The Course Project I Built in 2003 Is Gone. The Bet Underneath It Wasn't.
Case of the Week, Min Wu, PhD · ai-public-health.com
In 2003, I was a graduate student in Medical Informatics at UNC Chapel Hill, and I took a Human-Computer Interaction course at the School of Information and Library Science to round out the IT side of my training. The final project was open-ended: build something that lets a person and a system meet around information. I built a small website — a digital library for DICOM, the file standard for medical images. Static HTML. No backend database. Hand-coded structure. The header banner was a 3D affair I made in Photoshop, with the bevels and drop shadows that felt obligatory then.
It was a course project. I expected a grade and not much else.
Then the emails started. People I didn't know — engineers at imaging companies, developers of DICOM tools — wrote asking to be listed. The site had become useful to professionals working in a field I was only adjacent to. I added their tools, fixed the typos they pointed out, and kept going until the semester ended.
After I graduated, I left it on the UNC server and didn't maintain it. At some point it went offline. The artifact is gone. There is no archived version I can link to. The Photoshop banner exists only in my memory.
The Real Problem: The Artifact Is Not the Bet
Years later, working across health informatics, I had language for what had actually happened with that small website.
Most of us evaluate our past work by whether the thing we made still exists. That's a reasonable shorthand, but it conflates two different things: the durability of an artifact and the durability of the judgment that produced it. The artifact is the website, the deck, the model, the workflow. The judgment is the quieter wager underneath — about what will matter, about where a field is heading, about which problems are worth showing up for.
The artifact in 2003 was a static website. The judgment underneath it was a quieter wager: that medical images were going to matter, that they needed to be findable and organized, and that the people building tools for them needed somewhere to point. I didn't articulate any of this at the time. I was a medical informatics student picking a course project. But the choice of what to build — bringing an imaging problem into a course about human-system interaction — expressed an instinct about where both fields were heading.
The site was the throwaway. The instinct was the load-bearing part.
From the Course Project to the Field
Now scale this up.
I have watched professionals — early-career and experienced — judge their own past work by whether the artifact survived. The startup that folded. The report that's no longer cited. The tool that got deprecated. The website that went offline. By that standard, most of what any of us has built in the last twenty years is gone or going. Frameworks deprecate in dog years. Platforms change their terms. The half-life of any specific thing you build is shorter than it used to be.
If the artifact is the only thing carrying your work forward, your work has a short shelf life. That's the position most professionals are in right now without realizing it — running on the durability of things that are not, in fact, durable.
But the bet underneath the artifact has a different timescale entirely. Twenty-three years later, the bet that medical images would matter has come true in a way that's now quantifiable. Image and video data dominate public health, healthcare, and surveillance. The curves are diverging fast: tabular data grows linearly, image data grows exponentially. None of this was visible to me in 2003. But the bet turned out to be correct in a way I couldn't have defended at the time.
The artifact has the shorter half-life. The judgment underneath it is what compounds.
The Question Worth Asking
The question I want to ask about your work is not whether what you built will last. That is the easier, noisier question, and the answer is almost always no. The real question is the one underneath:
What bet are you making with the work you're doing right now — and is that bet likely to age well?
I don't think the answer is a single statement. A bet has a subject (what you're betting on), a timescale (how long before it pays out), and a stake (what you're giving up to make it). All three have to be right together. You can be correct about the subject and wrong about the timescale and lose anyway. You can be right about both and still be staking the wrong things. The bet that lasts is one where all three line up.
For now, the claim I am willing to stand behind is the smaller one: the question worth asking about your work is not whether the artifact will survive. It is whether the judgment underneath the artifact is one you'd still make today, knowing what you know now.
The Observer's Insight
At twenty-five, I could not have told you any of this. I just knew the project was interesting, the emails surprised me, and when the semester ended I moved on to whatever was next. Decades later, I have the vocabulary: I was not building a durable website. I was placing a quiet, undeclared bet on where visual data was heading.
Most professionals do this without noticing. They look back at what they made and judge it by whether it survived; they look forward at what to build and ask how to make it last. Both questions are pointed at the artifact. Neither is pointed at the bet.
The DICOM library was not a piece of work I'd point to as my best. It was a small course project that went offline a long time ago. But the instinct underneath it — that medical images deserved organizing, that they needed somewhere to live, that the people working with them needed a place to meet — turned out to be a bet worth making. That may be the honest picture of where most of us stand with our own past work. The artifacts have mostly disappeared. The bets, if we made any worth making, are still working.
Pick one piece of work from your own past — a project, a paper, a tool, a job — that didn't survive in the form you built it. What was the bet underneath it? Did that bet age well? Reply or leave a comment — I would love to hear your story. I will pick this up in next week's Case.
As a member of the Springer Nature Author Affiliate Program, I may earn a small commission from purchases made through this affiliated link to my book. Support the author by checking out my textbook, Artificial Intelligence in Public Health: https://tidd.ly/4mH9389
