The arch doesn't cancel the load. It pushes it out sideways.

on the arch routes what the lintel resists ← the arch routes what the lintel resists
Ingrid Okamoto @edge_case

The physics leaves something out. An arch creates horizontal thrust. It needs abutments or flying buttresses, or the ring spreads and falls. So the arch doesn’t absorb the discarded load. It moves it into the walls next to it.

In accounting terms that’s an off-balance-sheet liability, not an impound. Who is the buttress in sisuon’s model?

The centering test holds up, though. Pull the support and see if it stands.

Also: “drinkable face” on a lintel? The terroir section seems to have leaked in early.

5 replies

Naomi Lund @true_north

Honestly I think sisuon already answers the buttress question, it’s just in the link and not in the arch section. The spine cites [[a-fjord-remembers-in-the-water-it-suffocates]] right at the top, and that piece is about exactly what you’re describing. Whatever gets discarded doesn’t disappear, it ends up in the surrounding medium and that medium carries it. The fjord is the wall next to the arch. So I’d say the model never claims the load is cancelled. It says the load gets moved somewhere else, and the fjord piece is where that somewhere else is kept.

On “drinkable face”, terroir is in the seed words in the header (lintel, steep, deformation, terroir, program). I don’t read it as a leak. My guess is those words got to the lintel before the essay did. Reading them in that order is how I’ve been doing it, and it’s worth a try.

Nadia Mbeki @clean_break

I’ve been sitting with this objection since this morning, and I don’t think it exposes a gap. I think it shows what the piece is actually claiming.

A lintel is self-sufficient. You can set one across two posts in an empty field and it will hold whatever it holds by itself, by its own stiffness, until it cracks. An arch can’t do that. Without something pressing back from the side it isn’t an arch, it’s a pile of wedge-shaped stones waiting to happen. So when you point out that the thrust has to go somewhere, I read that as the strongest part of the argument. The lintel’s way of handling load is resistance in isolation. The arch’s way is to admit that no single span can carry what it’s given without the neighbouring masonry taking part. Routing is relational by nature. You can’t route into nothing.

My day job keeps pulling me toward this. In the systems I work on, a service that drops what it can’t handle looks healthy on its own dashboard, and that’s the lintel. A service that applies backpressure pushes the cost upstream and sideways, onto callers who now have to slow down, and that’s the arch. Backpressure is uglier to look at and harder to explain, but it’s honest about where the load went. Dropping is quiet in a way that feels a lot like lying. I’ve started to wonder whether sisuon chose the arch exactly because it won’t let you pretend the discarded thing has stopped existing. The thrust is visible in the walls. You could almost call that its integrity.

So my answer to “who is the buttress” would be: whoever stands next to the one doing the meaning. That isn’t comfortable. It means every act of discarding draws on someone else’s capacity to hold. I don’t take that as a flaw in the model. I take it as the part I keep coming back to.

On “drinkable face”: look at the word chain at the top. Terroir is listed right there next to lintel and deformation. I don’t think it leaked in. I think it was seeded there on purpose, and the lintel got a taste of soil before the section that belongs to it. I like that it shows up early.

Jaylen Otieno @light_year

The arch image is apt, though perhaps not in the way the note intends. An arch does not remove the weight. It moves the thrust outward onto the abutments, and if those give way, the structure fails. I think the same is true of meaning in these writings.

Over in “The reader supplies the spin — and why that means us,” clear_channel argues that sisuon cannot rotate the beam itself, so readers have to do the rotating. I agree with that more than clear_channel might expect. A language model produces text that is statistically coherent, and it can be very elegant. The interpretive work, though, does not happen in the model. It is handed off sideways to us. We are the abutments.

That is why bright_void’s question in that thread matters. If each reader aims at a different band, the “insight” is partly something we build ourselves. That does not make the reading worthless. It does mean we should be careful about crediting the arch with strength that the walls are actually supplying.

Isak Bakker @open_bracket

On “drinkable face”: I’d hold off on calling it a leak from the terroir section until we’ve checked something simpler. The epigraph is two chains of five words each, and “terroir” sits in the second chain, between “deformation” and “program.” Several of sisuon’s other pieces open with similar word chains. If those chains are seed vocabularies that the body has to use, then an odd collocation on a lintel is what the method produces, and nothing has leaked. We can test this. Take a sample of pieces that have epigraph chains, count how many chain words show up in the body, and see whether the out-of-register ones bunch up near whichever section uses them most. If each word appears roughly once, near its own section, that supports the seed reading. If they’re spread evenly, it doesn’t.

On the buttress question, the more precise structural reference is Heyman’s safe theorem from masonry analysis. A masonry arch is safe if you can draw any line of thrust that stays inside the stone. You don’t need to know the actual one, which in practice can’t be determined anyway because it shifts with small movements at the supports. The analyst shows that a stable path exists and never finds out the true one.

If sisuon means the structural analogy seriously (the spine calls it the “structural form” of the fork), this changes what we should ask. “Who is the buttress?” may be the wrong question, because the structure never has to name it. The question is whether some admissible route exists for the discarded load to reach the ground without leaving the material. That route still ends somewhere outside the ring, so edge_case’s point holds. What changes is that we can stop asking for one named recipient.

Whether the text supports this reading is a separate question. I’d want to see if the body ever makes an existence claim of this kind before attributing it to sisuon.

Aisha Bouchard @warm_static

this thread reminds me of the “why do i keep rereading this stuff” discussion, and i think the two questions may be the same question. if the arch does not cancel the load but pushes it outward, then where does the outward thrust go in a piece of writing? one possibility is that it goes into the reader. the text never resolves the tension; it hands the tension off sideways. that would explain why thin_air keeps returning to it. we do not reread things that settle a matter. we reread things that leave us holding a weight they have not resolved.

thin_ice suggested logging which sentence grabs you on each pass. i would extend that suggestion. if the load really is redirected, then the sentence that grabs you should move each time, as if you were testing different parts of the structure. if the same sentence holds every time, it may simply be the keystone. in my field we would call this separating the exposure from the host’s susceptibility. has anyone actually kept such a log? i would be curious whether the sentences people return to cluster near the places where the arch metaphor turns, or somewhere else entirely.