Thomas Jefferson, Needles and Haystacks

Scott Moskowitz, CEO, Blue Spike, Inc.

When Thomas Jefferson said "information wants to be free," he meant freely accessible.

Part 1: Copyright and Copyleft
Part 2: Internetworkingmycontent
Part 3: Give It Away ... Nah!
Part 4: Recognition + Responsibility = Rewards
Part 5: Value Based on Perception and Trust
Part 6: What of the Future?

The explosive debates surrounding the unfettered exchange of music represent a rare opportunity to consider the intrinsic value of intellectual property. Formats, such as the networking capability of Napster and the file storage of MP3.com, are inexpensive and readily available to any consumer.

The crossroads for the music industry is bound to the complex decision to fully unshackle content from prepackaged media formats, such as CD, and leverage the effectiveness of digital storage and transmission technologies. What remains is the invaluable need for title, or establishing responsibility for individual copies of digitized media signals. I argue for a greater emphasis on the uniqueness of a given work.

First on the table are the assumptions being challenged by the technologies available to artists, rights holders and consumers. Then on to the relative merits of technologies which have traditionally been used for data security. A discussion of secured digital watermarking and related technologies is offered as well to point out the hurdles for the future of the music business. Finally, I suggest a renewed call for closer relationships with consumers, which will enable music and all content businesses to grow.

What are those instruments? They are the unexciting but nonetheless indispensable building blocks of the new infrastructure: e-labels, e-tags, e-money, e-receipts and e-boxes that will animate, and automate, the retailing and auditing functions on the Web will make online commerce infinitely flexible and in most every way more secure than offline commerce. For the automated models of online commerce to be fully realized, vendors will have to find ways to durably embed or attach all of the labels, tags, receipts, subscription data, premiums and the transactional data that vendors and consumers create and share offline.

Digital watermarking is ideally suited to provide much of the infrastructure upon which this instrumentation will be based. Of all the technologies that can be called into play to drive commerce on the Web, digital watermarking has the unique capacity to animate business models and serve the interests of consumers while protecting the rights of copyright holders. It is, for the consumer, the least burdensome and troublesome technology to provide these key commerce mechanisms, placing all the moving parts out of their hands, working invisibly and with the lowest computational demands.

Copyright is a difficult concept to characterize.

A copyright is a means for a creator to establish ownership over a given "work" when created for the first time in tangible form. A neutral third party, the Library of Congress in the United States, acts as the repository for the works. Intended as a means for promoting creativity, a copyright is designed to encourage public accessibility to a creative work by enabling the recognition of property rights over the work. The scope of the copyright, as a legal contract between the creator of the work and the public, is bounded by "fair use."

Essentially in consideration of a publicly sanctioned monopoly of sorts, consumers have rights too, originally for criticism and research but also including such acts as parody.

A seemingly simple concept is further complicated when the copyright is then used to represent the various parties who contribute to a particular work. For instance, a "song" may have a number of bundled rights: a song right for the songwriter, a mechanical right when a copy is made by a device, a performance right, etc. The limiting factor is only how much any particular creator contributed to the work as it is considered in tangible form and how each person is to be compensated. Yet the simple fact is that by and large, consumers care little about the mechanics of copyright. Further, in trying to separate promotion from the act of piracy, the legal, not technical, merits of making music available for creating recognition, sales and an impetus for purchase is maddeningly complex.

For instance, a promotional CD provided to program directors at radio stations, the radio broadcast of the song itself, the shrinkage evident between discrepancies of what was ordered and what was actually manufactured at a pressing plant represent leaks by which any song can be exploited by dedicated pirates. Any unauthorized copy of the music signal can be made available to all points of access to the Internet.

Encouraging distinctions between genuine media signals and pirated media signals is the best course of action for owners of copyrights to ensure financial gain.

Fact: In order to evaluate the potential salability of a song, it must be stored in a freely accessible form. Typically a standard audio workstation with several tracks stores the recorded signals, and any manipulations and mastering represent the final product. Once stored, only subsequent recognition and exchange can attribute value.

