and actionable at internet speed as opposed to human speed and avoid something that I call the trap door into the legal system. You don't need to hire a lawyer. You don't need someone to interpret and explain something to you. You can just act on a description of rights by automated process. So the more specific goals of rights expression languages are generally twofold. These are overlapping but distinct goals. The first goal is to enforce rights, and this first in chronological order of how these things were designed, to enforce rights in a technological scheme for enforcing rights. And then the second goal is to automate licensing processes to just conveniently and efficiently automate the process of licensing rights. So we're going to talk about both of these things. We're going to start by talking about RELs or rights expression languages for rights enforcement. And this, of course, means a DRM mechanism, digital rights management. This is something that came into being in roughly the mid 1990s and one of the handful of pioneers in this field and the one who specifically pioneered the use of rights expression languages for DRM systems was a researcher at Xerox Park named Mark Stefik. Stefik is a researcher who had an AI background from Stanford before he went down the peninsula as it were. Maybe it's up. I can't remember it to Xerox Park. And he wrote a paper called Letting Lose the Light, Igniting Commerce and Electronic Publication. And in 1995, he and some colleagues from Xerox walked into my office when I was working in the publishing industry and gave me a copy of this paper, which I thought was fascinating. The paper was subsequently published as a chapter in Stefik's book, Internet Dreams, Archetypes, Myths and Minniforce. That was published in 1996 by the home team by MIT Press. And it contains chapters by a bunch of different authors. It's a very interesting book about concepts and ideas around the early days of the internet. And he also took out a patent on this subject area, one of several patents that he took out on DRM systems. This one that what we call the 403 patent, which actually expired recently, is for system for controlling the distribution and use of digital works, having attached usage rights where the usage rights are defined by a usage rights grammar. And this has been claimed at least to to mean a DRM system that uses a rights expression language. There has been licensing, there's been litigation over this, and that is what the claims are said to read on. I'm not expressing an opinion on that. I'm just saying what has been said to have been the case in in litigations and licensing discussions. So the rights expression language that Dr. Stefik invented was called Digital Property Rights Language. It was based on the list as befits his background in AI. And what Xerox did was it commercialized this system, a DRM system using the DPRL language, and they called it content guard. So content guard over the years in the 1990s morphed into something called XRML, the extensible rights markup language. I'm not going to go through this in detail. But basically, in the late 90s, 9798, XML came into being and that, of course, is the very, very widely used language or meta language for expressing structured data and structured content that's been been used in a million different applications. And so Xerox and Stefik recast DPRL as an XML based language. And then they spun out content guard into a separate company called Content Guard to market this DRM system. Microsoft became a part owner of the company as did media company Time Warner, which is why I call this rights language the establishment rights language. It was owned and controlled by a large major media company, a major technology company and also a French technology company called Thompson. And essentially, and then they presented XRML to MPEG to try and make it into an official standard. MPEG did adopt a slightly modified version of XRML as something called MPEG rel. And then pretty much nothing happened. Nobody really used MPEG rel for anything. Microsoft did use XRML, not the MPEG variant of it, but XRML in its software licensing scheme. So for example, Windows Office, other Microsoft products use XRML for managing licenses to that product that they grant to PC users. So that's that was the first sort of major commercial rights expression language. And then I'm showing here some example code, which basically is the code for you can pay $15 to read an ebook and to print it. And I'm not going to go into this in detail. If you're, you know, conversant on an XML code level, you can go look at this code. Then there is what I call the indie rights expression language, which is ODRL, the open digital rights language, which was created a few years after XRML by a guy named Renato Hianella in Australia. He worked for a systems integration firm called IPR Systems. And he did this, along with a woman named Zuzana Gutt in Vienna, Austria, as opposed to Australia, she was a business professor at a university there, with a tech background, very conversant in this stuff. They developed this thing. And they did it independently of any company or standards body. So they kind of had no particular institutional affiliation. But they did socialize this in various places. And the main place where they got some adoption was in the mobile telephony industry, particularly in Europe. There was a standards body called the open mobile alliance, which had a whole family of mobile telephony standards. And they were helping to develop standards for mobile digital content services, such as digital music services. And they adopted a small subset ODRL for the OMA digital rights management standard. And that engendered a whole bunch of implementations in the mid 2000s in Europe, got a lot of momentum going. But over here on this side of the pond in the US, people were skittish about all the patents that were out there that might cause some issues on patent licensing. And so there was a reluctance to embrace this