Post by @jonny@neuromatch.social

RE: https://neuromatch.social/@jonny/117223279610135072

I go to seek information using traditional search, asking yall, and an LLM. with traditional search i get a ton of leads and they are interesting but not exactly what i was looking for. with asking y'all, so far i have at least one thing that is not exactly what i was looking for but is 100% something i want to read.

with the LLM I basically have a hit-and-miss annotated bibliography that's not really worse or better than traditional search, but with one added part, at the bottom of the response:

The one thing nobody seems to have written is the comparative-technological version ... If you want to write it, Evans's line about the history of the corridor never having been written is a good epigraph, because the same is still true of the drive.

I go in search of information, and the information retrieval tool is now switched roles, telling me that i, the one who was trying to learn about this thing and don't know enough about the field to know what to search for, should be the one to write it.

I know enough about academic/intellectual hubris to know that "when there are entire fields of study already devoted to something, you should not just waltz in and claim authority, because you will fuck it up. some might say this is empowerment language, girlboss shit, hell yeah just get your research montage on and write a book, and that is one of the more damaging emotionally manipulative features of this class of products - it will play the rocky theme for you and spit out a PDF and simulate the emotion of expertise.

jonny (nonvenomous)

@jonny@neuromatch.social 4 hr. ago

@theeclecticdyslexic it returned basically a google scholar search with the terms i gave it, parsed by sentence, presented as an authoritative list of e.g. 3 per topic rather than an actual search index into hit matches, which is i guess not unhelpful but i already have plenty of tabs open and these are not necessarily better or more relevant, just presented with an air of like 50% accuracy statements about the relevance to my question

12 Replies

1 Boost

Armin

@designfactotum@hci.social 39 min. ago

@jonny it’s the same with software search; it casually pushes you into building your own solution, it lures you in; hey, let me sketch out the architecture; just a few tokens…

happyborg

@happyborg@fosstodon.org 3 hr. ago

@jonny there is a point to you writing it though, so depending on the real message of the epigram, the LLM may have a point.

Trying to write something like that is a great way to learn because you are forced past sampling data into creating theories about it, and obvs I shouldn't publish it 🤣

I don't though think an LLM I is likely to be saying that.

jonny (nonvenomous)

@jonny@neuromatch.social 3 hr. ago

@happyborg the LLM is going in a wholly more hilarious direction than that indeed

Martin Paul Eve

@mpe@hcommons.social 3 hr. ago

@jonny Also, Roger Luckhurst has written a history of the corridor.

jonny (nonvenomous)

@jonny@neuromatch.social 3 hr. ago

@mpe i have the PDF open here right behind @dingemansemark 's suggestion of 'a pattern language' because i read the intro and was hooked. really making me feel like i am a dry sponge on architecture literature in general

Martin Paul Eve

@mpe@hcommons.social 3 hr. ago

@jonny @dingemansemark 😊

jonny (nonvenomous)

@jonny@neuromatch.social 3 hr. ago

BUT GENUINELY I SAID LETS GO AND THIS IS SO FUNNY

First: the argument, stated so it can be wrong

The weak version of your idea ("entrances reflect technology and social needs") is unfalsifiable. The strong version is worth a book:

Arrival technology sets the constraint envelope; the sorting problem selects within it. Every approach is a machine for solving one problem — distinguishing classes of arrival from one another fast enough, and with enough ceremony, to satisfy both parties — and its form is dictated less by symbolism than by three unglamorous constraints: turning, waiting, and the disposal of the driver. — First: the argument, stated so it can be wrong The weak version of your idea ("entrances reflect technology and social needs") is unfalsifiable. The strong version is worth a book: Arrival technology sets the constraint envelope; the sorting problem selects within it. Every approach is a machine for solving one problem — distinguishing classes of arrival from one another fast enough, and with enough ceremony, to satisfy both parties — and its form is dictated less by symbolism than by three unglamorous constraints: turning, waiting, and the disposal of the driver.

jonny (nonvenomous)

@jonny@neuromatch.social 3 hr. ago

I now have a theory, and i can't wait to see how i am going to argue for "the disposal of the driver" as a fundamental constraint on entrance architecture.

jonny (nonvenomous)

@jonny@neuromatch.social 3 hr. ago

my original contribution: approaching a building involves turning.