Recognition buys the artist a larger initial potential market for her works. It does not guarantee sales. It is also not possible to definitively say with consistency how many copies of any song should be sold given some expectation of sharing by any number of purchasers.

This blurring between copyright sale and successful creation of a fan is the timing hurdle evident to anyone involved in the speculative business of selling media content. When Picasso said people only seek his signature, not his art, his statement relates to the profound value of recognition. In Picasso's prolific career, recognition for recognition's sake created value independent of his contribution to art. One approach for recognition is to give away lower quality versions of work to entice purchases of the same higher quality product. Another is to encourage widespread recognition by encouraging fans and critics alike to record and compare works.

Think for a moment of that oft-cited example of the Grateful Dead, whose hundred or so copyrights represent a small fraction of the number of available and unique performances of any single copyrighted work in their catalog.

Many artists currently spend more time developing an understanding of their audiences and work to encourage an active exchange of ideas, ranging from e-mail vote requested song lists for upcoming concert events to collaborative efforts with other artists who may have similar audiences. They may broaden their appeal by working with artists with separate audiences to enable the creation of new fans. A third approach has gained currency in the software business: seeking to leverage the talents of many individuals for works which require the efforts of many, i.e. the open source movement. Recognition of the work generally precedes individual contributions.

A result of the open source movement has resulted in the Linux operating system, Apache server software. Collaboration by a number of people for the purpose of improving a particular software application would seem to require compensation, but the means of recognition of the work itself, not its creators, forms much of the incentive to keep the work ongoing. That the work can later be monetized into cash, a job offer or an initial public offering, is a matter of the timing of events in how value is created. There are plenty of examples of open source models that have not achieved success; arguably the opening of the source code for the Netscape browser is such an example. The assumption is made that whatever work is first created, the "rights" would be shared amongst the creators who volunteered efforts with foreknowledge that the work would be made freely available to any and all. Intent and market response form the ruler by which success is to be measured in any open source project.

The Internet, MP3, Napster. Simple, inexpensive means for enabling the storage, in bits, and the transmission, in bits per second, of any audio information.

Easy replication of music, easy transmission and no barriers to the exchange. The result: information freely accessible to larger numbers of people. Should be a good opportunity to have a larger potential market. But, not all of the additional audience is interested in paying for the information. Of course, observers claim critics do not buy music. But there cannot possibly be a market full of critics causing the shortfall.

As with any form of commerce, the stepping stone to profitable exchange starts with recognition of a particular good or service. The creator has not abandoned the work and thus the economic value of the work needs to be established in the marketplace.

Because music can be characterized in quality terms (AM radio, MP3, CD), freshness (live, recorded), recorded versions, as well as the uniqueness of a particular performance (Madison Square Garden 1980) there is a lot of room for artists and their managers to get busy identifying profitable market opportunities, and seizing them.

Fact: All industries will sooner or later need to evaluate the open accessibility of value-added information. Increasingly it is information alone which differentiates between the strength of a particular company vis-a-vis its competitors. Design, sales lists, distribution plans are all information valuable to any given corporation. The cost of advertising is likely to rise at a rate much faster than other costs as recognition gives the chance for a sale. For musicians, the market is "easily" defined as those who want to pay for the work of the artists.

The music business just happens to be the first where the value-added information, or music, has been de-coupled from the means of distribution.

Seventy years ago, radio was intended to provide wireless communications points for consumers. The cost benefit versus laying and maintaining wires was deemed opportunistic. The initial stations for radio broadcast were thus intended as wireless communications ports, while the ability for third parties to listen in on a transmission was thought of as an inconvenience. The advantage of radio transmission was economically powerful. Until business interests recognized an even greater economic benefit was afforded by radio transmission in aggregating "listeners," advertising and broadcast industries were incapable of real time performance. It is with the origination of radio that music was de-coupled from recognition. Essentially, live performances could be substituted with radio broadcasts making identification of the performers more difficult but less expensive than with the promotion and sale of phonographs. That the business of selling music was able to relate performance to the sale of prerecorded songs regards the value that is bound in an easily recorded, stored or broadcasted signal.