technology, not just the Xerox patents, but others as well. And not much happened with this technology in the US. And eventually, the whole scene kind of went quiet. And ODRL went into a state of dormancy for a while. But hold that thought, because we'll get back to that later. So this is just some information about what ODRL looks like conceptually. It specifies these permissions, which include usage permissions or rights, if you will, display print play execute, reuse permissions, the right to modify to excerpt, annotate, aggregate, these are your derivative work type permissions, then transfer permissions, which are like your distribution rights, they map to distribution rights and copyright law, sell, lend, give or lease. And then there are rights that have to do also with distribution, such as move, and also reproduction, take a backup, make a duplicate and so on. So these map fairly well to core concepts of copyright law. And I'll just also mention a couple other enforcement oriented rights expression languages that were around and in one case continue to be around real networks had one, you may remember real networks, they were the streaming media pioneer of the late 90s through 2000s. Microsoft sort of came in and see and proceeded to dominate that market, but real networks had a whole ecosystem around its streaming media technology, including a DRM system for which they had a rights expression language. Adobe had an e book ecosystem, which they acquired from a startup called Glassbook in 2000. And that came with a rel called EVX, which Adobe subsequently integrated into its own ebook offering. And the Adobe ebook platform is still widely used around the world. In the United States, it's most notably used by the Barnes and Noble Nook system. And in a system called overdrive, which is very widely used in public libraries and also school libraries, if you quote, borrow unquote an ebook from a library, then you're probably using the overdrive system, which is based on that Adobe technology. So now we're going to switch to the other main purpose for rights expression languages, which is license automation. And that's again simply machine readable license terms to use in a license automation scheme, typically between businesses, not from businesses to consumers, but B2B. And if there's any enforcement of this scheme, it's done through the traditional legal means of, you know, hey, you're in breach of our contract, our license agreement. And then there's a discussion or a lawsuit, what have you, but there is no enforcement angle to the technology around license automation rails. The origin, at least what I consider to be the origin of this work was a paper written in 1993 by a Villanova law school professor named Henry Perret, who got both undergrad engineering and Sloan MBA degrees from the home team MIT. And he wrote a paper called permission headers and contract law, which was presented at a conference that I consider the seminal conference of digital rights management, which took place in Harvard in 1993. And in the footnote of this slide, you can see the way you can still to this day, get the proceedings, which make for very interesting historic reading, if you're into that kind of thing. And what Perret essentially said was you can have contracts, they can have machine readable headers, machines can read them and act on them in certain ways. And here are the legal implications of doing that and the benefits of doing that, a very, very influential piece of work. The first standard to take this approach was something called ICE, the information content exchange. And this came out during the first dot com internet bubble, when there was kind of a fad for something called content syndication, which was this idea that, let's say you're in your Condé Nast or Hearst, your magazine publisher, and you want to send articles and photographs out into the world that people can buy for 50 cents a piece or what have you. And this can all be, you know, sliced and diced up into individual content items and the whole process can be automated. Interesting idea, never really took off commercially. And the ICE standard was an attempt to, you know, create a standard to automate the licensing. RSS came along much, much simpler than the very complicated and Baroque ICE or ICE standard. RSS, of course, is very, very widely used today. ICE kind of faded after a few years. So in order to talk about where we are nowadays with these license automation languages, we first need to talk a little bit about creative commons. And again, some of you may be familiar with this already and in which case I apologize in advance. But creative commons is very important. And there's also an MIT angle to it. So Larry Lessig is famous for among other things, being the founder of Creative Commons, but the famous computer science professor Hal Ableson at MIT also had a big role in that and his name will pop up again shortly. The basic idea of Creative Commons is that it's based on a point of view that copyright has gone too far and is too lengthy in duration and gives copyright or rather create creators of copyrighted works too much rights and not enough rights for the public. So therefore, we need to rebalance things a little bit and create a system whereby the public gets more rights. And the incentives for copyright for creative people to create copyrighted works are more in line with reality, etc. etc. That's the basic idea behind Creative Commons. And they have a motto, a motto, some rights reserved, as opposed to all rights reserved, some rights reserved. And as a practical matter, what Creative Commons does is it provides a set of licenses that in which the licensee gets more rights than the licensee would under the copyright bundle, which we discussed a few minutes ago. And it's also important to bear in mind that although Creative Commons can apply to any type of copyrighted work, it can apply to something physical as well as something digital. It's based on an assumption that the copyright bundle