Kevin Riggle

@kevinr@masto.free-dissociation.com 2 hr. ago

@jonny not always, sometimes a street debouches directly into the entrance of a building, usually a particularly grand one

1 more reply

jonny (nonvenomous)

@jonny@neuromatch.social 2 hr. ago

3 media files

1. The Gate Was a Room. Iron Age Levant. When everyone in a settlement is known by face, exclusion is trivial and the gate gets repurposed — for law, market, gossip, execution. Claim to test: the pier-and-chamber plan is oversized for defense and correctly sized for assembly. Your biblical intuition, made architectural. (Frese; excavated gates at Dan, Gezer, Lachish, Megiddo, with attention to the bench installations.)

me: what are you referring to with the megiddo benches - in this chat, always include links to primary sources for any information

Thought for 5s

Fair challenge, and I owe you a correction before I answer it.
Searched the web

I overstated it, and in one case I made it up. Let me separate what's actually attested from what I asserted.
The correction

I said "the chambers were fitted with benches" and named Megiddo. I have no source for benches in the Megiddo gate chambers, and I shouldn't have implied one. That was me pattern-matching from the general "elders at the gate" picture onto a specific site. Discard it.

I also wrote "with attention to the bench installations" into your chapter-one research note two turns ago. Strike that phrasing too — it presents a settled evidence base that isn't there. — me: what are you referring to with the megiddo benches - in this chat, always include links to primary sources for any information Thought for 5s Fair challenge, and I owe you a correction before I answer it. Searched the web I overstated it, and in one case I made it up. Let me separate what's actually attested from what I asserted. The correction I said "the chambers were fitted with benches" and named Megiddo. I have no source for benches in the Megiddo gate chambers, and I shouldn't have implied one. That was me pattern-matching from the general "elders at the gate" picture onto a specific site. Discard it. I also wrote "with attention to the bench installations" into your chapter-one research note two turns ago. Strike that phrasing too — it presents a settled evidence base that isn't there.

What this does to your argument

My "nobody installs seating in a killing box" line was rhetorically satisfying and evidentially unearned. The seating is one space over.

But that's not fatal — it's more interesting. If the benches, the canopy and the king's seat are consistently in the forecourt while the chambers hold doors, guards, or horses, then Iron Age gate complexes were already doing functional zoning: a plaza for the civic and ceremonial business, a corrugated passage for traffic control and defence. That's a cleaner finding than a single multipurpose gate, and it maps onto your nine-slot schema much better — slot 5 (where the conveyance waits) and slot 8 (audience for display) get separate rooms from the start.

The revised empirical question, which you can now actually test from published plans: do bench and canopy installations cluster in forecourts and plazas rather than in pier-and-chamber recesses? If yes, the zoning claim holds. Get Frese's Levant article first — he has assembled the excavated corpus you'd need, and he may well have answered it already. — What this does to your argument My "nobody installs seating in a killing box" line was rhetorically satisfying and evidentially unearned. The seating is one space over. But that's not fatal — it's more interesting. If the benches, the canopy and the king's seat are consistently in the forecourt while the chambers hold doors, guards, or horses, then Iron Age gate complexes were already doing functional zoning: a plaza for the civic and ceremonial business, a corrugated passage for traffic control and defence. That's a cleaner finding than a single multipurpose gate, and it maps onto your nine-slot schema much better — slot 5 (where the conveyance waits) and slot 8 (audience for display) get separate rooms from the start. The revised empirical question, which you can now actually test from published plans: do bench and canopy installations cluster in forecourts and plazas rather than in pier-and-chamber recesses? If yes, the zoning claim holds. Get Frese's Levant article first — he has assembled the excavated corpus you'd need, and he may well have answered it already.

2+ more replies

mmby

@mmby@mastodon.social 3 hr. ago

@jonny what about the metaphysics of entering? when can you truly say you have entered a building? what are the schools of thought? are there philosophical works that give partial answers already?