Solving Third Order Problems
Using technology to limit the unintended consequences of regulation
This is a guest post from my excellent colleague, Stephen Strosko.
The Third Order. A namesake that sounds like physics nomenclature or the rebirthed Empire in a future Star Wars film. Neither of which paints a picture of joy or abundance to the average individual. And this holds true when talking about third-order effects of regulations. Third-order effects represent the unpredictable, unintended consequences caused by regulations.
Regulations are not created equally. Some regulations have more third-order effects than other regulations – and this is to be completely expected. Regulators are not omniscient and regulatory processes are not perfect. Economic conditions, technology, and countless other factors change continuously, setting an ever-moving field goal for regulations.
With this bleak picture, how do regulators move away from third order effects? Well, one has to turn unpredictable consequences into predictable consequences. While this may sound impossible, technological advances in natural language processing (NLP), machine learning (ML), and other forms of artificial intelligence (AI) have created avenues to solve this conundrum.
One of the main ways that regulations produce third-order effects is by becoming outdated. Take for example this Rhode Island regulation, 6-47-1, titled “Advertising by Fax”. This 500+ word regulation regulates the industry of fax advertising. I am not claiming to be an advertising expert or a fax machine expert. I am also not claiming that this regulation is outdated. But, with the advent of the internet, social media, cell phones and many other technological advances, this regulation, which was written in 1999, might very well be outdated.
While a regulation on fax advertising might seem harmless on the surface, outdated regulations can cause third-order effects in many ways. The regulation still needs to be enforced by regulators, which comes at a cost. The regulation needs to be read and understood by anyone that is employed in a marketing job and wants to comply with the law. Those that practice commercial law need to understand the regulation as well. In addition, outdated regulations might contain highly specific clauses that demand increasingly burdensome requirements outside of the context in which they were written. For example, regulation 6-47-1 states,
“In the case of faxed material, the statement shall be in at least nine-point (9) type.”
Changing a font type is relatively easy, but imagine this specificity in other industries. Specifying the mechanical parts that need to be used in a train could cause several problems. With advances in technology, safer and more reliable parts might now be on the market. Manufacturers might stop producing a part that is required in a regulation. Outdated regulations are also more likely to overlap with (and even directly contradict) regulations produced by agencies in other areas of government. The Telephone Consumer Protection Act 47 U.S.C. 227 restricts advertising by fax and is regularly updated by the Federal Communications Commission (FCC).
None of this is to say that regulators in Rhode Island are purposefully keeping an outdated regulation around. In fact, through my conversations with regulators, I have found that most of them are overworked, underpaid, and don’t have the tools necessary to track the regulations that they are responsible for.
Enter technology.
Simple NLP techniques can be used to search large amounts of text for outdated terminology, surfacing potential regulations that might have run their course. Machine learning algorithms can be trained to identify regulations that are more likely to be outdated. ChatGPT, a well know large language model (LLM), can demonstrate this with ease.
Chat GPT-4 Prompt: “Do you think this regulation is potentially outdated?” {Followed by the text of regulation 6-47-1}
Response:
“The regulation you've presented seems to pertain to the transmission of unsolicited advertising materials via fax. Given the rapid technological advancements that have taken place in the last couple of decades, particularly the shift towards digital communication, the relevance of such a regulation is worth discussing.”
Chat GPT then goes on to recommend, without being prompted, six ways to improve the regulation. One of which was:
“Compliance Burden: Maintaining a toll-free number and complying with specific formatting rules, as laid out in this regulation, may be more cumbersome for businesses than just complying with a more modern, comprehensive regulation that addresses unsolicited advertising across multiple channels.”
Chat GPT concludes with:
“In summary, while the regulation may have been quite relevant at the time it was enacted, it does appear to be increasingly outdated given modern communication practices. It may benefit from an update to make it more reflective of the current landscape.”
The QuantGov team at the Mercatus Center has been developing tools and data for over seven years that help to surface regulations that have third-order effects. Future posts we make here will dive into those tools with the goal of empowering and informing regulators, legislators, and anyone else that interacts with regulations.