applies to piles of bits, which as I said by themselves do not are not covered under copyright. A pile of bits independent of a storage medium is not covered under copyright. It's covered under a license agreement. And Creative Commons is a set of license agreements. So that's all okay. But it's just important to bear that kind of fundamental assumption in mind about Creative Commons. Creative Commons is a set of licenses that are expressed as machine readable code as well as the legal text of the licenses. And you choose from a set of licenses to assign to your work by choosing among options or attributes for things like are you going to require the licensee to attribute your work to you the author properly? Are you going to allow them to reuse your work for commercial purposes? Or are you going to forbid that? Are you going to allow them to create derivative works from your work? Or are you going to require them to copy your work on an as is kind of basis? Very interesting system very widely used. So what Hal Abelson told you we'd get back to him did about 10 years ago was to create a more formal rights expression language for expressing Creative Commons licenses. Creative Commons licenses had been expressed and indeed are still expressed to this day as comments in HTML if you're dealing with an HTML page. And what he wanted to do was to make the license terms sort of more integrated with the content instead of being relegated to comments. And so he created a set of terms that could be used as a rights expression language that are directly expressible in RDF in XML content and the RDF is simply a way of expressing metadata or semantics in XML and also can be expressed using the scheme called Adobe XMP which is the scheme that Adobe uses for expressing metadata in multimedia files such as photographs, illustrations, video and so on. And again, the idea here is just to integrate rights expressions with the content more directly eliminate redundancy. That's the dry principle. Don't repeat yourself and it is not not not at all meant to turn Creative Commons into a DRM scheme. That is absolutely not the intent here. In my view not enough attention has been paid to CC rel. It's a very interesting idea. I don't see it being used that much. I think maybe it's being used somewhere, but it's certainly worth talking about. It's a nice piece of work. And I have some example code here. This is a Creative Commons license for a photograph. It's in two pages and it requires and the license here you can see where is it in the next page. Near the top you can see that the license is the bi-nc-nd license which is the Creative Commons attribution non-commercial no derivative works license. That's the license that this photographer has chosen for his work and he wants you to if you're going to use this content to attribute the photograph to Scott Beal slash slash laughing squid and there's a URL that he wants you to stick in there so that someone who sees this can click on it and then view more information about laughing squid which presumably is his company. So that's just some example CC rel code that you can look at. ODRL has also returned from dormancy as a license automation language and it was morphed into a major revision called ODRL 2.x and in 2011 Renato Ianella and others submitted ODRL to the World Wide Web Consortium to attempt to make it an official standard which as I mentioned had not been done as yet at least not successfully. So roughly a year ago the World Wide Web Consortium did advance ODRL to recommendation which is their official standard stage. So now ODRL is in fact an official web standard. It is a syntax independent rights language that is expressible in various XML RDF syntaxes. There are a number of there's a few of those that are popular and it's also expressible in JSON which is my preference it's cleaner it's easier to read more widely used on the web and so on and they're very adamant in assuring everyone that this is not DRM. This has nothing to do with DRM schemes and one of the reasons why they are adamant about ensuring people of that is they want to make sure that this doesn't come into the realm of the patents that that I explained previously the Xerox patents and so on. So the language is really based on this core concept of permissions and obligations or duties and in fact the original working group within the W3C was called the permissions and obligations expressions or co-working group and the primary application of this language is in the news industry with something called writes ml and writes ml is a standard of IPTC the international press telecommunications council which is a standards body of the news industry and the Associated Press is using this today for their rights language or rights license automation and in the footnotes you'll see a reference to a slide share presentation done by Stuart Miles from the Associated Press which is which is really worth a look as an excellent explanation of the practical application of rights expression languages for license automation they're really doing this they're really you know walking the walk over the Associated Press which as you may know is the largest news gathering organization in the world. So this is a diagram taken from the ODRL website that shows you the kind of data model that's used in the ODRL world with the latest version 2.2 which is a very recent version and it's based again on permissions and duties you see that on the right side and then on the left side you see assets those are the media objects parties those are the licensees licenseeers and then various constraints on the on what the licensee can do within and so on and so I'm providing here a couple of ODRL 2.1 examples that are taken from again this is actually the GitHub repository of for ODRL these are expressed in JSON and this first sample says here's a movie you can only play it in Germany after 2018 and if you look here you see there's a URL for the movie under target and then you'll see there is a date a spatial