Protecting and Tracking Content through Forensic Watermarking
Content is now more important than ever in the tech sector. Twenty years ago, user-friendliness and technological advance were the main drivers of success for the big tech companies of the world. Google serves the user’s mind by creating a search engine that is easier to use and provides more relevant results to the user. Facebook serves user’s desire to communicate by allowing you to connect with friends in an easier way, and by suggesting more people that are relevant to you.
Now, however, content is the key for the tech giants. All of the four FANGs — the best performing tech stocks, Facebook, Amazon, Netflix and Google, have expanded to the content market. Facebook has been looking to produce its own TV shows; Netflix’ entire business resolves around video content. Amazon launched Prime Video as well as Prime Music. Google spent many millions on acquiring YouTube, recently branching out to a paid subscription model for YouTube Premium.
Piracy is a big threat to content owners, creators and distributors. Original content is a key draw for users and hence controlling the distribution of the content is of prime importance. Through proper protection, content owners can ensure that they are accredited for the distribution of their original assets. Content distributors ensure that users are drawn to their sources, and revenues are placed accordingly to the parties that invested in content creation. This means that original creators are funded and can continue creating content. However, protecting such content is and has always been a difficult task.
Over 2017, illegal music sites logged more than 100 billion visits1. That’s 10x more visits than spotify.com receives2. In January 2019, users made an estimated 363 million visits to sports piracy websites3. If these visits were redirected to ESPN.com, that would have increased its monthly pageviews by 46%4. In fact, London-based consultancy Ovum pegs the loss to industry at 16% of all digital TV and video earnings — that’s an estimated $37.4 billion USD3.
It can be seen easily that pirated streaming video and music sources are bigger than ever, competing for the same users and stealing attention from the original, legal sources of content. This renders content protection more important than ever.
Protection is achieved through what is often called “anti-piracy” measures. Generally, anti-piracy measures fall into two strategies; preventive measures to prevent transgression and reactive measures to detect and respond to piracy after it’s happened. DRM
Preventive measures aim to limit content to subscribers only. Often, such measures are called DRM measures, Digital Rights Management. (Strictly speaking, Digital Rights Management describes the full suite of solutions around content rights, including reactive measures. In common use, however, DRM tends to be used to indicate preventive measures). Popular DRM systems are Widefine (Google), Microsoft PlayReady, Fairplay (Apple) and Adobe.
Preventive measures tend to be technically implemented as a combination between some form of encryption and an internet validation service. The encryption serves to make sure that the content cannot be opened without being validated, i.e. behaves as the lock. The internet validation service determines whether the user has the rights to access the content and serves as the key that opens the lock. In the past, one would often see “serial codes” or similar serve as a key, but these are easily distributed or even reverse-engineered and generated, resulting in uncontrolled piracy. Additionally, by validating access through an internet service, the content owner can make sure a user accesses content only from a reasonable number of devices, and therefore isn’t sharing their access uncontrollably.
DRM has been very successful in managing rights access to online interactive content like multiplayer games. In those areas in which users are online perpetually, DRM can be implemented nearly invisibly and very effectively. However, in those cases were DRM isn’t invisible, it may impact the user directly. This may significantly impact the user friendliness of your product. Just try a search in Medium for “DRM”. Search results have titles like: “If you use DRM, I hate you”5, “DRM & Loathing in Los Angeles”6, “DRM is still here, and it still sucks”7.
Furthermore, DRM is often not a complete solution. It only requires a single pirate to “crack” the DRM system, after which the pirate often can redistribute the content uncontrollably. In fact, this pirated content in some cases is distributed so well and in a user-friendly manner that users of pirated content often don’t even realise they’re consuming illegal content.
Therefore, it is wise to ensure illegal distribution can be detected and responded to. One of the most flexible and powerful approaches to identifying illegal content is forensic watermarking. A forensic watermark is an imperceptible marker that is embedded into target content. The marker itself contains a unique identifier that references metadata relevant to the content. Metadata can include further information regarding the content, e.g. authors and contributors, title, album, genre, etc.
As an example of how a watermark can be used, imagine a copyrighted music track is used as background audio in creative works posted on YouTube. In this case, if the music was watermarked, one would still be able to identify the original metadata in the background audio, such that the original artist can be compensated for usage of their material.
Forensic watermarks can further be used to embed information relating to the distribution and/or subscriber of the content. In this case, a watermark is inserted deeper in the distribution chain. The watermark refers to information about the distributor and/or subscriber of the content. For instance, the subscriber of OTT (Over-The-Top) sports content can be embedded into the content, so that if that sports content is pirated and redistributed, the pirated content can be traced back to the subscriber, who can subsequently be warned or even banned.
To be useful, a good forensic watermark must be robust to attack, whether the attack is intentional or not. For instance, content often changes form and format during redistribution. A simple example of this is transcoding, in which format changes in order to make it more suitable for the next stage of distribution. A video uploaded to TikTok, for instance, is automatically transcoded. Traditional metadata often gets lost during transcoding, but watermarks survive the process. In a rapidly changing digital world, media redistribution is a continuous and unpredictable factor for any content creator. Forensic watermarks provide a strong guarantee that no matter what form the redistribution, the content can be traced back to its origins.
< Previous Page