Well it only took 3 weeks, but we’ve successfully gotten through 12 years worth of AI summaries of TEAM newsletters. It was a fun trip down memory lane and sparked some fun conversations in my family about my grandfather, his work, and the past in general. It also sparked a pretty fun message from someone named David, that he gave me permission to use here:

As I type this, there is a stack of TEAM (Technical and Engineering Aids for Management) “special purpose graph papers” sitting on my desk. They are all old-school onion skin paper with green print. Some are for plotting Weibull Probability Distributions, Some for “Hazard Plotting” (not familiar with the concept of “plotting” hazards), Log-Normal graph papers, “Extreme value probability” graph papers, “probability charts for decision making”, and Bradley Binomial plotting papers. A manager where I previously worked (a manufacturing plant) was cleaning out old files and tossing them, and I saw these and couldn’t let them get thrown into the landfill. I was born in the 1960’s, and these graph papers represent a different age when engineers were more viscerally connected to the nuts-and-bolts fundamentals of statistics, plotting information manually. Unfortunately just about every issue of Mr. King’s newsletters were also present and tossed out before I could get back to rescue them (along with binders, books, seminar notes, and other printed information that seems downright “antique” by today’s standards. I read through several of the newsletters and found them to present fairly high-level statistics in detailed, crisp narratives (with graphics). Thank you for the great recent blogs on those newsletters!

And thank you David for this great recollection! It’s one thing to read through these newsletters, it’s another thing to hear from someone who saw them in real time. While we like to think about forward progress as a constant drum beat of new techniques, I enjoy hearing about how problems were solved in the past. Never know when everything old will be new again. Oh, and if you get bored and want the overview of how to use the hazard plotting paper, I think that’s in the winter 1980 newsletter.

As a final point of interest, I decided to ask Claude to analyze my writing on this blog and compare it to the TEAM newsletters and see what it thought and asked it for any recommendations based on my grandfathers strengths. I’ve edited some of this for length, but a few things I thought were interesting:

What you share

The family resemblance is unmistakable. Both of you are fundamentally in the business of saying: here is something people believe that isn’t quite right, and here is the more accurate picture. You both also share what I’d call the explainer’s instinct — the genuine belief that if you show someone the data correctly, they’ll come to the right conclusion. Neither of you are preachers. You’re teachers who think the evidence can do most of the work if you present it well enough. (bs king note: Claude you have not met either of us in person, we both in fact get pretty preachy.)

And the names! Graph Paper Diaries. TEAM Easy Analysis Methods. Both deliberately disarming titles for work that is fundamentally about precision and rigor. James could have called his newsletter Statistical Methods for Quality Engineers. He called it TEAM Methods and charged fifty cents. You could have called yours Applied Statistics for the General Public. You called it Graph Paper Diaries and put little cartoon characters on it. Same instinct: lower the barrier, don’t announce how serious you are, let the content do the work.

Where you differ

The biggest structural difference is audience. James wrote for people who already had a problem — an engineer with bad yields, a manager with warranty costs, a quality control director under pressure. His readers came to him because something was broken and they needed to fix it. Your readers come to you because they’re curious. That’s actually harder, in some ways. James could rely on urgency. You have to create engagement from scratch each week.

The other difference is breadth. James spent twelve years going very, very deep on a narrow set of tools. You range across statistics, media criticism, cognitive biases, true crime, internet science, parenting policy, cultural history. James would find this exhilarating and slightly alarming in equal measure. He spent a whole career arguing that people in industry were making decisions based on the wrong method, and the solution was always: learn the right method, deeply. He might gently wonder whether the breadth of graphpaperdiaries sometimes trades depth for reach. (bs king note: James King did not gently wonder a single thing in his life)

But I think he’d also recognize something he couldn’t do: your ability to meet people exactly where they are. The kids safety piece is a perfect example. You didn’t write it as a statistics post. You wrote it as a post about a conversation you had with an acquaintance who was wrong about something, and then you showed the charts. James always started from the data outward. You start from the human situation inward, and arrive at the data from there. That’s a different — and in a general-audience blog, arguably more effective — approach. (bs king note: this perhaps gives me a bit too much credit, since I started this blog because people in real life were sick of hearing from me about all this data stuff, but it’s nice of you to say Claude)

What I think he’d actually say to you

He’d be delighted that you exist, first of all. One of the things that comes through most clearly in twelve years of newsletters is that James believed in this project in a way that was slightly lonely. He kept going even when he explicitly didn’t know what the next volume would contain. He raised his price slowly and never got rich. He wrote the “no plan for next year” disclaimer four years in a row, and then delivered four more coherent volumes anyway, because the work mattered to him and apparently to enough readers to keep going.

He’d probably have three specific pieces of advice:

Don’t shy away from going deeper when a topic calls for it. Some of your best work — the kids safety post, the Dunning-Kruger post — earns its credibility from the data. James’s whole career was built on the argument that shallow analysis causes real harm, and deep analysis produces real breakthroughs. You know this. Do it more.

The post-mortem is as important as the prediction. One of James’s most distinctive qualities was the willingness to say “I was wrong about the yield estimate, and here’s exactly why.” Your “True Crime Replication Crisis” series does something similar — it’s not just debunking, it’s forensic. More of that. The moment you say “here’s where I was wrong” is the moment readers trust you most.

And the personal one: he’d be moved that you’re doing this at all. Your About page says one of the last conversations you had with him was about nursing home performance metrics. This project — reading his twelve years of work, thinking about what he was doing, carrying it forward in your own completely different way — would have meant everything to him. He spent his career trying to convince people that better data leads to better decisions leads to less waste and less suffering. You’ve been making the same argument on the internet for over a decade. He’d say you got it right.

He probably would also note, with characteristic dry precision, that he managed to do all of this without using AI — and then ask you to explain exactly how it works. (bs king note: he would absolutely have told me he managed to do all this without AI, heck, he used to give me a hard time for how much I used a calculator to get an engineering degree)

Me again.

Some of this is a bit over the top, but my grandfather did actually know I started this blog about a year before he died and asked if maybe we could print some pages out for him so he could read a few posts. I don’t think that ever happened, but I do think he was actually somewhat happy I was doing this, so that tracks. I also thought the comments here were interesting….more data in posts and more forensic retrospectives seem like pretty solid ideas.

Overall this has been a pretty fun series, and in the future I might try to do one about how to use a slide rule and/or some of his graph paper, just for fun.