constraint for well first there's a date and time constraint it's got to be after January 1st 2001 and it's there's a spatial constraint which means it has to be within Germany. Another example is here's a movie you you can only distribute it over mobile networks as opposed to let's say home broadband or satellite or or what have you and here you have a distribution right with delivery channel must equal mobile essentially and again this is a JSON version of this example you can look at the code pretty easy to read pretty straightforward here's another example specifically from writes ml which again is what they call a profile of ODRL like a special case of it or a subset of it this is expressed simply as a data structure in kind of pseudocode which makes it even easier to read you have a picture that can be displayed by a licensee within Germany and it's being distributed by the EPA which is the European Publishing Association or the European Press Association and so you have permissions the assigner is the EPA the assignee is EPA partners the constraint is spatial and it must equal a DEU which is the code for Deutschland or Germany and here is the same code written in JSON you can also do it in various RDF syntaxes and again this is real this is used in production today through the AP and possibly others there are a few other license automation rights languages that are worth mentioning and they tend to be specific to particular niches within the media industry in the stock image industry commercial image licensing photo licensing industry there is something called plus which stands for picture licensing universal system it's one of these reverse engineered acronyms and a guy named Jeff Sedlich runs this initiative and it's meant for various flavors of image licensing very specific to that space then there is something in the nominally magazine consumer magazine publishing community um underneath the umbrella of a standard family called prism which again reverse engineered acronym I think it stands for publishers requirements for industry standard metadata and they have a right language for that type of content in the government and academic library community so library of congress university libraries and so on there is something called meth rights which if you're from that community you'd be familiar with the mark standard it's a related standard if you're sending digital materials from one library to another here are the rights that you have with these materials and then the other one that i'm aware of that's interesting is actually I think the most recent one onyx pl onyx is a set of metadata for books and that includes physical books it covers things like it was originally created in the late 1990s to deal with online physical book retailers like amazon and it deals with things like here's how big the book is here's how many fit in the box here's how many pages long it is and so on and so onyx pl is the permissions language for onyx and those are all languages that are in use for license automation so that kind of covers rights expression languages for license automation I want to take a step back now before I go into music and the music modernization act example and just talk about provide a little context provide a little big picture here and talk about how these things interrelate rights expression languages DRM's licenses and copyright and have these all fit together and in a way what we're really talking about is less x4 factors we've got code we've got technology we've got laws and we've got behaviors how do these things how do rights expression languages cause these things to interrelate so licenses what are they they're legally enforced rights and restrictions in a contract DRM that is a set of technologically enforced rights and restrictions on content independently of whatever legal terms may apply DRM is technology rails can be used in either paradigm to enable precision in a DRM system or in a pure contract license system rails can encode licenses that confer more rights than the copyright bundle on the licensee that's the basic idea of creative commons to give the licensee more rights than the licensee would get under copyright law they can also be used to to describe licenses that confer more restrictive rights or less rights than the copyright bundle and if you're looking at the end user license agreements for many commercial content services like amazon or apple or barns and noble or what have you it's going to be true of those agreements they will be more restrictive with regard to digital content you can't resell it you can't lend it you can't publicly well you can't do that anyways and never find out more restrictive the relevant point of law here is again that you don't get the copyright bundle with pure digital content pure digital content is something that is covered only under license agreements and this was just reaffirmed by the second circuit in a court decision that just came down within the last few weeks in a case called capital records versus redigie very interesting case redigie is a boston they still exist boston area startup in fact in boston across the river from mit and their basic business model is you buy songs on itunes you can resell them as quote used unquote through their platform they were sued and the court found that what they were doing was a violation of copyright law they are still in business because they have a new architecture which they think might pass muster with the law but that has remains to be seen and so the second circuit kind of reaffirmed this idea that you've got to tie content to a physical medium in in order for copyright to apply and the judge said something like again mr pierre leval judge pierre leval who we mentioned earlier having written that seminal harvard law review article he said you know you could apply a copyright to a set of music tracks that are on a thumb drive for example then the whole thumb drive full of music would be something that's covered on the copyright but the songs just