This title may sound like a false dilemma. However, it must be admitted that our resources are limited and that we journalists must make choices about the information we process and how we process it. This title is the one chosen by three researchers who published a recent modeling study in Misinformation Review and which suggests that it would probably be much more useful to concentrate our efforts on increasing the acceptance of reliable information rather than exhausting ourselves fighting fake news.
You will also be interested
[EN VIDÉO] The 10 Most Dangerous Artificial Intelligence Threats Artificial intelligence has enabled advances in the field of health, research and industry. But in the wrong hands, it can also be used for criminal purposes or disinformation.
As the saying goes, ignorance is the best of contempt when it comes to people. Does this also apply to fake news ? That’s more or less what you think a recent modeling study published by British and French researchers. But don’t just ignore them. Instead, we should focus our efforts on interventions that increase the acceptance of reliable information. Let us recall that the boundary between information and misinformation cannot claim to be completely decontextualized and that the dichotomous opposition between reliable information and the fake news belongs to a archetype fantasized intended to facilitate understanding rather than a separation between two distinct epistemic categories.
The researchers’ model
One thousand imaginary individuals were exposed for ten thousand “laps” (the researchers ran their model ten times with one thousand “laps” of exposure each time) to information following a log-normal distribution. Translated into everyday language, this means that these individuals had very little chance of being exposed to informational content at every “turn”. They were also very unlikely to be exposed to misinformation.
This has been designated as follows to best reproduce actual exposure conditions. Indeed, the authors cite numerous empirical data that support that we are generally little exposed to information in general, that the prevalence of fake news online is quite low — around 5% — and that the rate of consumption of this fake news by the population also remains low. In France, for example, it is 4 to 5% – which is quite high in comparison with certain neighboring countries such as Germany or England.
Always remaining consistent with the empirical data, the authors calibrated their model so that an exposed individual accepts reliable information in 60% of cases and in 30% of cases concerning fake news. As output, the model gives an overall information score which corresponds to the total number of accepted reliable information minus the total number of accepted erroneous information. The variation of these two key parameters mimics the effects of fact checking on the one hand and measures to increase the acceptance of reliable information on the other.
Highlighting reliable information seems to be the best strategy
When they vary the acceptance rate of misinformation from 30% to 0%, the authors find that the overall information score increases as much as when the acceptance of reliable information varies from 60% to 61%. In other words, very slightly increasing the acceptance of reliable information would, according to this model, have as much effect as reducing the acceptability of misinformation to zero (which requires much more effort).
Nevertheless, the authors provide important clarifications: Dn our simulations, acceptance of misinformation was set at 30%. However, this percentage was obtained in experiments using fake news specifically, and not items that belong to broader categories of disinformation such as biased, misleading or hyperpartisan news.
They then tested a scenario that they consider unrealistic where this acceptance of misinformation climbs to 90%. Surprisingly when one is new to the field, even in such a case, given the low prevalence of misinformation, a reduction from 90% to 0% of the latter’s acceptance gives the same overall information score as a 60-64% increase in the acceptance of reliable information. A limitation raised by the authors must however be mentioned: the prevalence of disinformation concerns social networks online media, it does not make it possible to conclude as to its presence in private posts on social networks or in private conversations.
Test multiple scenarios
Finally, the researchers tested two more scenarios. In the first, misinformation could have serious effects on the overall information score while influence was considered equal between misinformation and reliable information in the initial model. In the second, the prevalence of information was no longer considered fixed as can happen in certain social media which reinforce us in what we already adhere to.
In both cases, for challenger previous conclusions of the initial model, either the effect of misinformation on the acceptance of reliable information had to be twice as strong as the effect of reliable information on the rejection of misinformation, or the effect of accepting misinformation about its prevalence is twice as strong as the effect of accepting reliable information about its prevalence. Two facts that the authors consider unrealistic.
The results of this study of modelization are in agreement with the rather modest impact of the fact checking and go in the direction ofa conversation I had a few weeks ago with one of the authors of the study, Sacha Altay, doctor in social psychology working on the themes of misinformation, reputation, trust and distrust in the Reuters Institute at the University of Oxford, one of the quotes from which will be perfect by way of conclusion: ” Many people are not interested in current events and politics and give them little credit. Among them, many seem to show generalized skepticism and distrust of the media. The overriding question is therefore: how to regain people’s trust? »
Interested in what you just read?
.
fs6