Week 2 Discussion
Observations, experiments and simulations, like journalism and news gathering, seek to find an objective truth.
Observations, like live news reporting, are grounded in uncovering the truth of viewable, direct insights into events. From Darwin’s original studies of Galapagos species to Penn's work with CRISPR and today’s advanced telescopes exploring distant galaxies, observation is a critical process for both science and the news. Yet observations can be constrained by the observer’s perspective, biases, and limitations of technology.
Experiments offer observational control and precision, allowing us to isolate variables and find relationships between experiences. They often require extensive and expensive resources, are also prone to potential bias, and social and ethical considerations can constrain their scope. We run many digital experiments in my teams’ product work at NBC News, and the results are always fascinating, especially when they surprise us or disprove a hypothesis. A good experiment is one predicated upon a deep curiosity, meticulously designed to isolate variables (in our case specific product features we are A/B testing), minimize bias where possible, and produce quantifiable results. It incorporates rigorous measurement, and in the scientific context, elements of randomization, and double-blinding where applicable, ensuring findings are robust and generalizable, as with vaccines.
Above: Running election exit poll simulations backstage at NBC News on election night 2024
Simulations bridge the gap between controlled experiments and real-world observations by helping to model the potential futures of complex systems. Schelling’s model of racial segregation demonstrates how we might reveal otherwise inaccessible patterns and predictive outcomes. However, as we also see in political polling and exit data from elections, simulations are only as good as the assumptions they are built upon. Such credibility is often undermined by mistrust or misuse of results to support a partisan, politicized agenda, especially in an era where the truth has (alas) now become a matter of opinion. Baudrillard extrapolates this idea to the point where the simulation supersedes the real experience it replicates, and describes it as setting a dangerous precedent in the field of virtual reality.
Learning from simulations involves recognizing them as tools for hypothesis generation, trend analysis, and decision-making under uncertainty. Climate models, for example, provide invaluable insights into potential outcomes based on varying carbon emission scenarios. We often run news event simulations prior to elections at NBC, where we simulate journalistic responses to potential events which might unfold on the night. It’s more than just rehearsal, it’s observable preparedness for an infinitely branching series of scenarios.
Ultimately, experiments, observations, and simulations are complementary, malleable scientific methods, each contributing to a richer understanding of what it is to seek understanding of the world. Transforming findings into insights, while remaining vigilant and curious about their limitations, we might navigate the complexities of both scientific inquiry and journalism, and safeguard their integrity in an era increasingly fraught with moral challenge.
