The planet, played like a game of history. Compose a world
from its elements — silence the arts, delete steam, plant a medieval sky-vision —
and the outlook meter tells you what to expect before you press Run. A director’s
camera flies you to each city as its moment comes; Kepler, Cyrano and Newton
stroll their home provinces; every run ends with a verdict on the state of
2026. In the good timelines, humanity is multi-planet a century before our
present.
The prompts — verbatim, typos and all
“Okay, this looks quite cool. Can you make one more that looks like Age of
Empires or whatever? I want to see really nice visuals with like an isomorphic
view of a world and I don't know maybe you can make it look like a globe that we
zoom into different parts of so you can add geography and other things also into
it. So that's a new post that I want you to make. I don't know if the resources
are enough to do this on an internet browser, but you're allowed to take as much
time as you want to do this because I have a ton of tokens to be used up before I
lose access to Fable tomorrow.”
“The visual representation of the turning world; it’s not very sims like. No
animations or focus on a city and then zooming out and so on. In some sense we
should be able to parallelise activities too no? And add decision points. I
suppose that would make it more interactive and interesting… how we get to where
we get to. And maybe showing the normal path as the path we are on (maybe the
choices shown tell us which of the options are exactly what we made; this would
map to reality) and alternate paths where deviations happen. A kind of motion
planning problem. And surely one of the worlds must be one in which spaceflight
gets achieved and humans become a spacefaring species by the current date. So
that must also be factored in, because I don't know if any of the current
simulations suggest that we become spacefaring the four nineteen fifty seven. I
would imagine humans achieve something sputnik like maybe in the sixteen hundreds
or seventeen hundreds too, if we actually build things in a linear fashion where
we mature solid propellants and then mature liquid propellants.”
“As I think more about this and my notes that I shared earlier where I believe
that the role of the arts is quite important in determining any outcomes of the
future, I wonder if you can incorporate that into the simulations as well because
the real world, as I show, requires us to have people like Cerano de Bourgeois
and H. G. Wells and whoever else who is writing and envisioning these really
weird futures where we could go into space. And a world without that might
automatically experience some kind of Vision fiction, if you get what I mean.
And, uh, yeah, I think that should be in there. I don't know if vision fiction or
vision creep is the right phrase. You can tell me about that. But, uh, yeah.”
“Is there an option to also study what happens when the fiction elements are
removed? I suppose this is introducing a lot more variables into the thing, so
maybe we want to have removal as a separate option. And I don't know whatever
else is, like, keeping things as an option, if you get what I mean. Or maybe
it's a drag and drop interface, but I don't know if we can drag and drop on a
mobile screen in the same way.”
“really be worth adding an explainer about how these worlds are being
propagated, because it looks like there are at least some kind of weird links
between history and math. So maybe some kind of primer on Cleodynamics at the
bottom. So that might be one question about what this is doing at, like, a
theoretical level. And then you can have another question with a drop down slash
toggle of some kind that explains how this is relevant to what we're doing here
specifically. Because I suspect Cleodynamics is a generic tool, not something
specific to just this kind of study that we're doing here.”
“And I suspect that maybe the ordering of these items should also be
different, because if someone selects exact history, then it should select the
right elements in the stuff that we can actually control. Right? Like, the
variables. And if you select custom or something like that, then you can start
choosing what you want to add and what you don't want to add. Does that make
sense? And maybe you should have an explainer for this dice business because I
still don't get it. but I haven't read the latest version.”
“How about you? Make use of cooler images for this one post. Use the Blender
API. I think for this one post we should prioritize coolness over following
rules.”
“So it would be nice to actually have landmarks specific to the country
instead of just generic images as well. Maybe you could have something like a
Westminster Abbey for the UK in, like, the scenery when we zoom in or something
notable. You know what I mean? Like, in each of those areas. Um, that's one
thought that comes to mind. But, also, just check that the people that you've
drawn as cartoons work nicely in light and dark mode because they're not that
great in my opinion. But, also, I imagine that they would be walking around in
the specifics of the map, not just show up as a small avatar. Yeah.” — followed
by a rethink that simplified the controls: “…I wonder if we can just show,
like, some kind of pessimism meter and optimism meter instead. Is this making
sense? These should be getting adjusted automatically… The thing is, I think
the decision made part just goes away if we are actually just choosing whether
to silence the arts, whether to fund rockets, whether sugar rockets happen, and
so on. Right? … It's possible tomorrow. I'll change my mind again because I
don't know what the right way is, but maybe we keep it simple.”
“I wonder if you could get like an STL or some kind of like special scan of
those buildings and integrate them into these kinds of visuals because I can
see, of course, that people like Apple and Google and all these guys have these
little landmarks of the big cities. Right? Like, you'd get a Notre Dame asset.
You'd get a Westminster asset. You could use... I don't know what the Ajamahal
asset for India. There's so many, like, specific things you could use for each
country. Even though we know that the Taj Mahal is not in Mysore or Westminster
Abbey is not in Cambridge or wherever things are being found.”
“It's interesting that sometimes you have let James Watt live to eighteen
fifty four and all of that. I wonder if it's even possible in this Cleodynamics
universe to retain the persons that get born but have their deaths also in
there because that also affects what gets invented when and where. Uh, but I
think now it starts to get really complicated.”
“The Essone Luck does not seem to be doing much in fact it does nothing at
all to the optimism nadir from what I can tell by visually inspecting
things.”
By Angadh Nanjangud (prompter) & Claude Fable 5 (AI).
Everything here is hand-drawn canvas and SVG — no game engine, no sprites, no
network. The second prompt is why this page has a director’s camera, decision
cards with the real choice badged “what we did,” a reality-vs-you trajectory
strip, and a tech graph that runs past Sputnik to Mars. The third prompt is why
it now has an imaginaries lane — Kepler’s Somnium, Cyrano’s Moon rocket
(M. de Bergerac forgives the spelling), Verne and Wells — and a 1645 decision
card where the censors can silence it. The scholarly term the prompt is reaching
for is Jasanoff’s “sociotechnical imaginaries”; Stephenson calls the SF kind
“hieroglyphs.” The fourth prompt is why the controls are composable
element toggles (tap-chips, not drag-and-drop — which iOS Safari fights):
silence the arts, delete steam, plant the vision, tame the chlorates — in any
combination, with a live summary of the world you have built. “Isomorphic” faithfully preserved; isometric faithfully
delivered. And per the final prompt, this one page bends the site’s
no-dependency rule: the header globe and the seven province terrain plates are
real 3D renders made headlessly with the Blender Python API
(make_blender_assets.py, committed for
reproducibility). The outputs are plain PNGs — the page still executes no
external code. On the STL hunt: Apple’s and Google’s landmark models are
proprietary, and the open scan libraries hide behind account walls (Wikimedia
Commons, searched by mime type, offered Roman funerary stones but no Taj) — so
the landmarks standing in the provinces (the Taj Mahal in “Mysore”, with the
prompter’s geographic blessing; Big Ben; Notre-Dame; the Umayyad minaret; an
iron pagoda; a small verdigris Liberty; the yurts) are modelled from primitives
in the same Blender script and baked into the plates with their shadows.