An Intro Video to a YouTube Channel I Never Built
102 slides, a script, and one argument I still make: math is not statistics.
AI-assisted adaptation. The deck and the script are mine, written in June 2025. Claude Code rebuilt them as the click-through below, cleaned up the artwork, and drafted the writing around it. The argument is mine. So is the part where I never shot the video.
In June 2025 I built a 102-slide deck and wrote a script for the first video of a YouTube channel about statistics. Then I did not make the video, or the second one, or the channel. The two files sat in a Drive folder for fifteen months.
The deck was built to be narrated. Its slides do not hold whole thoughts; they hold one line each, stacked up until a screen is full and then wiped. That is a strange thing to read and a fine thing to click, so that is what it is now. One click, one line. Eighty-seven of them. The arrow keys work once the frame has focus, and it goes full screen if you want it to.
Hi, I’m Thommas.
What survived
The argument, mostly unchanged. People say "math and statistics" the way they say "salt and pepper," and then they are surprised when a statistician tells them the answer is it depends. Math is deductive and closed. You are handed premises and you are asked what follows, and what follows is not a matter of opinion. Statistics is none of that. You are handed a mess and asked what is probably true, and probably is carrying the whole sentence.
The Intel and AMD section survived too, and it is the part I would defend hardest. A cost-cutter and an engineer ran two companies in the same market at the same time, and the thing that separated them was not spend. It was what they thought spend was for.
What did not
The script has a stretch in the middle where the dictation software gave up and spliced in an argument about whether LeBron is supposed to tell me that MJ is the guy. I have no memory of saying this. It is in the file, mid-sentence, between a clause about bank accounts and a clause about growth. I cut it, which I mention only because I enjoyed finding it.
I also said twenty years in one place and ten in another, about the same span of time. The deck opens with ten slides of me introducing myself to an audience of nobody. And one of the stock images — the up and down arrows for bank and spend — still had the watermark and the stock ID baked into it, which is the kind of thing you notice on a public site and never notice in a Drive folder. It is redrawn here. So are the mean-median-mode figure and the Intel/AMD chart, the second of which is labelled as illustrative because it always was: it is a shape, not data.
What I could not have written in 2025
Bob Swan was out in 2021. Pat Gelsinger replaced him and was himself ousted in late 2024. In March 2025 Intel appointed Lip-Bu Tan — physics degree, master's in nuclear engineering, an MBA as well, but the engineering came first. A year in, he told an earnings call that Intel was going back to being data driven and engineering driven, and that the conversation a year earlier had been about whether the company would survive, where now it was about how fast they could scale supply to meet demand.
The deck ends with AMD taking share and Intel not understanding why. What I did not know in June 2025 is that the story would keep making my point for another year without me. The cost-cutter shrank the company. The engineer is rebuilding it. Lisa Su, who has run AMD since 2014, is still running AMD.
Intel and AMD details are current as of September 2026.
The one thing the page does better than the video would have
The video would have run about nine minutes and you would have watched it at whatever pace I read it. Here you set the pace, and you can go backwards, which matters in the stretch where three variables contradict each other and the resolution is that the relationships were never linear to begin with. That is a slide you want to sit on. In a video it goes by in four seconds because I was already talking about the next one.
The channel still does not exist. But this was always the argument I wanted to open with, and the format it ended up in is closer to how I actually think than a talking head would have been. Collect the data, estimate the shape, accept that you will land near the top of the mountain rather than exactly on it.
Pretty damn close is the whole discipline.