Monolith is a simple tool that takes two arbitrary binary files (called a Basis file and an Element file) and "munges" them together to produce a Mono binary file (with a .mono extension).
Monolith can also reconstruct an Element file from a Basis file and a Mono file.|
In most cases, the resulting Mono file will not be statistically related to either file. If you compare the Mono file to the Element file, the Mono file will contain none of the information present in the Element file. In other words, the Mono file by itself tells you nothing at all about the data in the Element file. Only when combined with the Basis file will the Mono file provide information about the Element file.
Monolith can be used for exploring the boundaries of digital copyright, and the rest of this website is devoted to such an exploration. The core questions: What happens when we use Monolith to munge copyrighted files? What is the copyright status of the resulting .mono file? These questions are considered in depth below.
DownloadsYou can use Monolith with whatever Basis file you would like. However, standard practice dictates the use of a particular Basis file, Monolith_7D4.wav, when generating Mono files. This particular WAV file was created and copyrighted by me (Jason Rohrer), and I have placed this file in the public domain for free and unrestricted distribution. By following this standard, you can ensure that others can easily interact with your Mono files. After downloading the basis file, drop it into the base folder found in your Monolith installation.
You can download the Monolith application for the following platforms:
Copyright ==> ?Things get interesting when you apply Monolith to copyrighted files. For example, munging two copyrighted files will produce a completely new file that, in most cases, contains no information from either file. In other words, the resulting Mono file is not "owned" by the original copyright holders (if owned at all, it would be owned by the person who did the munging). Given that the Mono file can be combined with either of the original, copyrighted files to reconstruct the other copyrighted file, this lack of Mono ownership may be seem hard to believe.
Consider this simple fact: for a given Element file and any other file of the same length (call it fileA), it is possible to choose a Basis file that, when munged with the Element, will produce fileA as the resulting Mono file. Therefore, if a copyright holder claims that she owns the information in all Mono files that are munged from her work, she is also claiming copyright over all possible binary files that are the same length as her work. For example, suppose that fileA is an MP3 of a Beatles song, and the Element file is an MP3 of a Britney Spears song copyrighted by Jive Records. It is possible to find a Basis file that, when munged with the Spears song, will produce the Beatles song as the Mono file. Jive Records certainly cannot claim copyright over the Beatles song (which is copyrighted by Apple Records), nor can they claim copyright over any other Mono files munged from MP3s of their songs.
What does this mean? This means that Mono files can be freely distributed.
So what? Mono files are useless without their corresponding Basis files, right? And the Basis files are copyrighted too, so they cannot be freely distributed, right? There is one more twist to this idea. What happens when we use Basis files that are freely distributable? For example, we could use a Basis file that is in the public domain or one that is licensed for free distribution. Now we are getting somewhere.
None of the aforementioned properties of Mono files change when we use freely distributable Basis files, since the same arguments hold. Mono files are still not copyrighted by the people who hold the copyrights over the corresponding Element files. Now we can freely distribute Mono files and Basis files.
Interesting? Not really. But what you can do with these files, in the privacy of your own home, might be interesting, depending on your proclivities. For example, you can use the Mono files and the Basis files to reconstruct the Element files.
Bring Taping HomeThe Internet, despite all of the intellectual freedoms that it bestowed upon us, had an odd side-effect: it moved the "Copyright vs. Home Taping" clash into the public sphere. Home Taping used to be a private issue, since the act itself (the Taping) happened in the privacy of one's home. There were laws forbidding Home Taping, but they were unenforceable laws. The Copyright holders got their "blank media" taxes passed and otherwise gave up on their crusade to stop Home Taping. However, with the Internet, Home Taping has moved into the public realm, and the laws forbidding such Taping are suddenly enforceable. Now we have a royal mess: a public that is fond of Home Taping is butting heads with Copyright holders who are trying to enforce laws that have never been enforced before.
Monolith moves Home Taping back into the private sphere. There are laws that forbid combining a Basis file with a Mono file to produce a copyrighted Element file, but since such combining happens in private, those laws are unenforceable.
To best explain the logic behind Monolith, we need to explore the odd properties of digital information.
Digital Information is WeirdHow can something so simple be called "weird?" We have our standard grade-school explanation of digital technology: "everything is ones and zeros." Nothing weird about that. The oddities crop up when you try to represent analog (in other words, "physical" or "non-digital") entities digitally.
