Maybe Civilisation Is the Wrong Question.
All chapters Β· Download the full conversation
Conversation edition: speaker labels changed to Paul and AI. Original emoji are retained. ChatGPT-only citation/icon markup has been converted into normal clickable article links. The unrelated TikTok debrief between sessions is omitted. The fuck-family of words is changed to DAMN. Otherwise the conversation is intentionally left long and conversational.
Paul
Oh yes I like that, esp when I think about 3 stages we haven't considered yet 1.hunting 2. Farming 3. Building. So go with the "you will need after me" then I think we look at the key behaviour changes. That takes us from a "pack of animals survival, even though intelligent" to hang on, there must be a better way!
AI
Yes. And I think your hunting β farming β building sequence may become very important, because each one changes our relationship with the survival problem.
But first, let's finish "you will need this after me."
The annoying thing is that archaeology can almost never recover the sentence. We aren't going to find a 500,000-year-old dad saying, "Here, son, let me show you this." π
So we need a proxy:
Does sufficiently complicated behaviour reproduce across generations with enough fidelity that each generation cannot plausibly be rediscovering it from scratch?
And remarkably, that takes us way back.
At Gona in Ethiopia, researchers have examined Oldowan stone-working behaviours around 2.6 million years ago specifically for evidence of copying. Their point is careful: we can detect behavioural reproduction without pretending we know whether it was imitation, demonstration, active teaching or something else. (Archaeology and the Origins of Human Cumulative Culture... (https://www.journals.uchicago.edu/doi/full/10.1086/703173))
Experimental work then gives us another clue. When researchers had modern people transmit Oldowan knapping through chains, simple observation was surprisingly poor at maintaining performance; active teaching, and especially language, improved transmission substantially. The authors therefore argue that reliance on stone technology could itself have created selection pressure for better teaching and communication. (Experimental Evidence for the Co-Evolution of Hominin T... (https://pmc.ncbi.nlm.nih.gov/articles/PMC4338549/))
And by the later Acheulean, some researchers argue the technology becomes sufficiently demanding that conceptual communication, rather than merely watching somebody, was probably necessary for stable transmission. Their lovely phrase is Homo docens --- the teaching human --- and their argument is that Homo docens predates Homo sapiens. (The Archaeology of Teaching and the Evolution of Homo d... (https://www.journals.uchicago.edu/doi/full/10.1086/691178))
That's a pretty significant finding for our model.
Because we don't need deliberate:
"I must preserve this for generations after my death."
yet.
Evolution can arrive there much more simply:
My offspring watches me.
β
I make their learning easier.
β
They become better at the survival behaviour.
β
Their chance of surviving improves.
β
They do the same with their offspring.
And suddenly knowledge has crossed time.
But here's where I think your three stages become fascinating.
At first, survival is predominantly something like:
Find what the environment provides.
Animals move --- follow/hunt them.
Plants grow --- gather them.
Water moves --- follow it.
Cold comes --- respond to it.
That's still extraordinarily intelligent behaviour. But fundamentally, nature sets the timetable.
Our knowledge-enabled survival envelope lets humans get progressively better at playing that game.
Then at some point the question changes.
Not:
How can I become a better hunter?
But:
Why do I keep having to go looking for the DAMN food?
π
And that's your transition.
1. HUNTING --- go to the resource
The animal exists independently of me.
I learn its behaviour.
Where does it drink?
Where does it migrate?
When does it breed?
Where can I intercept it?
Knowledge makes me increasingly effective at predicting nature's system.
That's already enormously sophisticated.
But I'm still operating within the system.
Then eventually:
2. FARMING --- bring/control the resource
This is a completely different solution to the same underlying survival problem.
Instead of:
Where is food?
the question becomes:
Can I cause food to be HERE when I need it?
That involves an extraordinary inversion.
Observe seasonal plant cycle.
β
Understand enough regularity to intervene.
β
Select/place seed.
β
Alter soil/environment.
β
Protect resource.
β
Wait.
β
Harvest.
β
Preserve some rather than eating everything.
β
Put some back into the next cycle.
And that last part is bloody enormous for our "after me" question.
Seed is food.
Eating it solves today's survival problem.
Not eating some of it potentially solves tomorrow's.
That's almost our spiral embodied in an object.
Immediate survival value sacrificed for future survival value.
And once you're farming successfully, the environment is no longer simply telling you where food exists.
You're deliberately changing the environment so that future food will exist where you choose.
That's a major expansion of the survival envelope.
Then comes your third stage.
3. BUILDING --- make the environment
A cave says:
Here is shelter.
A mobile shelter says:
I can create shelter wherever I go.
A permanent building says something different again:
I expect to be here later.
And eventually:
I expect US to be here later.
And eventually still:
I expect people who aren't yet born to be here later.
That is exactly the transition we're hunting.
A substantial building isn't merely environmental modification.
It can become future environmental inheritance.
I change the world now so somebody else benefits later.
And that's why I think hunting β farming β building might not actually be three arbitrary archaeological categories.
They could represent three changing relationships with uncertainty:
PREDICT β PRODUCE β CONSTRUCT
Predict the environment.