Recognition continues to be the single most important aspect of selling any good or service. That radio sells CDs is well known but rarely analyzed.

How many stations state the name of the artist and song? How quickly do listeners convert a listening experience into a purchase? The Internet represents similar barriers to the successful marketing and sale of music as all of the previous formats for both storage and transmission of audio signals. That a network can offer both prerecorded and broadcast versions of the same signal is an issue of significance: why is the stored version any different from the streamed version? The difference is that while perfect copies are possible, the breadth and variety of music now available is likely to have more difficulty in finding an audience: stored or streamed. The anticipated debate over rights between two very profitable industries seems to hinge on an old concept. To the chagrin of computer-related interests seeking to "restrict" use of content on digital platforms, with concepts such as digital rights management (DRM) and "trusted systems," media companies maintain consistence on the single issue evident since the advent of consumer electronic devicesresponsibility over copies.

With the introduction of digital compact discs (CD), the audio industry experienced the benefits of the profitable boom of the silicon chip and digital signal processing.

With the introduction of digital compact discs (CD), the audio industry experienced the benefits of the profitable boom of the silicon chip and digital signal processing. Similar to the benefits afforded to the computer industry as a whole but with far less competition in distribution channels, the present reversal of fortune regards the ability to better define and capture rent from copyrighted audio recordings. Labels consolidated to take advantage of the fabulous drops in manufacturing and distribution costs: retail pricing stayed fairly uniform. Digital signal processing also allowed for the quick introduction of a huge number of new acts and music genres. Again, not all the flowers that bloom are roses, but the cheapness of pressing and distributing a CD afforded the artist a much larger potential audience than even the best tour of sidewalks. Yet no major media company is ever able to predict the hits. Music is an acquired taste.

Vanity, fashion, trend, culture are concepts with frequent changes and more frequent mistakes. Risk defies permanent success. Without open access, experimentation is impossible.

Artist and consumer: who made whom? The biology of psychoacoustics (roughly how a human processes audio information) can enable comparisons between different schemes for storage and replication of audio signals, but can never substitute for the human ability to critique. A label is simply a venture capitalist of music: one hit must pay for (at least) nine duds. That we face an increasing number of calls to limit or restrict access to music is simply wrongheaded.

Similarly, the adoption of the Digital Millennium Copyright Act of 1998 (DMCA) is a bold attempt to deny consumers fair use and first sale doctrine rights, which are, essentially, the rights of the consumer, once a purchase has been made, to manipulate a signal or decide what to do with the audio-based intellectual property that has been transferred. The legality of the Act will become a greater question since a number of suits have been filed from all manner of interests in the music business. Broadcasters, software providers and music industry lobbyists have all complimentary and competing interests which are not yet clear without a success model for conducting business in this brave new world.

Wherever value resides, a business model can emerge.

Because it is difficult at best to clearly define how value will reside in a world of instant access to music recordings, title must first exist for the music itself. What is singular is the invaluable importance of establishing responsibility over individual instances of a song, as is the case with any property in a market economy.

The technology of the Internet is the same technology which has enabled the major record labels to become extremely rich in the face of market demands for changes in formats (more singles), price competition (cassettes should be more expensive than CDs based on manufacturing costs) and transparency in the promotion of music (a distribution channel issue).

The other proverbial shoe has dropped anyway. Consumers have found alternative means for enabling their demands to be met.

The file format MP3, the portable devices supporting the format, and network technology which represents a large bazaar for the exchange of any song with geographically dispersed participants. Music is a good whose value is bound in the ease with which it can be shared and exchanged. Unlike computers, which offer increasing performance in information processing, the typically analog device we know as consumers can only process information, such as music, at relatively the same quality level as at the dawn of history. There are problems with the overly simple analysis that music is simply "intellectual property".

Capturing rent is the basis of any market for which property is defined. Recognition of value and the cost of value exchange are the basis of commerce.

