Medical Slop — A Short History of How We Learned to Look Busy

Documentation didn’t get worse. It just scaled.

Happy new year everyone and welcome to a bunch of new subscribers. For the OGs in the crowd you’ll notice a new name and look to my Substack. There’s an explanation at the bottom of the post. But thanks for staying with me.

Now on to the slop….

---

Slop was Merriam-Webster’s word of the year in 2025. Its importance became evident when a Chrome extension shipped that allows you to evade slop and browse the web like it’s still 2022. For some historical context, the slop era officially began in November 2024 when the quantity of AI-generated articles being published on the web surpassed the quantity of human-written articles.

From FivePercent — The quantity of AI-generated articles being published on the web surpassed the quantity of human-written articles in November 2024

As I thought more about it, slop’s been part of our standard operating procedure for years in medicine. It’s in our dot phrases, copy-forward notes, and WNL notations.

So what counts as medical slop? Where do we find it? And what happens when synthetic authority scales up?

What is slop?

For simplicity, I’m talking about written copy. That is, chunks of text on screens.

It might be helpful to think of medical slop as copy, usually tech-mediated, applied at scale in a medical context where it didn’t exist before. It’s optimized for all the usual suspects: billing, risk mitigation, engagement, or to showcase that we did (or claim to have done) something.

Sometimes we paste or drop it as a template. It can be an empty acronym written with a pen. Sometimes it’s generated by ChatGPT or Claude.

And as we’ve all experienced over the past year or so, AI-generated slop seems strangely out of context and tone deaf. It’s missing voice and lens. I think of it as vacuous in a way that I can’t quite place.

Brian Merchant of Blood in the Machine characterized it nicely:

This agglomeration of content living and dead, marked by a faintly unpleasant aesthetic homogeneity, that evokes in us a weary inability to discern what’s real and what’s not.

When we add authority to slop, it becomes more of a problem.

Why medical slop is more concerning

Physician creators are a different. What we deliver is watermarked with trust and authority. Especially in the public realm. Readers assume that what’s written reflects true expertise. James H Stein, MD did a nice job raising concerns over the ethics of machine-made expertise in Synthetic Prose Is Annoying, but Synthetic Expertise Is a Moral Problem:

When a physician publishes health explanations, the “M.D.” after their name signals training, judgment, and accountability. Readers reasonably assume the analysis reflects real expertise

And medicine is uniquely exposed here because the public can’t tell the difference between me, speaking from 30 years of practice, and a naturopath who knows how to engineer a prompt.

Weak authority to synthetic authority

Before going further I think it’s important to understand what the internet did to authority. Because before everyone had access to expertise, only experts had access to information.

This kept any schmuck from publicly disseminating fake stuff.

Take TV talking heads, for example. In the pre-AI days the only people who could hold their water in front of a camera were those with real expertise in the subject.

Post web, any physician with a modicum of social intelligence could read up on something they’ve never worked with and ape expertise on the national stage. Still, feigning expertise took talent. You had to read and synthesize from original sources and be a talented translator.

But it went a step further with AI.

Synthetic authority started with AI

Now all of us can create dangerously authoritative looking stuff in just minutes. Me included…

With the right prompt, for example, I can create a detailed analysis/opinion on the latest electrophysiology study published in The New England Journal of Medicine. The problem is, I can’t even read an EKG or reliably calculate a p-value.

The consuming public doesn’t know this. They figure I’m a doctor so I must have the ability to critically appraise a study from NEJM.

And if you think this is only about longevity Substacks, you’re missing the bigger picture. Because medicine has been manufacturing slop for decades. Consider any context in healthcare and you’ll find tech-mediated filler. From paper charts to patient message responses, it’s there.

The rise of medical slop

Here are a few places you’ll find medical slop in the wild:

Analog slop

Before EHRs, paper charts were full of pre-fab language: cranial nerves II–XII intact, WNL, no acute distress, etc. This was manilla folder filler that was sometimes true but always raised suspicions. As trainees, this is where we cut our teeth with slop.

And in full transparency, I was an early booster for more insidious forms of analog slop.

During busy nights on call in the early 90’s when we knew a child with ALL and fever was on his way up, we’d begin the H&P on paper. We knew we were going to find. We’d leave blanks and then fill in the details when the child landed on the floor. It was pre-documentation gone wild. An exercise that defied the whole point of taking a history and creating an impression and plan.