Copyrightable entities are inherently analog. Music, painting, sculpture, writing---all of these must be presented in the physical realm to be consumed by a human audience. Even mediums that are always created and represented digitally, such as digital photography, must be translated into the physical realm (for example, into a lighted display on an LCD monitor) to be consumed. The bits (the "ones and zeros") used in the representation mean nothing to us by themselves---we cannot experience or otherwise consume them.
The bits are fully devoted to the representation, not to the presentation. A digital photograph, represented as a JPEG file, can be presented in many ways---on a monitor, on an envelope with an ink jet printer, on photo paper, on the side of a building with a projector---the possibilities are endless. Furthermore, the same digital photograph can be represented with many different file formats (JPEG, PNG, TIFF, etc.). Each of these formats might use its own unique bit-level representation of the photograph, though the presentation (for example, the color of each monitor pixel) might be identical. In other words, many different sequences of bits can be used to represent the same photograph. In fact, if we count all possible formats, including those that have not been (and may never be) invented, an infinite number of different bits sequences can be used to represent that same photograph. We have not even brushed upon the impact of various resolutions and quality settings: there are thousands of ways to represent the same photographic content as a JPEG bit sequence.
But could we not make the same statement about analog representations? For example, there are perhaps hundreds of analog audio formats: 45 RPM records, 78 RPM records, reel-to-reel tapes (of various widths and speeds), 8-tracks, and cassette tapes. We can certainly imagine an infinite number of possible analog audio formats. How is this space of possibility any different from the space associated with digital encodings? For true analog formats, the difference is in the form of the representation. Analog formats represent one physical property directly using another physical property. For example, the intensity of a sound wave is represented directly by the depth of a groove on a record. As another example, sound wave intensity is represented directly by magnetic intensity on an analog tape. Though analog formats are representations of content (a record groove is certainly not the same thing as a sound wave), they are not encodings of content, since no "decoding" is necessary to reveal the represented content. For example, if you drag a sewing needle through a record groove, you can hear a faint rendition of the sound, and this particular operation cannot be described as decoding. True analog representations are isomorphic to the content they represent (in other words, they have the same structure as the content). Digital encodings, on the other hand, are generally not isomorphic to the content they represent, especially when examined on the binary level.
Looking at the issue from a slightly different angle, we can see that digitization involves representing something that is infinitely detailed (an image) with a format that is inherently finite (a sequence of bits). Digitization, therefore, is the process of deciding what to keep and what to throw away. Of course, an infinite amount of information must be discarded in this process, and there are clearly an infinite number of ways to do this.
The point is that, in the digital realm, the content is copyrighted, while the binary representation surely is not, given that there exist an infinite number of representations for the same piece of content. This point is well-demonstrated in the realm of digital music. The copyright holders record a physical act (plucked guitar strings, vibrating vocal chords, etc.) and eventually give it a digital representation (as a CD track). In most cases, the copyright holders never translate their content into other formats, like MP3 or OGG files---those translations are usually made by "unauthorized" third parties for the purpose of exchanging the content online. There are a myriad of different settings and algorithms that can be used to create an MP3 file, and each of these results in a unique bit sequence. The copyright holders do not construct any of these bit sequences themselves, so the sequences cannot be rightly called part of their work. The content, in this case the music, is what is copyrighted.
So, why is this weird? Because the Internet cannot be used to exchange content; it can only exchange bit sequences. When people download an MP3 file via the Internet, they are downloading a bit sequence, one of an infinite number of possible digital representations for a particular piece of content. It is only when that representation is rendered, or decoded and played through speakers, that the content itself comes into existence. Thus, it is the act of playing an MP3 of an unlicensed song that is actually illegal under a true-spirited interpretation of copyright law. Downloading or otherwise exchanging the bit sequence surely cannot be illegal, unless the copyright holders are ready to lay ownership claims upon an infinite number of different bit sequences for each copyrighted song.
People have been getting sued for sharing MP3 representations of copyrighted files, and its not even illegal, huh? That would be pretty weird if it was true, but it is still hard to believe, so I will tackle the issue from yet another angle.
Digital Content is FiniteSorry Britney, but I am going to pick on you. After all, what kind of digital copyright treatise would this be without an example based around a Britney Spears track? Consider the song "Toxic," which was released on Spears' album "In the Zone" in 2003 by Jive Records (an RIAA member). This song has undoubtably been translated many times from CD format into MP3 format. For all of the various MP3 quality settings and encoders used, there are probably hundreds of different bit sequences being used to represent "Toxic" on peer-to-peer (p2p) networks. None of these bit sequences were created by the people who hold copyright over the song itself (Jive Records), and there are (as argued earlier) an infinite number of other bit sequences that could be used to represent the song "Toxic."
