No Silver Bullet in Healthcare
On November 17, 2022, Frederick Phillips Brooks Jr., a pioneer in the computer science world and 1999 recipient of the A.M. Turing Award, passed away. I accidentally stumbled upon Fred in an internal Google Health chat where a colleague gave a nod to his contributions over the years and referenced one of his most influential ones, "No Silver Bullet". A contribution that impacted not only him but very likely most, if not all, of the software development industry.
I spent the first hour of my day today reading through some of the highlights of No Silver Bullet (I'll read it in its entirety at some point), published in 1986, along with some reflections on it 20 years later in the 2008 Object-oriented Programming, Systems, Languages, and Applications (OOPSLA) panelist discussion. My rudimentary summaries of both are below:
No Silver Bullet Takeaways
Fred argues that there are two kinds of tasks in software development: essential and accidental. The former deals with the "complex conceptual structures that compose the abstract software entity" and the latter with those arising from the implementation of the abstract. Further, the major breakthroughs at the time were all focused on tackling the accidental. In turn, Fred postulates:
There is no single development, in either technology or management technique, which by itself promises even one order-of-magnitude improvement within a decade in productivity, in reliability, in simplicity.
Despite Fred's skepticism, he does go on to suggest that the existence of a silver bullet, or rather a solution that comes close to achieving an order-of-magnitude improvement, is those that address the essential complexity in software specifications, design, and testing.
OOPSLA Takeaways
The group of panelists, which included individuals such as David Parnas, Linda Northrop, Dave Thomas, Ricardo Lopez, and Martin Fowler, seemed to suggest that the original arguments laid out by Fred 20 years ago, still persist to this day. The focus on rapid prototyping and frameworks cultivated tackling simple issues at the expense of long-term gains. Modern developers were more concerned about the how and not the what, being certified vs. being competent leading to "certifying incompetence", and creating more work for themselves in search of a silver bullet.
Reflection
As I read through the various ideas and perspectives, a constant stream of similarities to healthcare came to mind:
The promise of electronic health records (EHRs)
The tunneled criticism of a singular group (e.g. vendors, government, healthcare administrators), often by, another singular group
The focus on what can you do vs. what should you do
And of course, artificial intelligence
As I reconcile the information I just read, I was compelled to jot down my thoughts because I think there is also a similar pursuit of a silver bullet in healthcare. Which, with healthcare expenditures as a percentage of GDP growth increasing every year from ~5% in the 1950s to nearly 20% in 2020, I can understand why. Personally, I think the entry of big tech companies like Google, Amazon, Apple, and Microsoft embodies the search for a silver bullet as they seek to disrupt healthcare. To be fair, I'm not at all against this and welcome the change.
What I am concerned about is the notion that a singular group, skillset, or solution will be the hail-mary that healthcare needs.
Data isn't interoperable - create a national EHR! Why not tackle the underlying data exchange solutions?
We need more informaticists - let's teach people to code! Why not learn biomedical informatics or computer science?
We need more clinical decision support - let's create an AI model! Why not ask if AI is even the right solution?
We need to fix clinician burnout - let's reduce clicks! Why not make required documentation a byproduct of the workflow?
We need to reduce healthcare costs - let's create a direct-to-consumer (DTC) business! Why not evaluate the healthcare policies that govern reimbursement for opportunities instead?
The solution to most modern-day healthcare problems, to me, appears extremely short-sighted because learning the nuances of healthcare, government, and tech is the exception, not the norm. This line of thinking is akin to Brian Foote's comment at OOPSLA in that the world is running on "bad code" and that we aren't working on being more effective at writing code, but rather focused on making bad code better.
Concluding thoughts
As per my usual posts, my ideas are really never fully formed, but writing them out helps me to refine my thoughts over time. More importantly, it opens the door to criticism. Personally, I'm not in pursuit of a silver bullet, but I am keenly interested in the discourse around tackling the essential problems in healthcare. Which, isn't usually solved with quick solutions.