floating around cyberspace are not relevant point of technology that's worth mentioning rights expression languages enable precision and certainty but they do not enable drms to be capable of emulating copyright rights so many people like to say drm is about restrictions to rights that you would get under copyright what i like to what i prefer to say is that they are simply not capable of emulating copyright it's apples and oranges in a way and there was actually a very interesting project that took place a decade ago over a decade ago the standards initiative called the digital media project that illustrates this this was started in 2003 by a man named leonardo perron in inlake he is the founder of mpeg so he knows a thing or two about tech standards and his idea was to create an open interoperable drm standard set that respected and emulated real world content usages including copyright rights and that worldwide not just in one country or another this still exists it's kind of on low simmer but the most interesting in my opinion thing that they did was they tried to create this set of mappings of drm constructs to what they called trues or traditional rights and usages and in this powerpoint there's actually a url to the webpage that shows you the extremely lengthy list of traditional rights and usages that they collected or kind of crowdsourced among their participants it's incredibly exhaustive and yet the work was never completed which just really shows you how this is perhaps an impossible task to take on so now we're going to talk about the music modernization act and we're going to talk about music licensing but first i'd like to just establish some basics about music rights and again some of you may know this already in which case i apologize and you can skip maybe if you're viewing a recording of this but one thing that's most fundamental about music recordings is that they carry two copyrights each one is for the underlying composition can think of as the sheet music with the lyrics and then the other is for the sound recording and if you think of the Beatles yesterday that was written by John Lennon Paul McCartney so that's the composition and then there have been hundreds and hundreds if not thousands of covers of that song and they are sound recordings by whoever covered that song Beatles or whoever else it was today's popular subscription digital music services such as Spotify apple music google play title amazon music unlimited etc typically offer three types of services that each require a set of licenses from copyright owners in music interactive streams which are you choose the track you want to hear it plays the track what we call conditional downloads which have also been called tether downloads which is you download a track to your device and it will play on your device until your subscription is no longer valid you've canceled it or what have you and that's often used for enabling you to play a song in your device without a live internet connection such as in in new york where i live you're on the subway and you don't have a connection you want to hear a song then there are permanent downloads which are drm free files that you buy and they are downloaded to your device and you have them in perpetuity notice i didn't say oh you're licensing them but you have them they are on your device and those are the basic three paradigms that today's big services offer and in order to offer them these services which we call dsp's digital service providers that's a music industry term they have to pay a set of royalties depending on what type of service they're offering and there are a number of these but here's some here are the most common examples and these are typical cases there are exceptions there are a lot of nooks and crannies to this but these are the typical examples you have to pay a a royalty for reproduction and distribution of the sound recording to a record label you have to pay for reproduction and distribution of the composition to a music publisher and again this is like the sheet music with lyrics and in the music publishing part of the music industry they refer to this as a mechanical right and a mechanical royalty you also have to pay for the public performance of a composition to something called a p.r.o. which is an organization like ASCAP or BMI and so on or in some cases you pay it directly to a music publisher if it's one of the big ones like Warner Chapel or Sony ATV the second one of those i've underlined it because we're going to talk more about it there is a compulsory license under copyright law which means the copyright law guarantees a license of that nature without you having to negotiate with the rights holder in advance and in the law the royalties are set they are set by law and so that's the compulsory license for mechanicals now there are private agreements that supersede that compulsory license but i'm going to through the purposes of this discussion pretend they don't exist because it just makes the discussion a lot simpler so the music modernization act i know is something that other people in this course are going to talk about and this is again probably redundant to what some others like george howard and vicky nowman are going to talk about and they're extremely knowledgeable about this stuff but just to set the stage for what i'm going to talk about here i want to go over the basics of the music modernization act which is a piece of legislation that passed very recently late last year basically what it does relevant in relevant part is it revises section 115 compulsory mechanical license in a few key ways one is that under the pre the previous or prior compulsory license you had to take a license to each track individually that license was guaranteed to you by law but it required that you did certain things like file paperwork with the license with the license or with a copyright office this process has proven to be a gigantic pain in the neck for these service providers it was fine when all you wanted to do