I have a particular MP3 representation here, a particular bit sequence, stored in a file called "britney_spears_-_toxic.mp3". If we believe that all possible bit-level representations of "Toxic" are actually copyrighted by Jive Records (in other words, if Jive Records holds copyright over this particular MP3 encoding), then sub-sequences of bits from these representations must also be copyrighted (just like a paragraph of a book is copyrighted by the same person who holds copyright over the entire book). Here is a particular bit sequence that occurs in my MP3 file:
Obviously, since that particular bit occurs in almost every other binary file, I have not broken the law by "distributing" it. There is nothing special about that single bit, and it is certainly not "owned" by Jive records. How many bits must I divulge before I break the law? How about 10 bits? Jumping into the middle of the file by 1,087,320 bits, we see the following sequence:
This bit sequence is quite meaningless to us---it certainly does not resemble any part of the song "Toxic." In fact, that particular sequence occurs in the MP3 file more than 60 times. Undoubtably, it also occurs in almost every other MP3 file: there is nothing special about a sequence of 10 bits. How about 24 bits? Jumping into the middle of the file by 21,442,264 bits, we see the following sequence:
This 24-bit sequence occurs only once in the MP3 file, so it might seem like a good candidate for a sequence that is unique to my "Toxic" MP3 file and therefore copyrighted by Jive Records (if so, I have just broken the law by divulging these bits). Of course, in raw form, these bits look meaningless to us, and they certainly do not resemble any part of the song "Toxic." In fact, the sequence is not unique at all, as it is found in the ASCII text encoding of the Bible. For instance, the sequence occurs in the file genesis.txt if we jump into the middle of the file by just 968 bits. Here is the 72-bit context from the Bible that contains our 24-bit sequence (our sequence is in bold):
We can convert this binary sequence into ASCII text, adding a few more characters for the sake of context (our 24-bit sequence is in bold):
Yes, the 24-bit sequence taken from the "Toxic" MP3 file is actually the word "was" when it is interpreted as ASCII text. Of course, when interpreted as part of an MP3 file, it has a different meaning, perhaps the loudness of a particular sound frequency at a particular instant in the song. Again, there is nothing unique about the bits themselves---only when they are interpreted in a certain way do they really become content. With one interpretation, the bits are part of the copyrighted song "Toxic," and with another interpretation, they are part of the Bible, which is in the public domain.
Because digital representations of content are inherently finite (each representation is made of a finite number of bits), these kinds of examples can be extended to cover entire files. The MP3 encoding of "Toxic" contains only 38,804,992 bits. Though there are a huge number of possible binary files with this many bits, there are a finite number of them. There are an infinite number of imaginable binary file formats, and thus an infinite number of possible interpretations for those 38 million bits. Interpreted one way, these bits represent a song, but interpreted another way, they could represent text or an image.
This all sounds great in theory, but in practice we are using those 38 million bits to represent a song, and if we try to interpret those bits using another existing file format (for example, if we pretend that the bits are a JPEG file), we will certainly get nothing but garbled content as a result. The point is that the content itself, which is copyrighted, lies only in the interpretation of the bits. The bits themselves are meaningless to us and uncopyrightable.
So, back to the interesting question: How many bits must I divulge before I break the law? Here are 240 bits taken from a point 80,000 bits into the MP3 file:
I have divulged quite a few bits here, and we would not be unreasonable to assume that this sequence is unique to the MP3 file for "Toxic"---this sequence is not likely to occur in any other file, MP3 or not. Have I broken the law? Have I distributed a bit sequence that is copyrighted by Jive Records?
You would have to be pretty special to be able to hear Britney singing when you stare at those bits, since without interpretation, they are just bits. This sequence is just one of a finite number of possible 240-bit sequences, and there is nothing so special about it. Since Jive Records did not create this particular sequence (recall that, though they created the song, someone else created the MP3 bit sequence), they certainly do not hold copyright over it, so I have not broken the law.
Are you still not convinced? Even if Britney's voice is not jumping out at you, you might claim that I have given you enough information by which you could reconstruct Britney's voice if you wanted to. But how many hoops would you have to jump through to carry out this reconstruction? You would have to strip the bits out of my HTML layout, pack them into a binary file, fit them into a properly-formatted MP3 file, and finally play the file through your speakers---all of this work, just to hear only a chirp of sound (240 bits do not encode very much audio data).
As another example, I could stand on a street corner and recite:
and you could stand there with a pencil and paper, scribbling down ones and zeros. Then you could rush home, pack the bits into an MP3 file, and play them through your speakers. But who would be breaking the law, me or you?
