The New York Times publication has revealed the successful completion of a trial for a new blockchain-based system to detect misinformation.
According to a June 12 blog post, the publisher built a prototype to provide reliable metadata on various pictures found online. Often, pictures taken in a different location at a different time will be used in connection with unrelated events. Through the platform, readers and social media users will be able to know who took the picture and when.
Technologically, the system used an IBM blockchain platform developed directly by IBM Garage, to ensure that the data remains tamper-proof. The system requires a permissioned enterprise blockchain “to ensure the right members of the network have the appropriate permissions for the metadata,” the blog post explained.
Following user feedback, the New York Times team realized that in addition to knowing when the picture was taken, readers are also interested in knowing where it appeared earlier.
The solution is still a prototype, however, and the publication’s team found several issues in the implementation.
The problem of real world data
The prototype was heavily simplified, and used a single smart contract that automatically approved all new pictures. In the real world, that content must be reviewed and validated by the organization that published the original picture. That can become problematic with some metadata fields, like the description of an event — these can be quite subjective, and the rules are difficult to formalize.
The biggest challenge is matching pictures with their version on the blockchain. Due to the fact that pictures can be modified through software, a computer may not recognize that a slightly altered image is in fact present in the blockchain.
Further advances in computer vision and image recognition seem to be necessary to make this platform a reality.
Finally, accessibility to the blockchain was also noted as a concern. As the team noted, “news organizations with varying financial and technical resources need to be able to participate.” A public blockchain could help solving this, though it may be more difficult to ensure adequate permission levels.
The new prototype follows similar efforts by Italian news agency Ansa, which uses blockchain to “certify” its news stories.
On the other hand, the system developed by the NYT provides a straightforward use case. If implemented, it could help curb a common source of misinformation across both traditional and social media.