The metrics for monitoring the health status of economics transcended long ago the professional journalism and became common in the public discussion. It's easy to find terms like "inflation" or "unemployment" in the media. But while several media speak about propaganda or "fake news", we still lack validated and widely adopted metrics to assess the health status of the information sphere. Thanks to technological advances, the science of monitoring information emerged in the last few years. Phenomena such as hate speech or misinformation have been the centre of the attention in the literature, leaving the general context unstudied and blurred in the background. Our study showed that information on critical phenomena might be better understood if analysed in the larger context where they take place.
For example, in the first days of the Covid-19 pandemics or the Ukraine war, citizens' need for information pressured the news supply system to unprecedented and critical levels, which forced news outlets to overflow the audience with news around these hot topics. These kinds of qualitative observations naturally raise questions. Is the news demand that drives the production, or vice-versa? What are the intrinsic timescales of these dynamics? And, perhaps more importantly, what is the role of misinformation in these dynamics? To answer these questions, we analysed the interplay between news supply and demand in Italy from December 2019 to August 2020 for the most trending topic. We also investigated the incidence of questionable sources regarding the keyword ``coronavirus''.
If we think about the infosphere as a market for news, it is natural to analyse its supply-demand behaviour. We learned some peculiar features of this trade with a simple modelling strategy. A linear modelling scheme was effective in almost all cases, which means that supply and demand direct relation can be quantified. In particular, the news production of a given day shows a strong dependence on the previous days' supply. This sort of resistance to change is usually referred to as an ``inertial'' effect. The model also allowed us to measure the ``memory'' of the process and the causality relation. The memory size (2 to 4 days) indicates that it does not matter if a topic was important one week ago, but it does if it was important yesterday. The causality analysis showed that the demand often drags the dynamics while the supply tries to follow it. But how well does it follow? What happens if supply is lagging behind? Here is where questionable sources come into play.
We studied the supply chase for news demand by comparing the general production and the subset of supply coming from questionable sources (annotated as such by professional fact-checkers), focussing only on the ``coronavirus''-related news. The model exposed that questionable sources showed lower inertia than general production. In other words, questionable sources seem to be more reactive to news demand than the average.
To better understand what was happening, we also dived into the semantics of supply and demand. We looked at the shares of the common keywords co-occurring with ``coronavirus'' in supply and demand. We discovered that news supply from questionable sources seems to be better semantically aligned with the demand than general production. For example, the ``coronavirus bulletin'' was the second most queried subtopic in news demand. While it was only ranked 14th in general news supply, the rank in questionable sources was higher (11th). In other words, news demand seems to be chased faster and more precisely by questionable sources than by general production.
This evidence could be awful news: misinformation seems to have an advantage in the news market. On the other hand, the analysis of the mismatch between news general production and demand revealed that this gap seemed to trigger boosts in questionable sources' output. By timely measuring the mismatch, we can highlight vulnerability conditions that could ignite misinformation production and prevent it. For this purpose, we developed an index based on the inertial and semantic mismatch between general news supply and demand. The mismatch index was able to show when ``coronavirus'' questionable sources supply was growing in almost-real time. This index could represent a crucial asset. Currently, misinformation assessment is a slow and sensitive process performed by professional fact-checkers. Instead, the mismatch index could timely detect questionable sources' production spikes without direct verification and, as we have seen, time might be crucial. This index could be a powerful tool to support editors, journalists and fact-checkers and regulatory authorities who could use it to launch public warnings on topics vulnerable to misinformation.
Protecting the health of the information sphere is a fundamental challenge of the Information Age, especially in times of global pandemic and international conflicts. Spontaneous glitches or deliberate attacks might result in disastrous consequences, ranging from traumatic individual and collective misbehaviours to institutional destabilisation. The challenge of avoiding these hazards is particularly complex in democratic contexts where fundamental freedoms of thought, word and association need to be preserved. Furthermore, the information sector is facing an unprecedented economic crisis, so every intervention should also consider the sustainability of the proposed solution. The approach we suggest is focused on preventing misinformation production rather than on its removal. Democratic rights are therefore entirely preserved. Furthermore, the analysis of supply and demand could be helpful for the news industry way beyond the fight against misinformation since it could be strategic to offer a better service to the citizens. Our work's promising results need to extend to other countries, languages, keywords and time frames. Nevertheless, we hope that our contribution could ignite a broad debate in the scientific community about studying the general dynamics of the infosphere and misinformation as part of a larger ecosystem.