The pie chart is mightier than the pen
Can data replace journalism?
A TikTok star will likely be elected as president of Romania in the coming weeks. The US just authorized $1.6 billion of funding to counter Chinese propaganda. And record numbers of bot accounts flood social media platforms, aiming to sway public discourse on just about any political topic. In light of these new, AI-leveraged propaganda techniques, traditional journalism stands little chance.
This is a problem for democracy: since democracies require rational decision-making from the electorate, disruptions of mass media will destabilize democratic nations in the long term. But help may come from an unexpected place: the very same quantitative methods used to build social media bots, can also be used to counter misinformation and establish factuality in the new, post-journalism era of governance.
How can data fill the role of journalism?
In democracies, journalism fills the following five roles:
Communicate issues to the electorate;
Establish how potential policies may affect those issues;
Connect policies to candidates;
Report the actual policies that were pursued by elected candidates;
Discuss how the new policies affected the original issue.
These five points correspond to five aspects of software testing:
Create a formal specification describing what a program should do;
Model how different programs could meet the specification;
Measure the performance of the programs;
Iterate both the specification and program, if needed.
By co-opting the process of software development, political platforms can be strengthened against the algorithmic manipulation of social media.
Example: economic messaging
The most important issue of the 2024 US election was the economy. Neither candidate provided much insight into how their policies would affect the personal finances of Americans, but it is clear that Donald Trump’s messaging resonated more with voters. His firm support of tariffs provided a consistent platform, albeit one with unpredictable results.
How could an ideal economic platform be created, to win over the greatest number of voters?
Following the 5-step software process above, the specification comes first. Microeconomic issues, such as grocery prices, were of particular importance this past election cycle, so minimizing grocery prices would be a sensible goal. From there, mathematical models of different policy options (changes in agricultural subsidies, tax code, etc.) can be created. Once the policy is in place, the measurement, attestation, and iteration steps may be completed.
A political platform that is optimized in such a way is less suceptible to political attack, even if repeated and amplified online. Voters do not have to be convinced of the “trustworthiness” of the policy: the trustworthiness derives from the mathematical methods and numbers used. The voter needs only to be convinced that the methods and numbers are correct, which is a much lower bar to clear.
Retooling campaigns to utilize data
In future election cycles, a successful political campaign will require expert statisticians, data scientists, and AI researchers on staff. Traditional pollsters have failed to make accurate predictions regarding elections - a new AI-based or computational approach could yield better results.
Dedicating staff to monitor social media accounts is a relatively new phenomenon, but it is already behind the curve: automated analysis of online sentiment, a typical application of machine learning, would provide much more useful information. That information can be used to build a platform tuned to the desires of the voters, similar to how live-streamers find their audience by reading messages sent by their chatters.
Today’s candidates for office are woefully unprepared for campaigning in the digital-first environment. Their speeches and debates are subjects of online mockery, and can hardly be said to attract undecided voters. While talking numbers is not a magic fix for a candidate with poor communication skills, it can help establish a clear and consistent platform and divert discussion away from damaging potential gaffes.
Data-driven messaging is also more likely to appeal to Wall Street and Silicon Valley investors, whose support is increasingly important in today’s PAC-driven campaigns.
Concerns about data-driven politics
Quantitative reasoning requires a level of mathematical maturity that many voters do not possess. The US lags behind other developed nations in terms of math ability, and COVID-19 has exacerbated that trend with the incoming cohort of voters.
However, rather than banning platforms like TikTok, social media could be used to distribute short-form, quantitative explanations of public policy. Through repetitive exposure, even undereducated voters can establish an informed knowledge base through repetition. The popularity of scientific YouTube channels attests to the tendency, or even preference, of younger generations to learn new concepts through social media rather than in the classroom.
A secondary concern is that any candidate using quantitative data will lose their advantage, as competing candidates will begin to collect favorable data of their own. However, that is a net win for democracy, as it would raise the standard of discourse to always include numerical evidence, increasing the effectiveness and accountability of political candidates as a whole, even if their data may be biased or inaccurate to some degree.
What can be done?
Establishing a consistent narrative in today’s post-truth environment is near-impossible. The sciences, previously thought of as trustworthy, are increasingly unpopular and perceived as being politically compromised. Facts are unwanted. Spending money to saturate mass media with one’s viewpoint is becoming unsustainable, as the cost of proliferating AI-generated nonsense approaches zero.
The current reality calls for a more fundamental form of reasoning, one that can only be achieved quantitatively. Compared to the subjective concept of “truth”, the foundations of mathematics are much more unassailable.