Version Control Can Help the Media Win Back Reader Trust

CoinDesk columnist Nic Carter is a partner at Castle Island Ventures, a venture fund based in Cambridge, Mass., that focuses on public blockchains. He is also the co-founder of Coin Metrics, a blockchain analytics startup.

Everyone knows the internet has changed the fundamental business model of journalism. Before the internet age, people subscribed to a physical newspaper that arrived once a day, and they were happy to do it, too. Newsrooms invested in shoe-leather journalists who had the flexibility to pursue a story for months. 

The press wasn’t perfect but it was weighty because subscriptions and advertising meant newsrooms were flush with cash, and long-running investigations could be financed. And the concrete nature of the news gave it a certain finality and weightiness. If a newspaper got something wrong, it had to publish a retraction because the editors couldn’t go back and edit the ink on the page once published. Retractions are embarrassing, so newsrooms generally tried to get it right the first time.

Today, things look very different. Data-rich advertisers like Facebook and Google now account for most ad spending online. Local journalism is feeling the squeeze as treasured smaller outlets get rolled up by private equity firms or simply go out of business. Larger publications remain but margins are being compressed, with the industry consolidating into a handful of winners. Due to the media’s pivot to an online publishing model, editorial standards have been relaxed. Instead of treating published articles as final, news outlets may take the opportunity to change content after publication.

But many outlets take an interactive approach, getting feedback from their readers and editing headlines after the fact and often without any acknowledgement. One particularly egregious example can be found in the Washington Post’s obituary of Islamic State head Abu Bakr Al-Baghdadi, in which they referred to him in the headline as an “austere religious leader,” rather than making his career as a terrorist the focus. In response to criticism, this headline was changed – but no mention of the alteration can be found in the article today. 

Of course, the best phrase to use to describe him is a matter of debate, but the point is the Post shifted between different treatments of the issue at will, with no acknowledgement of its mistake. It’s important for the public to know the Post initially issued a lukewarm obituary for a terrorist, as this materially affects the newspaper’s credibility on these issues.

News outlets should commit to the contents of their article by hashing it along with the headline, and publishing that hash in a costly-to-reverse blockchain like Bitcoin.

These stealth edits are done partially out of convenience, and due to the fast-moving nature of the news today, but also because the changes allow news outlets to more conveniently tailor their coverage to the cultural zeitgeist, effectively becoming malleable to the whims of social media critics. As mistakes are rarely admitted, awkward turns of phrase can be sent down the memory hole, with the public none the wiser. This effectively allows the press to avoid confessing to its mistakes and limits the process of accountability. Unless a keen-eyed commentator archives an article when published, stealth edits can occur without anyone knowing.

So I’d like to propose a very simple solution. At the time of publication, news outlets should commit to the contents of their article by hashing it along with the headline, and publishing that hash in a costly-to-reverse blockchain like Bitcoin. This situates a specific piece of data (the text of the article and the title) in time, localizing it to an hour-or-so window (if using Bitcoin) and makes evident any subsequent changes to the content. If at any point the contents of the article are altered, the hash of the text will no longer match and this will be trivially observable. Of course, the publication can always create a new hash with the edit incorporated, but it won’t be indexed to the time of initial publication. Unless the publication is willing to lie about when articles were published (and this deception would be easy to spot) the system holds up. A publication could timestamp multiple versions of an article and switch them out after the fact if it knew ahead of time which edits it might want to make. So the system isn’t perfectly ironclad, as the publication would have to demonstrate uniqueness as well as chronological localization. There are some initiatives to work on unique proofs of publication like Peter Todd’s Proofmarshal or CommerceBlock’s Mainstay, but ultimately the solution does not lie in tech alone. Readers will have to demand more accountability from the news organisations that they rely on, and these publishers will have to respond in kind.

To demonstrate a primitive version of one such system, I have hashed the exact contents of this article into Bitcoin’s blockchain using Open Timestamps on July 9. (Use the .ots file and the .txt file linked at the end of the article to verify the timestamp.)

Contrary to the occasional utopian proclamations of enthusiasts, public blockchains aren’t magical machines that ensure  all data inserted into them is valid or true. But in this very limited context, they thrive. Timestamps on-chain prove some specific data existed as a specific time. This suffices: We are trying to prove a specific article configuration existed at the time of publication and did not change thereafter. 

I am not the first to propose something similar: for an earlier take see Janine Römer’s discussion of Revision Controlled Journalism. These ideas have existed for a while, and they can easily be instrumentalized now that we have effectively immutable cloud notaries in the form of public blockchains.

I don’t expect this proposal to be enthusiastically accepted by mainstream publications, but I do believe challenger outlets aiming to compete on credibility and accountability will consider embracing the practice. It costs very little, generates a strong signal that the publisher is willing to stand behind the content and demonstrates a commitment to getting it right the first time around. In the ephemeral internet age, the discipline of immutability is worth defending.

Thanks to Peter Todd for his feedback on this article. Find the .ots and .txt files here.