The relative importance of distribution strength for packaged goods will continue to drop in the face of heated competition for the attention of consumers. Instead of location, location, location, a better mantra for the Internet Age would go something like: recognition, recognition, recognition. Recognition, and the ability to encourage it through advertising and word-of-mouth, goes hand in hand with responsibility, which makes necessary the technology of digital watermarks.

The Internet Age probably cannot give the consumer too much more of what is already a high quality sound.

However, the potential in microsegmenting marketing efforts, music genre "gurus" filtering vast numbers of titles, providing virtual reality concerts, "aficionado" pricing by sampling rate, personalized optimization of audio signals, real-time listening bars, better than drive-to-Tower Records-download times, etc. is the ultimate challenge to profitably benefit from electronic distribution.

It is difficult to balance arguments for open access with commercial reward. However, technology has made the debate real. Denying the tremendous amount of turmoil which exists in the music business as a result of CD ripping, real time exchange of audio files with networking protocols, and anonymous theft, is no longer an option.

Enabling the individualization of copies of a song with provably secure digital watermarks and a supporting infrastructure for third party authentication of music are likely to be the best way for artists and rights holders to continue to profit from their work.

We originally defined a digital watermark to mean "an ability to establish responsibility for a digital copy." Thus, we first describe the technology of digital watermarking invented by my company, Blue Spike, and then evaluate other competing technologies.

Perhaps best described as a means for binding a "digital signature" to a music signal, in a manner which ensures that attempts at erasure cause audible damage to the song, secure Blue Spike watermarks can be used to tamperproof individual instances of a digital copy of any media content.

Any suspect copy can be checked with the appropriately generated "key," or keys, in the case of multiple rights that were used to embed the signature at the time of purchase. If the information cannot be securely embedded, it is likely that any sacrifice of the signal's quality should be avoided.

Essentially, watermarking is strictly a security technology; the embedded digital watermark information has intrinsic value independent of the audio signal itself. Where the consumer is not bothered by the inaudible tag, the rights holder is able to differentiate between authentic and pirated instances of the song.

A watermarking key is basically a string of cryptographically generated binary digits, or "bits." The key is also a map of how the watermark has been embedded into the target signal. This simple improvement over traditional cryptography is the dramatic difference between digital watermarks and strict digital signature or related encryption technologies. If the key is needed for third party authentication, even by consumers, we use commonly used mathematical tricks to split the key into a key pair. These tricks were discovered in the 1970s and form the basis of public key cryptography.

For watermarking, encoding and encryption is handled by the key, not just encryption.

The private key is used to encode the digital watermark into the music. The public key is used to decode the digital watermark from the music without revealing the private key. The consumer can even authenticate a copy of a song themselves with their public key, just like a purchase receipt.

I originally conceived the technology because of experiences with theft. A ruler with a permanent red marker signature was stolen way back in the eighth grade. Luckily, a friend recovered the stolen item. Though only one of many of the varied experiences of junior high school, I subsequently used a pen knife to etch my initials into CD jewel cases and CDs in a manner not obvious to any casual observer in the early 1980s. The reason: at the time, CDs were scarce and confrontation over the theft of a CD in college is not a preferred means for maintaining control over the property in question!

Extending this means for intentional acts of hiding or "obfuscation" to tagging a media signal is really very straightforward. The parallels with physical good distribution are also uncanny and relevant.

The same way physical media companies seek to monitor materials and distribution of goods by marking those goods with serial numbers (Cartier, for instance) or limiting "authenticity" with "holographic patches" (Levis, for instance), digital watermarks enable artists and rights holders to better track and differentiate between two seemingly "identical" digital recordings. As is commonly the case with highly sought-after goods, such as US$100 bills, Channel bags, and Nike clothing, piracy will exist; but tracking will allow the rights owners to continue to exercise a measure of control over their works.

Moreover, clinical approaches to rights management which ignore the serendipitous nature of recognition, and its big brother, fame, miss the point that works need to be freely accessible with their rights intact to provide beneficial marketing data to musical acts.

