LikeChain: How to manage content quality and distribute content with the blockchain — some first thoughts
With the blockchain, we could create an open, decentralized system to assess the quality of content on the Web. We could further build transparent filter and recommendation algorithms that avoid many problems of today’s content distribution on the Internet and in social media. The system would make it easier to discover good content and make it more difficult to spread fake news. Nevertheless, we could take full account of the individual preferences and interests of users.
[German translation of this article here./Deutsche Version hier.]
Too much content for effective filtering
On the Web, there is a considerable amount of content that is poorly researched, biased, or simply fictitious. On social media, that type of content is widely spread and sometimes achieves some amazing reach. It can therefore affect public opinion, although the content should not have been presented to a large number of people because of its poor quality.
Before the Web, the editorial departments of major newspapers and news magazines were in most cases a quality filter for content. They decided which content was relevant and good enough for publication.
Today, centralized teams could no longer fulfill this function alone. Because of limited resources, like before, they could only take care of the content that is of interest to a broad range of recipients. They would not only filter out bad content, but also almost everything that simply does not fit into the mainstream.
However, there is a lot of interesting and relevant content on the edges of the network. And there is lots of content by good authors, who just have not yet appeared on the radar of the editors. So we need a quality management system that also includes the niche content and that can value the contributions of specialized and new authors and of those who think out-of-the-box. We need a quality management system that can handle an ever growing amount of content on the Web.
Indicators for the quality and relevance of content
There are four main indicators for the quality and relevance of content: Content, Creator, Curator, Context (4C):
* Content refers to the content itself. Are the facts correct? Are the scope and depth of the presentation adequate? Are all relevant aspects considered? Are opinions and facts clearly separated? In summary, does the content adhere to journalistic and scientific quality standards?
* Creator stands for the originator or author of some content. What is the quality of the content she has created in the past? How much content of good quality has she created so far? What is the reach of the content she created? Is the author competent in the area she is dealing with, i.e. how much high-quality content has she already published in a specific subject area?
* Curator is the channel through which a piece of content reaches the reader. What is the reputation of the person or organisation who recommended the content? How much good and relevant content has a source recommended and disseminated in the past? Is the source highly regarded in the reader’s social network?
* Context stands for the environment in which a content appears. What other content appears in the immediate vicinity? How much other content with which reputation links to the content in question? How many sources with what reputation recommend and disseminate the content? (Remark: “Curator” and “Context” cannot be easily separated from each other. Possibly both have to be combined into one factor for concrete implementations of the idea presented here.)
In order to be useful, a quality management and recommendation system for content must be able to evaluate the four above-mentioned criteria (4C).
Content and dissemination as transactions in the blockchain
All basic events in the media landscape can be defined as “transactions” between actors and content. Transactions would be, for example
* An author publishes a content.
* A reader evaluates (for example, “likes”) a content.
* A curator recommends (“shares”) a content.
* A reader subscribes to a curator or to a stream of content by an author (“follows”).
Blockchain-based systems are particularly suitable for storing such transactions securely and transparently in a decentralized way. Thus, all basic processes in a media system could be represented in a blockchain or a similar distributed ledger system.
In a first step, all elements in the content system receive a unique ID: For pieces of content unique hash values can be easily calculated. Authors, curators and readers can be identified by their public signatures in a public/private key system.
So for all four examples above, the elements of the transactions are uniquely defined: authors, readers, and curators can be identified by their public keys. And pieces of content are clearly defined by their hash values.
All transactions in the media system result in a network of actors and actions (see example and illustration below). (Remark: Computer scientists call such a network a “directed graph”: contents and actors are the nodes in the graph, the transactions are the edges of the graph. See also https://en.wikipedia.org/wiki/Directed_graph.)
For example, some transactions:
* User “Goethe” publishes the content “Faust”.
* User “Shakespeare” publishes “Hamlet”.
* User “Jenny” shares the content “Faust”.
* User “Tim” likes the content “Faust”
* User “Tom” likes the content “Hamlet”.
* User “Hannah” likes the content “Faust”.
