Facts as Annotations
You may have noticed back in December that the Washington Post released a fact checking plugin for Chrome that provides inline context for Trump’s tweets.
A few days later an equivalent Firefox extension was released as well. At the time I looked at the plugin source that was installed (I couldn’t find it on Github) to see how it was gathering facts:
The WaPo browser extension to provide context to Trump’s tweets has a local store of facts https://t.co/gg6RYxYsKj https://t.co/DMsStcr0Ks— Ed Summers ((edsu?)) December 17, 2016
Just a few days ago the Washington Post updated their story to indicate that the extension now will fact check tweets from the POTUS account as well (thanks for the heads up Neil). I took the opportunity a look under the hood again and can see that now it is fetching the facts dynamically from the web from this URL:
Now there are 73 facts about 98 tweets, which is very cool. I put a snapshot I created this afternoon up as a gist if you want to take a look at them pretty-printed.
But it’s not just that there are more facts that’s exciting here. The big improvement in my opinion is that the plugin is loading the facts dynamically. So as new fact checking is performed the plugin can respond in near real time. The plugin doesn’t need to be updated to get the new facts in front of people.
This raises the question of what workflow is producing the facts. I think it would be interesting to know a little bit more about how the facts end up in this JSON data being served up at pbump.net. Presumably there are journalists watching what Trump is tweeting and somehow adding them to a database that is being used to serve up the data. It feels like there could be an opportunity to formalize the data structure for the facts, and bootstrap a mini-ecosystem for the sharing of facts, by trusted authorities.
Having used the Hypothesis Annotation plugin for a few years I can’t help but wonder if it might o share these facts using something like Web Annotations. Web Annotation provides a distributed data model and format for annotating web content. What if there were a plugin that could be configured to display facts from the Washington Post and the New York Times, or any other authority that wants to put in the work, and someone wants to approve? As much as I loathe the idea of alternative facts and how they are being used politically at the moment, I still recognize that facts are based on trust, and trust is fundamentally a social problem.
Truth be told I don’t think using Web Annotation as a technology in itself is going to solve this problem of trust. Been there done that. What’s actually needed is openness about how the architecture underpinning the behavior of fact checking, and letting others participate in it. This obviously isn’t a fully baked thought, but more of a provocation for further thoughts. One small, perhaps risky but practical step forward would be for the Washington Post to publish their plugin on GitHub, and to start a dialogue about what an small ecosystem for fact checking could look like.