Originally, the ability to watermark was defined by me as akin to hiding needles in a haystack. But informationally, the haystack, or song, is not difficult to search quickly. The increases in computational processing speed and access to bandwidth makes the notion of information hiding and other forms of obfuscation more problematic. Simply, it is computationally easy to search vast amounts of information quickly, no matter how cleverly the information is hidden.

There are existing means for enabling permanent changes to be maintained and subsequently detected: we call these "digital signatures."

Since copyrights are intended to last for long periods of time, more secure means for embedding authenticity information, a digital signature for example, into copyrighted media signals, has become invaluable. We can only embed in a finite number of locations in any given song, since the song has a limited amount of data that comprises it.

However, the additional uniqueness required to differentiate between two similar songs can be saved in a watermarking key. This prevents successful attempts at collusion: the comparison of watermarked songs to reveal watermark locations to make erasure easy. The arguments I make here are not unlike arguments being made for any form of security and I will focus on why advanced data security, called cryptography, needs to be bridged with the ability "to hide things in plain view," or steganography.

Digital watermarking is a secure form of steganography.

Steganography has a history at least as old, or older, than cryptography, the art of secret codes. That the application of digital watermarking first happens with music is instructive to present interpretations of security, trust and value. Security needs to be bridged with human perception to enable the many wonderful network technologies and artistic expression possible to be profitable.

With an Information Economy, manufacturing, labor and distribution costs are now commoditized, so that establishing, maintaining and fostering "trust" becomes the most valuable asset.

"Trusted transactions" are the best means to identify and authenticate market participants, value-added information and value adding components in real time. The emergence of a large anonymous marketplace for goods and services demands greater transparency and more realistic security applications. Security in trusted transactions can be easily demonstrated to foster trust by any market participant. In fact, Blue Spike has been shipping a server product that is able to handle all of the complex copyright issues in a manner consistent with market realities and legal remedies. The comparison with other security technologies reveals that only digital watermarking can enable title to be created for individual instances of digital signal copies.

Public Key Infrastructure (PKI) applies strictly to transmission security.

Meaning ensuring that information sent from point A to point B can be said to have integrity, confidence and trust. Non-repudiation of the information sent, a form of guarantee to the recipient of the information, is also accomplished, but only for the information as it is transmitted. Once the information has reached its destination, no further security is enabled. Persistence of identity and authenticity of information, including music, is not possible.

Cryptography, including public key cryptosystems and digital signature applications, have never been designed to establish responsibility over digitized media content such as music. Digital signatures are related to digital watermarks only to the extent that changes are made to target data that can be detected later a means of tamperproofing. The underlying technology of digital signatures simply proves that there is space to hide additional information in practically any set of data.

Digital watermarking technology, however, enables the imperceptible concatenation, literally "permanent binding," essential to the persistent modification of the music or content signal, of a digital signature to media content.

Simply, encrypted music can be decrypted and subsequently copied and redistributed in an uninhibited manner thus, the "digital copy problem" known commonly as "piracy." Digital watermarking leverages the security of cryptographic transmission and combines said security with tamperproofing of a signal such as music. Encryption and PKI cannot solve the "digital copy problem" because the media signal being copied is not considered in the generation of a digital signature or digital certificate.

Only a secure digital watermark which combines cryptographic techniques with data hiding, or steganography, are apt to successfully enable audit trails for media content.

Digital watermarking for images is not the same for audio, since the information hiding exploits psychovisual effects for visual information and psychoacoustic effects for audio information. An interesting argument for consideration is that any scheme to replicate a signal, so-called digital signal processing or communication schemes, such as compact disc or MP3, are logically related to any ciphering or encryption scheme. Could it be that value resides in the ability to share a secret? We know encryption is able to randomize a signal in such a manner that the key used for randomization is a shared secret between the sender and receiver.

When we evaluate the merits of an audio signal, we must observe the signal in "coded" not "ciphered" form. Copyright is not secret sharing, it is value sharing.

