Today we’re launching the first phase of some new features to allow the Who Targets Me browser extension to become part of the workflow of any reputable fact checking organisation in the world.
The features will allow all users of Who Targets Me to refer political ads they’re targeted with to a fact checker, and, when the fact check is complete, see the results, which will also be shared with every other Who Targets Me user who saw the same ad.
Context: The difficulties inherent in fact checking political ads
Not every political ad makes specific claims that can be checked. Even when they do, checking them is a complex, nuanced business.
As a result, most democratic countries don’t try to regulate for ’truth’ in political advertising, leaving the heavy lifting to established norms of campaigning behaviour and accountability via the media. For the most part, these norms hold and campaigns that lie do so at a cost to their credibility.
The alternative to norms – regulating truth – risks introducing dangerous biases, heavy-handedness and other forms of manipulation. Just observe how the political parties in the US lean on various institutions of democracy (e.g. voter registration, gerrymandering) to load them in their favour.
We’re very clear that governments getting involved in regulating the content of political advertising is a bad idea. But if governments shouldn’t do it, who should?
It’s well known that Facebook doesn’t fact check the political ads it carries. Though it makes many people angry, there are some understandable reasons for this:
The first is practical. There are a lot of ads. Checking and pre-clearing them all would create an impossible bottleneck (for Facebook, Governments or anyone else). This would be true even without bad actors purposefully clogging the system (as they would surely try to).
The second is more philosophical. Writing neutral rules for what does and doesn’t get fact checked is fraught with complications. Though Facebook has many rules to govern content and behaviour on its platform, it has barely dipped its toe in the water of routinely fact checking political content (a few labels about election-related misinformation in the US apart). In the end, we do actually agree with Nick Clegg and Mark Zuckerberg when they say that Facebook should, as much as possible, avoid being an “arbiter of truth”.
Because we think governments and platforms aren’t the right organisations to manage fact-checking, we plan to work with ones who are – established fact-checkers.
Given the complexities, why even get involved?
Like many things we do with software, introducing a fact-checking ‘layer’ is an experiment to explore the creation of a voluntary, independent system that sits on top of the platform as a “middle layer” between the user and the social network.
By creating this new layer, we point to the possibility of other new layers existing in the future – other tools that do new things to help people understand and contextualise the information they see (read Mike Masnick’s excellent Protocols Not Platforms for more on this).
Similar (though far from identical) ideas include Twitter’s recent announcement that they will allow users to report mis/disinformation, their Birdwatch initiative and Public Editor, the last two of which lean on communities to perform fact checking and contextualisation.
We know the community of Who Targets Me users is curious about political campaigns and cares about good quality information. If we can provide tools that encourage people to request fact checks of ads they’re interested in or suspicious of and, in doing so, help other Who Targets Me users get additional context, that feels like a positive outcome and an experiment worth conducting.
So, to start with, we’re approaching the problem in a limited way:
- We’re going to restrict ourselves to Germany to start, followed by the UK (we are hoping Spain will follow next).
- We’re not planning to try and fact-check every political ad.
- In the next phase of the tool’s development, we aim to leverage some automation provided by Full Fact to pre-identify ads that contain claims that can actually be checked.
- We’re going to rely on our users to indicate which ads they’d like checked.
- We’ll rely on the judgement of our fact checking partners to decide which to check.
- We’ll relay the outcomes of checks to Who Targets Me users verbatim via their personalised results page (https://results.whotargets.me).
The fact checking workflow, in detail
1. Validating checkable claims
- All text (and only the text) from political ads seen by Who Targets Me users is passed to the Full Fact “Claim Detection” API (coming soon).
- The API identifies whether any part of that text contains a checkable claim.
- The API works in a number of languages. Initially, we will only pass it text in English, Spanish and German.
- If the text of an ad is found to contain a checkable claim, we’ll provide a button for a user to “request a fact check” (initially, this will be available for all political ads).
2. Creating a referral pathway to fact checking organisations
- If the user wishes to request a fact check they click the “request a fact check” button next to the ad in the Who Targets Me results panel.
- An entry is created in our database, recording the request. Some ads may get requests from multiple users.
- Periodically, a list of requests and the number of times the request was made will be forwarded to the relevant fact checking partner.
3. Fact checking
- The fact checkers will do their normal work with no interference from us.
- They will choose which, if any, of the referred ads to fact check.
- Working at their own speed, they’ll check claims made in ads.
- They’ll publish their fact checks as normal – usually on their website.
- They’ll share the URL of the fact checks with us.
4. Displaying fact checks via the Who Targets Me results panel
- We’ll upload the fact checks to Who Targets Me’s database.
- All instances of ads whose claims were checked will now be labelled as having been checked.
- Clicking the “View fact check” button next to an ad will open a new tab and show the fact check on the fact checker’s website.
- We’ll publish data about the number of checks requested, how many users request checks and so on.
Misuse: Some things we’ll be watching for…
Every internet problem you propose a solution for creates other sets of problems.
One problem specific to the approach we’re trying is the risk of ‘brigading’ – people from one political persuasion piling on to the ads of their opponents and requesting checks of everything.
We hope that having a “human in the loop” (actual fact checkers and their standard editorial process) mitigates this, but we’ll monitor the tools for this type of abuse.
For this to work, we need your help
Please give the tools a go. If you’re unsure about something a political advertiser is saying, make a referral. It may take a while to get feedback from the fact-checker, or they may choose not to check the ad at all, but your initial action is the vital part of the process.
Furthermore, please do send us any further ideas on how we might improve the tools. We’re in an experimental, listening phase and want to make this work as well as we can. These are complex issues with lots of trade-offs. Let us know where you sit on them and why we should consider or reconsider some of the choices we’ve made in developing this new feature.