was do a cover version of the Beatles yesterday or what have you then it was a simple process you filed some paperwork you got your license and you were often running but if you're a Spotify or apple music or google play you are acquiring tens of thousands of tracks from record labels and independent music aggregators every day each one of those requires some paperwork to be filed and each one of those has some data about the composition that underlies the sound recording that could be that you're not getting directly from the music publisher or songwriter so it could be missing it could be incorrect it could be even in dispute so you're taking in all this information you have to get this in this information about music compositions that enables you to get a license you may not know where to get that information it's a source of trouble and it's led to lawsuits with lawsuits against Spotify and Apple and it's basically a big become a big problem that's reared its ugly head now that services like Spotify Apple music and so on have become the dominant source of revenue in the recorded music industry i'm going to pause take a drink of water because i'm about to lose my voice okay i'm back okay welcome back thank you so um what services like google play Spotify Apple music title and so on have been doing is they've been processing all this paperwork by themselves or more commonly hiring a third party service provider such as the harry fox agency or a company called music reports to do it for them under the music modernization act all that is replaced by a single um licensing agency that is appointed by the us copyright office that will process all of these mechanical licenses for all the streaming services that are eligible under certain conditions so it's a it's effectively a de facto monopoly nonprofit agency that will do all this work and the copyright office will choose an entity to do this in response to proposals by i believe it's july of this year so a lot of big changes will will happen as a result of this um the music industry in general is very happy that it's passed i think that it's got frozen cons but unbalanced it's a good thing for the music industry you'll remove a lot of uncertainty and paperwork it will also increase royalty payments to songwriters and music publishers as a result of other aspects of the legislation that i won't go into here so now let's talk about how this would relate to rights expression languages now that we understand the fundamentals of mechanical licensing and all that stuff so what we're talking about here is rights that a publisher a music publisher is granting to a digital music service provider and they are rights that exist because of a blanket statutory license that exists for all tracks you know all music tracks period um and one of the things that you can draw on to represent this in a rights expression language is an existing standard in the music industry called DDEX which stands for digital data exchange this is a set of messaging primate or well messaging constructs in in in the language which can be expressed in xml or i believe common delimited ascii or something of that nature which include all the information that a record label wants a service provider to know when they have released a recording so if i am you know universal music group and i'm sending a feed to youtube or to Spotify i'm going to send information about all these releases that i'm putting out in this format and one of the pieces of information that i'm going to send in addition to the basic metadata artist title certain standard identifiers such as an isrc which is standard sound recording identifier is the set of rights that i'm licensing to this service provider and among them are the three rights that um are typically provided by these dsp's on-demand streaming conditional downloads and permanent downloads there are others and there's DDEX documentation that i believe daza has put up on the course website that you can look at so yes and one just a little bit of context setting um if i may when you say dsp's to give people a sense of the type of market participant you're referring to would that be for example spotify or apple music yeah so dsp is the is the music industry jargon term for digital service provider and what we're talking about here is exactly that your spotify is your apple music your google play is amazon music unlimited napster title deezer um and and a few others that i can't pick up at the moment um but but those are the those are the biggies and they they all offer this kind of combination of services for end users and they're the ones who are specifically targeted for the benefits of the music modernization act yep perfect thanks i appreciate that uh kind of context setting and um and i did ask you to um talk a bit about this part of this part of the marketplace partly because i think it's something people understand and because there are some as you said like major reforms and changes coming with this new legislation music modernization act and uh when we get down into student exercises we may select other license types such as the so-called sync license for putting music to like you know film or tv but um but at a high level i i hope this is something everyone can relate to because most people do get their music this way nowadays well your stride i just wanted to provide a little context for what you're saying no that's excellent context and in fact two comments back at you first of all the other reason why it's important to focus on these dsp's is because they have become the majority revenue source for the music industry over the last few years they now represent collectively about two-thirds of music industry revenue it's not about downloads anymore downloads are in free fall vinyl is a nice little market but it's not going to be more than single digit percentage revenue share um cds you know on low simmer nowadays radio it's still there but it's not that big of a deal the major major part of the music industry nowadays