I would simply be reciting a sequence of ones and zeros that are meaningless at face value. You would be the one exerting the effort required to transform those bits into the audio content playing through your speakers. The "unlicensed copy" of the song "Toxic" would exist only in your living room as it plays out of your speakers, and it certainly would not exist on my street corner. Thus, you would be breaking the law, while I would simply be annoying passersby with meaningless gibberish.
Legal InterpretationsIn the real world, copyright law is interpreted and enforced using a different logic from that presented above. Thus, certain binary sequences are deemed illegal to distribute because they are explicit representations of copyrighted works. For example, it is hard to mince words about an MP3 encoding of a copyrighted song. First of all, an MP3 file represents the song's sounds in a direct manner. Second, the method of converting that representation to a presentation is straight-forward, since an algorithm exists that can convert an MP3 file into sound mechanistically without any additional information.
But how far away from direct and explicit representations do we have to go before copyright no longer applies?
Mono files, given that they contain no information from the original Element files, are not explicit representations. The binary data in a Mono file cannot be directly interpreted to produce a presentation of the copyrighted content, so they cannot be seen as representational at all. Mono files take the data a step beyond any explicit representations, and I claim that this step goes far enough to leave copyright behind.
A Monolith ExampleGoing out on a legal limb, here is the Mono file for the "Toxic" MP3 mentioned above:
According to standard practice, Monolith_7D4.wav was used as the Basis file to generate this Mono file, and you can see this reflected in the file name.
How Monolith Munging WorksThe "munging" done by Monolith is surprisingly simple, given the strong statistical properties that result. Monolith combines the Basis and Element files by applying the logical XOR function bit-by-bit. For a pair of bits, the XOR function has the following behavior:
The XOR function is similar to standard addition, except that the carry is ignored in the case of adding 1 and 1.
As a concrete example of XOR applied to a sequence of bits, consider the 8 bits that occur after jumping past the first 160,000 bits in the file britney_spears_-_toxic.mp3. We can XOR these bits, one-by-one, with the corresponding bits from the Monolith_7D4.wav Basis file. This operation results in the corresponding bits from the Mono file that is available for download above.
Statistical AnalysisComparing the Toxic MP3 and the resulting Mono file bit-by-bit, the files have the following correlation coefficient:
For positive correlations, coefficients fall in the range [0,1]. A coefficient of 1 would indicate that two files are identical, while a coefficient of 0 would indicate that the files are completely uncorrelated.
Derivative WorksSome people would argue that, since I used the "Toxic" MP3 to create the Mono file, the Mono file is a derivative work of Spears' song. Distributing derivative works without permission is illegal, since US Copyright Code gives copyright holders rights over all derivatives. For example, Title 17, Chapter 1, Section 106 gives the owner of a copyright the exclusive right:
But what constitutes a derivative work? Certainly, some part of the original work must be present in a recognizable form. The munging done by Monolith leaves no part of the original work in place, so Mono files therefore cannot be counted as derivative works. However, this feature of Mono files is hard to understand fully for sound recordings, since the resulting Mono file is not even a playable sound file at all (the munging completely obliterates the MP3 header information that would be necessary to even interpret the bits as audio). We can better grasp the derivative status of Mono files by applying Monolith's munging to text data.
Here are the first 800 characters from the Bible, which is in the public domain:
Here are the first 800 characters from Thomas Pynchon's novel The Crying of Lot 49, which is still protected by copyright:
What happens when we munge these two bodies of text together with Monolith? Is the result some kind of bizarre Bible/TCOL49 hybrid and therefore a derivative of Pynchon's work? If so, then distributing the derivative would be illegal without Pynchon's permission. Here are the first 800 characters from the resulting Mono file (cleaned up slightly so as not to completely confuse your browser):
None of Pynchon's words are present here in a recognizable form, so this certainly does not qualify as a derivative work. Though the results are more difficult to inspect, the same logic applies when we munge audio files: since no part of the original work is present, a Mono file does not count as a derivative work.
In fact, US Copyright Code places special limitations on what can be counted as a derivative sound recording. Title 17, Chapter 1, Section 114, Part B states:
Since none of the actual sounds from the original file are present in the Mono file (the Mono file is not even a playable audio file), it cannot be seen as a derivative work.
DevelopmentRead the Mono file format specification.
See the SourceForge project for CVS access and bug reports.
CreditsMonolith was created and developed by Jason Rohrer.
Monolith was partially inspired by email exchanges with White Raven.
The graphical user interface for Monolith was programmed using the wxWindows cross-platform toolkit.