Automating Truth: The Case for Crowd-Powered Scientific Investigation in Economics

TL;DR

This paper argues for a crowd-powered, collaborative approach to scientific investigation in economics to improve reproducibility and control amidst the challenges posed by advanced computational tools and data proliferation.

Contribution

It proposes a novel framework leveraging human diversity and crowdsourcing to establish a controlled, reproducible scientific investigation pipeline in economics.

Findings

Computational technology has accelerated knowledge acquisition but risks data unreliability.

Uncontrolled data generation hampers reproducibility and scientific integrity.

Crowd-powered collaboration can restore control and verification in scientific research.

Abstract

Scientific investigation procedures have been evolving to follow an ever-changing cultural landscape, the sophistication of the technology available and an ever-growing knowledge base. This continuous evolution brought investigation practices through distinct historical phases, mostly marked by different types of participants and organization, from individual natural philosophers to science driven by large institutions. There is clear evidence that we are now getting to an age of drastic disruptive change. Increased complexity and mandatory multidisciplinary thinking have moved research from an initial phase of disjoint polymaths into a current phase of widespread uncontrolled use of computational tools and data generation, the "informatics crisis". The use of advanced computational technology for communication and generation of data in large scale without proper controls is compromising our ability to conduct an adequate reproducible investigation, causing a dangerous drift from the scientific method. To counteract this deviation, we advocate the use of a next-generation investigative approach leveraging forces of human diversity, micro-specialized crowds and proper computer-assisted control methods associated with a "pipeline of proof". This paper outlines the impact of advanced computational technology, not only as an accelerator of the rate in which humanity acquires objective knowledge but also as a dangerous side effect as a generator of massive amounts of uncontrolled, unverified and untraceable data and results that cannot be reproduced. We propose an alternative for methods of investigation based on collaboration in large-scale through standard procedures of proof and crowds in building a "collective brain in which neurons are human collaborators".