Attempts to limit observations of the value of the audio signal can never yield economic benefits to those seeking to increase the value of their work. Value is enhanced in authorized exchange.

Digital Rights Management (DRM) relies on associations between media content and predetermined rules to be governed in the exchange of such content. Rules exist in a particular domain in order to have applicability. Should a containerized or DRM-associated song be compared with a freely available but identical version of that song in an unsecured CD (for instance the industry standard, Red Book audio compact disc)? Any and all rules, including use parameters and restriction, can be simply removed or defeated. What is also problematic is that music must be played to be recognized. Once played, the music signal itself is in freely accessible format. Capturing the signal in the ether is a trivial pursuit.

Any unsecured format can easily be compared with a secure version of the same media content and the difference between the signals is the supposed security. Essentially the notion that secure formats can succeed if unsecured and less expensive versions are concurrently available makes the assumption that consumers will give up something, their fair use and first sale doctrine rights, for nothing.

The problem with DRM, cryptographic containers, "cryptolopes" and related technologies is the over reliance on predetermined rules, not on the market reality that content achieves value when freely accessible. Not free, just freely accessible.

A system for trust, or "trusted system," is inherently command-based and increases the likelihood for systemic failure since media content requires open and accessible means for recognition in order for markets to work correctly. Denying legacy media and legacy file formats (for instance currently available unsecured CD and related file formats such as WAV and AIFF) which are not "containerized," is likely to diminish the returns for any given media work.

Recognition is highly valuable and the time in which to maximize returns shrinks.

Traditionally record companies had 6-8 weeks to sell 80% of a given released title. The time in which sales are similarly reached presently is between 2-3 weeks. This excludes long sellers. Thus the "paparazzi effect" is when media content owners must maximize recognition in order to enable maximized sales. Value cannot be created secretly. Secrets are not maintained in promoting music, thus assumptions that a trusted system or DRM solution can assure security and payment for media content, are poorly constructed arguments. Moreover, each crack in the system requires that all outstanding players be replaced. Systemic risk is higher than the cost for supporting any supposed illusory security over the packaged digital signals.

Comparing performance: a trusted system will likely require ten times as much computation to offer no more security than can be offered by secure watermarking at a cost of ten times the overhead to the system.

Furthermore, models for "superdistribution" simply ignore traditional and well-founded market effects that when value is established for a particular asset, control over that asset is rarely successfully shared with downstream parties. The argument that a digital rights management system can guarantee payment ignores the importance of understanding the ways and means for maintaining the value of a recognized asset, including music. Labels do not share control over copyrights, why will they in the future?

Trust is misplaced in a system which cannot provide concurrent support for existing unsecured media formats (CD, DVD-Video, etc.) or transmission channels (radio, wireless, satellite, etc.) and new formats (electronic distribution).

The issue will continue to be whether DRM or other trusted system technologies can add value to the music to be transacted. There is no security by obscurity and obscurity should not impact the ability to realize value in media content. Consumers should not be bothered by any copyright security or audit trail and should be given the ability to maintain privacy.

The weakness of "all" other watermarking systems is encompassed by a simple test we originally proposed: can your watermark survive itself? Can the watermarked media content be watermarked again and still establish responsibility of the original watermark?

We base this assertion on existing patents and cannot be certain that infringement or alternative technologies will not be introduced. To the extent that secure digital watermarking represents a fundamental intellectual property tool in which to create responsibility for intellectual property, we believe that piracy can be limited as it has in the successful experiences of all media industries, including music.

The key factor will remain creating recognition and converting recognition into a valuable consumer experience. It is not likely that any general standardization of the technology of watermarking is likely to meet all of the various potential applications. Similarly, business models will only be possible to test by extending currently profitable business practices into the electronic domain.

Markets benefit from confidence and trust that is enabled with transparency and liquidity. We expect that the future of music and the successful sale of media content will continue to be a symbiosis between youth and art as an ongoing melody in the making.

©This article originally appeared in dmm v3.3 :: Beyond the Battleground: Get Digital-Music-Smart - or Get Out :: music report