In my defense, we were young and full of ourselves...It was 1991 and our killer technology was the first generation of gel pens. We could write faster than those who came before us.

And nothing could stop us.

Slop & paste

Then the EHR came along and evolved as an efficient supply chain for slop.

The dot phrase is the animal by-product filler of the EHR. Fine in moderation, but no one uses them that way. This and other forms of keystroke-generated copy has created the famed Frankenstein notes.

But string a couple of these babies together and you have a ready made H&P in minutes. Sure, it looks like the last 47 admissions you did for reactive airway disease. But it’s an H&P, right?

Ultimately, cut and run paste and forward notes have turned the medical record into a landfill of plausible text that does more to pollute the record than advance the quality of care.

Ambient slop

Ambient AI scribes were supposed to set us free from the keyboard. In 2025 they edged into revenue cycle management and began to craft documentation to support higher complexity billing. But then reality crashed the party. Predictably, this lead to AI-supported rebound denials that have only fueled the slop arms race.

(All together now...) More technology will not fix a broken healthcare system.

And as doctors we created one of the early slop prototypes to counter denials: long, seven page, AI-generated appeal letters. But it turns out the medical directors making the decisions never went to the library to access the three pages of references that came with our letter.

I suspect it won’t be long before we see GPT-generated ‘impressions’ inside the EHR. I put this in quotes because a machine can’t have an impression, only a human can.

And who’s gonna read all of this?

Public slop

The public stuff is the most visible..

• Pre-social SEO slop. The early medical blogosphere was the golden age of hand made, artisinal content optimized for search. The pressure to perform for the Google spiders was staggering and so we had to craft our own performative rubbish.

• LinkedIn. If you want to explore the expanding layer of goo settling on the medical landscape, look no further than LinkedIn. It’s ground-zero for GPT-fueled filler angled to the healthcare vertical. Pro tip: LinkedIn slop is best served with a technoutopian flare that assures readers the best days of doctoring are just around the corner, thanks to our Silicon Valley overlords.

Medical slop is ultimately a human problem

It’s tempting to blame the tools. But copy and paste is a human problem, not a technology problem. Dropping a gigabyte of lab data into the middle of a progress note is a choice. Using documentation to only appear thorough is a sad human failure.

But we’ll figure it out. The past couple of years have been a testing period. We’ve done this with earlier technology.

I do suspect we’ll see a renaissance of voice and lens. We’ll crave human writing that makes a connection. This will be the good part of all of this. The further we value volume over ideas, the more we’ll crave the rare, human stuff.

And as writers we need to raise our game. As Seth Godin suggested, “Technology begins by making old work easier, but then it requires that new work be better.” I suspect this reality will ultimately drive human attention in a world where information is more exponential than ever.

What slop did I miss?

---

The explanation that I promised: So no big thing here. I am just working through what I want to do with my old site, 33 charts. I started this letter with that name, since folks associated it with me. And I knew I wanted to ultimately give this letter a new identity that represents a new chapter. So here we are. I thought the new year would be a good time to turn the page.

Why liminal? The word describes a state of transition. For physicians, it captures where we’re at right now. But the problem is my colleagues (and me) don’t always see the transition as it’s happening. We feel it but we can’t quite explain or place it. And when the tension is unnamed it creates problems. And so the transition may show as burnout, moral distress, or awkward workarounds in our clinical flow.

So The Liminal MD is the same great letter. You shouldn’t notice too much different. For now I am going to center things on this kind of weekly long-form post that is a mashup of cultural criticism and analysis of things in medicine. I want to try to explain how AI, healthcare systems, and other things are mutating medicine. I remain obsessed with how the system is changing doctors and their agency.

I want to do a brief, weekly send of curated stuff I’ve read and found. But I don’t always have enough for substantive letter. So I’ll send that along when I hit a critical threshold. It will be like the old Digital Exhaust format, but more focused and with brief, pointed commentary from me. I am committed to staying quiet unless I have something good to say.

Also, I am reviving my podcast for 2026 under the same name (formerly, FreerangeMD, which scared lots of people away) with a tighter focus on a physician audience. And I’m always looking for provocative angles on the changing healthcare space.

As always, I’m available by email at fox42@me.com. I’m always happy to jump on a call if you want to chat. I’d honestly love to learn about each and every one of you.

As I have always said since my earliest days in the blogosphere: I want to create more value than I take. That’s my commitment to you and I am so appreciative to all of you for being a subscriber.

As always, if you could share these letters with someone who might be interested

Bryan