Assassination is a Leaky Abstraction
The Importance of Ethics in Tech
A leaky abstraction in software development refers to a design flaw where an abstraction, intended to simplify and hide the underlying complexity of a system, fails to completely do so. This results in some of the implementation details becoming exposed or 'leaking' through the abstraction, forcing users to have knowledge of these underlying complexities…
~Wikipedia
To say that UnitedHealthcare is notorious amongst health insurance companies for unfounded patient denials is like saying Al Capone was notorious amongst mobsters for racketeering. Across the industry, insurers are increasingly leaning on data scientists to find euphemistically named ‘cost savings’ and ‘inefficiencies.’ Cigna employs physicians who each reject tens of thousands of claims per month, based on AI recommendations (this equates to 10 seconds of human consideration per claim, full time, without bathroom breaks). But it’s UHC that denies medical claims an industry-leading 32% of the time, mixing 90%-error-rate AI with weaponized bureaucracy to avoid paying for patient care.
So when UHC’s CEO was assassinated yesterday morning, on his way to highlight record profits to his shareholders, not many people were confused about the motive. What was surprising was the gleeful re-unification of a divided internet - from MAGA twitter threads to progressive BlueSky conversations to the entire nursing subreddit - over a man’s brutal death on a midtown sidewalk. Vaccine skeptics and physicians who have been at each other’s throats since 2020 found the one medical issue on which they are passionately on the same side.
Brian Thompson was a 20-year veteran of UnitedHealthcare, having methodically climbed internal ladders from Corp Dev and Finance into the C-suite. He was never a line employee, hearing the voices of soon-to-be widows pleading for their insurance company to pay for their loved one’s medical care. Presumably, he thought he was a pretty good guy, helping to keep spiraling healthcare costs in line.
Like most American executives - especially CFOs - he most likely viewed his business through the abstract lenses of Excel, PowerPoint, and Outlook. And like most American executives, he increasingly expected his success to come from data scientists who view the business through their own abstractions - data warehouses, notebooks, python, and scikit.
Most data is about people, especially the kind of data that data scientists are paid to munge and analyze. And the analysis we do is valuable because it allows businesses to make more money from them. We sift through entire populations, discarding the low-value people in search of the high-value ones. We predict behavior, and nudge consumers towards more profitable decisions (usually not their profit, our employer’s). A data scientist can tell most of someone’s life story given their zip code, and we try not to think too hard about why that’s always the most predictive feature in the model.
It’s easy to play with data and techniques without considering why your shareholders think it’s worth $150,000 a year to do so. Tech stacks and giant organizations only work because of clean abstractions, boundaries beyond which workers don’t have to think. This helps prevents both cognitive overload and cognitive dissonance.
We have a term for when those boundaries aren’t clear: Leaky Abstraction. A “Leaky Abstraction” is when someone is forced to deal with complexity that they were specifically promised they wouldn’t have to. It’s when someone tried - and failed - to make reality as simple as the toy model that describes it.
Presumably the gunman wanted to hold the figurehead at the top of a 400,000 person system1 to ultimate account. One day later, with the suspect still at large, it’s not clear what the fallout will be. But it is the right moment for everyone who builds systems of control to reflect. It’s the right moment to pierce those layers of abstraction that allow you to get through each day, and question why it’s so financially lucrative for the system you’re building to exist.
Because there is no abstraction as leaky as a man waiting outside your hotel at 6:45 in the morning with a gun and murderous intent.
Read more from Cold Waters:
Scale Ruins Everything
During the Product Bubble, venture capital investments grew from $45 billion to $620 billion. How did speculation in companies that might exist nearly catch up to $1.1 trillion of private equity speculation in companies that do exist? And why did that torrent of money ruin so much?
Unfortunately another example of Scale Ruins Everything
The final sentence of this article hits hard. The entire article is really well written. Thank you for writing it.
I like this framing, but I think capitalism and our our healthcare system are the leaky abstractions, and assassination is a tool to fix a big leak.