Speaking with Objects
A field note, off the numbered sequence
FIELD NOTE
Speaking with Objects
On a hall of pillars, an exam sheet, and the oldest question memory ever asked.
John Callahan · 2026
In seminary, in a Latin class built around Cicero, a teacher once showed us how ancient orators memorized long speeches without notes. You picked a hall you knew well. You walked it in your mind, in a fixed order, pillar to pillar. At each pillar you set an idea — something vivid enough to stick — and when it came time to speak, you walked the same hall again, out loud this time, and each pillar handed you back what you’d left there.
The story behind the technique is older than Cicero. A poet named Simonides was at a banquet when he stepped outside; the roof came down behind him, and every guest inside was crushed beyond recognition. He was the only one who could identify the bodies, because he remembered exactly where each person had been sitting. From that terrible fact he drew a plain principle: order and place are what memory actually holds onto. Give a thing no place, and it has nothing to hold it. The technique we were taught — the pillars, the hall — comes down from that.
There’s a small, honest irony worth including here, since the whole subject is memory: the fullest ancient description of this technique survives in a text that was attributed to Cicero for something like two thousand years, and almost certainly wasn’t his. Even the discipline built entirely around remembering things correctly got its own authorship wrong, for longer than most civilizations last. I don’t think that undermines the lesson. If anything it’s a good demonstration of it — the technique outlived the argument about who taught it to us.
· · ·
I found a smaller version of the same thing on my own, years later, without knowing that’s what it was.
Studying for exams, I’d write my prep notes onto a single sheet, and I’d put each important section in a specific place on the page — not organized by outline or hierarchy, just by where it physically sat. And I could remember what I’d written by remembering where I’d written it. I didn’t think of this as a technique. I thought of it as a habit that happened to work. It took me decades to realize I’d rediscovered, on a sheet of exam paper, exactly what a hall full of pillars had already taught a room of Roman senators.
Later still, in sales presentations, I noticed the same shape again, wearing a different coat: PowerPoint. Fixed sequence. Discrete slides, each one a station. Walk through them in order and the argument arrives in the order you built it to arrive in. A slide deck is a memory palace someone built out of software instead of stone.
A slide deck is a memory palace someone built out of software instead of stone.
But here’s the difference that actually matters, and it took me a long time to see it as a difference at all rather than just a convenience.
The ancient version required you to build the hall yourself. You chose your own pillars. You knew, because you’d placed it there, why this idea belonged at this specific point in the sequence and not the next one. PowerPoint hands almost everyone a hall that’s already built, whether they understand why the sequence works or not. Most people using slides are getting the mechanical benefit of a two-thousand-year-old memory technique without ever having done the part that used to be required to earn it. Two presenters can click through identical decks and be doing entirely different things — one of them walking a hall they built and understand, the other reading someone else’s floor plan out loud, fluently, having never actually walked it themselves.
· · ·
Which is, I think, exactly the question AI puts in front of anyone who speaks or writes for a living now, just asked at a steeper angle than PowerPoint ever asked it.
PowerPoint pre-built the architecture and left you the substance. AI can pre-build the architecture and generate what stands in each room. Structure and content, both arriving finished — something Simonides never had. He built both from nothing, out of a collapsed roof and an exact memory of where each guest had been sitting.
I don’t think this makes the tool the danger, any more than the slide deck was. The danger was never the capability. It’s what happens when nobody has to be Simonides anymore — when a hall arrives fully built, fully furnished, and gets walked out loud by someone who never placed a single thing in it themselves. Used that way, AI is PowerPoint’s failure mode at full depth: something that sounds complete because it has the shape of completeness, without anyone’s own attending ever having happened inside it. Used the other way — as a collaborator in building a hall that’s still actually yours — it’s closer to what a good research assistant or a sharp editor has always been. You still have to know why this idea goes at this pillar. The tool can help you build and test the placement. It cannot want to place it there instead of you.
The honest test hasn’t changed since a poet stood outside a banquet hall in the moment before the roof came down.
Could you walk the hall without notes, and have it still mean the specific thing you meant by putting it there — or would you be reciting someone else’s floor plan, correctly, fluently, having never once walked it yourself? That was Simonides’ question before it was Cicero’s, and it was Cicero’s question, misattributed for two millennia, before it was ever mine. It is, now, exactly as much anyone’s question as it ever was. The pillars just keep changing material. The question standing at each one hasn’t moved.
John Callahan is the author of The Attending series. The series is published by Abendzeit Studios.