It is not necessary to distinguish readers from curators or curators from authors. Every actor can be everything at the same time. The are all just users/actors. Actors must be known only by their public key and do not have to determine whether this key is for a specific role in the system, i.e. author, reader, curator, etc.
Reputation for content, authors and curators with the blockchain
In such a system, it is easy to determine a quality value or a reputation value for each piece of content or each actor. To do this, we only have to consider the transactions that affect the element to be evaluated.
In the example above, we could calculate a quality value of “4” for “Faust” by evaluating each “like” with one point and each “share” with two points. “Hamlet”, on the other hand, gets only one quality point, since the content has been “liked” only once. If we stick with this simple mechanism, we should read “Faust” rather than “Hamlet”.
But what if we think that “Tom” has given such outstanding recommendations in the past that his “like” should count ten times more than the “likes” and “shares” of the other actors? Then “Hamlet” would rise to ten points because of the tenfold count of Tom’s “like” and move ahead of “Faust”.
Individual, yet transparent filters are possible
There is an infinite number of ways to evaluate and filter the transactions on the network. Theoretically, each user can configure his own filter and rate those aspects that are most important to him.
On Facebook, for example, this is not possible because Facebook is the sole master of the data and the filter algorithm. Only Facebook knows how its filter algorithm evaluates the content. It is completely intransparent to outsiders how the platform decides who gets to see which content.
Unlike on Facebook, in the system proposed here the complete blockchain is publicly available. Everyone can work with all data and all content. Developers could compete on who will develop the best filter algorithm. And since the data is completely separate from a particular filter, it is easy for any user to switch from one algorithm to another that seems more appropriate.
So it would be possible that a user chooses to weight the actions of women twice as much as those of men. Another user could instruct its filtering algorithm to include only those actors in the evaluation who reach a certain minimum reputation value. Thus, although they work on the same basic data, both users would determine a different content quality and probably get different content recommendations (see illustration).
Integrity and origin of content can be easily determined
With this system, it is easy to determine who was the first to publish a piece of content and who has distributed it since then. This has two advantages.
Firstly, the system makes it easier than ever to identify the true author of a piece of content and to better enforce his rights, for example in the form of royalties or attribution.
Secondly, it gets much easier to identify creators or distributors of fake news or criminal content. Thus, sources that often publish problematic content can be filtered out. And in more serious cases, they can also be prosecuted.
In addition, it is also possible to quickly and reliably determine whether content has been modified since its publication as unique hash values are generated and stored for each piece of content.
By clearly identifying a content through hashes, a block-based system ensures that authors can better enforce their rights, identify foul spots on the network, and also protect the integrity of content from unauthorized change.
Some open questions
Apparently, an open, decentralized system based on a blockchain-related technology is well suited for publishing and disseminating content on the web, and could avoid many of the problems in existing social media networks.
Technically, it would not be a big problem to provide users with the necessary public / private-key key pairs. What’s needed is a user-friendly implementation in client software for reading, publishing and curating content.
However, many architectural issues have to be decided. For example, it is open which form of blockchain or distributed ledger is suitable for a worldwide use with adequate performance.
And it has to be defined whether to store the complete content in the blockchain. On the one hand, the system would be very resistant censorship; on the other hand, existing implementations of distributed ledgers work with rather small amounts of data per transaction. Images or video content would require transaction sizes of megabytes or gigabytes, which are not common in today’s implementations of distributed ledgers.
But the proposed system could also be implemented if only the metadata (users’ public keys, content hashes, etc.) were stored with a reference to the actual content.
However, it will be important to define a mechanism that allows content to be modified after publication, without losing the link to the first publication and the first author. This is important, for example, to be able to correct errors or to supplement new information. A class of change transactions might need to be introduced. This would allow a complete change history to be stored in the blockchain.
Possible implementation
An open-source project with broad-based support would be the most promising environment to implement the proposed system. Support from key stakeholders in the online and media industries would be helpful. And public authorities could support the project through standardization measures, legal regulations, and public procurement contracts.