is these interactive streaming type services so that's why it's important to focus on them and the data problems that they've had we're always there but it's just that no one particularly cared about them until they became the huge businesses with tens of millions of subscribers that they are today the other comment that i want to make about what you just said as a regarding context is sync licenses are very amenable to implementation in an automated more online marketplace using rights expression languages and i haven't talked about them today but i leave it as an exercise as they like to say to to explore that and it would be a welcome addition to the industry to do so because there have been several attempts over the years to create online automated marketplaces for sync licensing and by sync licensing what we mean is the licensing of music for use with things like video content tv commercials tv shows movies video games things of that nature big business kind of the wild wild west in terms of the way it's organized very amenable to automation for the lower sort of cost transactions not like negotiating with taylor swift to to make her song a theme song for a netflix sitcom or what have you but the more every day tasks of music sync licensing of which there are many many so with all that in mind i'm going to go back to my slides great thank you attempt to thank you for joining me on that little excursion back on the road we go okay back on the road we go and here we go so talking about representing mechanical licenses in a rights expression language rights expression language one issue that you have so that first of all there are two licenses that you can represent one is the license that the publisher confers on the music service provider and those again are these three types of things on demand streams permanent downloads conditional downloads then there are the license terms that the dsp confers on users through apps and devices through drm systems and yes for everything but permanent downloads drm is still used certain flavors of drm and one thing that's interesting here is that a publisher can issue a conditional download license to a service provider but just issuing a conditional download license doesn't really tell you anything about the parameters of that conditional download it doesn't say like you can play this x number of times or you can play this until the end of the month at which point you have to pay another month subscription things like that so there's a bit of a disconnect there which puts kind of a crimp in the in the automated ability of this so to speak but it's just something to look at if you're looking to automate this through a rights language so with that in mind what i have offered is an exercise to represent these mechanical rights that the mma music modernization act touches in a rights expression language and the simplest exercise is number one here just to take these mechanical rights and represent them as a data structure in the style of the rights ml data structure that i've showed a few slides back just as a as a pseudo code data structure then the second one the extra credit one is to represent it as an odrl profile in json or rdf and then the extra extra credit one would be to expand that beyond the mechanical license to all of the relevant rights for interactive music dsp's and to do that as an odrl profile in other words rights ml took odrl and made a profile of it for the news business which mainly has to do with photos and text articles they're working on expanding it to video and other types of news but right now it really is applied to photos and text and if someone were to take um take that example and create an analogous set of rights expression language constructs for music licensing to digital music services that would really move the needle of the music industry and could be very useful as things like the open music initiative which george howard talks about in this class move forward with their automation on blockchain so that kind of ties it all together and if someone were to create this odrl profile for the music industry for these dsp's usages or even for sync rights that would really be a contribution to the industry and as far as i know and i have looked nobody else is doing this right now so with that this is all my contact info i'm happy to talk to anyone and i look forward and you can always look this up later i look forward to meeting every participant when i come up to cambridge uh next week for the class excellent well thank you so much bill that was um exactly what i had hoped you would say and the information i hope you would provide and it was so clear as well and so we'll post your a pdf and a powerpoint version of your slides um in the presentation actually i think we already have it on github but we'll link it to um to your session page so the people can get those very important links that you provided um down to the source material especially perhaps including the um odrl specifications and i think yesterday when we were prepping for your session uh we identified a few more helpful links to examples uh developer like the dedex document and and a few other odds and ends yes a few other odds and ends some example uh license types or um kind of um pseudocode squibs uh from odrl which i thought were helpful and maybe even some leads on some parsers or you know some uh some processors so still working on that right with it if i have more to add i will certainly add it um still working on that great um okay then um so then we look forward to hosting you uh at MIT on Tuesday and um and we should have um it's an online session as a reminder but we should be able to actually see the almost like brady bunch style all the all the people joining on and and we'll be continuing to collect pigeonhole um feedback for questions and ideas and other comments on your marks in the lead up between now and your session so that we can really get off to a sprinting start super okay thanks again and we'll thank you soon