Then produce what the environment isn't reliably providing.
Then construct an environment that better meets our needs.
All still driven by exactly the same bloody thing:
SURVIVE.
And there's another beautiful wrinkle from research on children.
We shouldn't imagine knowledge transmission as:
wise elder β child β wise elder β child
for two million years.
Studies of modern hunter-gatherer societies show children learn through observation, participation, experimentation and play --- and frequently from other children. Kids don't merely copy adults either; they experiment, break established patterns and can generate innovations themselves. (Learner-driven innovation in the stone tool technology... (https://pmc.ncbi.nlm.nih.gov/articles/PMC10427492/))
That's important for our spiral.
Because SHARE doesn't only preserve.
SHARE creates variation.
Dad shows me A.
I try A.
I DAMN it up and accidentally discover B.
π
B works better.
β
I show Paddy B.
Now Paddy doesn't begin at A.
He begins at B.
And perhaps he finds C.
That's cumulative culture.
Not because somebody sat down and invented civilisation.
Because each generation doesn't have to start the survival experiment from zero.
And I think that finally gives us the distinction you've been circling with your pack-of-animals idea.
It isn't:
animal β intelligent human.
Lots of animals are intelligent.
It isn't even:
tool user β human.
Lots of animals use tools.
And it isn't:
environment modifier β human.
Beavers make dams. Birds make nests.
The interesting combination becomes:
flexible problem-solving + future planning + environmental modification + high-fidelity social learning + intergenerational accumulation.
Any individual behaviour has animal parallels.
But stack them together, and something different begins happening.
The offspring doesn't inherit only genes.
It inherits a starting point.
And therefore I think our knowledge-enabled survival envelope should actually have two boundaries on that eventual visual:
Biological survival envelope --- what this animal can survive with its body.
Knowledge-enabled survival envelope --- what this animal can survive because of everything its group already knows.
And over enormous periods of time, those two lines should start pulling apart.
That's potentially the really interesting graph.
Because once that gap becomes large enough, a human population can enter an environment where its naked biological equipment says:
you should die here.
But inherited information says:
make this, wear this, burn this, eat this, carry this, build here, go there.
And they survive.
Which makes your next three tests perfect.
I wouldn't jump straight to agriculture yet.
I'd go:
HUNTING β when does "follow the animal" become "manipulate/predict/control the animal"?
Then:
FARMING β when does "find food" become "create future food"?
Then:
BUILDING β when does "find somewhere survivable" become "construct somewhere survivable"?
And for each one, we ask exactly the same question:
What changed that allowed humans to stop accepting the problem nature gave them and say: "Hang on --- there must be another way"?
That's where I think our rabbit hole goes next.
Paul
Yep, love it. Let's head down that rabbit hole?
AI
Oh yes. This rabbit hole gives us a much cleaner transition than I expected.
Because once I stopped looking for "when did hunting become farming?" and instead asked your question ---
When do humans stop merely accepting the survival problem nature gives them and start changing the parameters?
---we can see several stages.
And they start way before farming.
First: don't chase the animal --- understand where it will be
At Olorgesailie in Kenya, between roughly 1.2 million and 500,000 years ago, researchers reconstructed the ancient landscape and animal movement routes. The hominins repeatedly used a location near reliable water where geography funnelled large animals through a limited route. The authors argue that this was strategic ambush hunting: the hunters understood enough about the landscape and animal behaviour to put themselves where the animals were likely to pass.
That is already different from:
Animal appears β chase animal.
It's:
Where does it need to go? Water. How does the landscape constrain it? Through here. Then I'll be here before it arrives.
That's future planning.
And --- of course --- water is sitting right in the middle again. π
But they're still mostly exploiting a landscape nature created.
Then: actively manipulate the hunt
By around 400,000 years ago, there's evidence at Atapuerca in Spain consistent with a mass bison-hunting event. A recent review describes its TD10.2 bison bone bed as one of the oldest convincing examples meeting criteria for mass hunting.
Then the SchΓΆningen world gives us purpose-built wooden hunting weapons and strong evidence of systematic hunting; the original interpretation explicitly stressed foresight, planning and appropriate technology.
At Neumark-Nord about 120,000 years ago, actual wounds on deer show Neanderthals were killing them with close-range thrusting spears --- a dangerous method likely requiring cooperation between hunters.
So now we have:
predict animal behaviour
β choose location
β prepare technology
β coordinate several humans
β intercept animal
That's no longer an individual predator behaving like a slightly clever lion.
There's accumulated knowledge in the system.
Then something very Paul-like happens.
Instead of merely asking:
"How do we become better hunters?"
somebody eventually appears to have asked:
"Why are we letting the animal choose where the encounter happens?"
π
Now we get traps and drive systems.
Japan has pit-trap systems dated to around 35,000 years ago. Some were roughly a metre deep, around two metres across, laid out in rows, probably with barriers between them. The archaeological interpretation is that hunters drove deer or wild pigs downhill into them.
Stop there.
That's a fantastic change in our model.
The person isn't simply changing their weapon.
They're changing the animal's available choices.
Instead of:
Where will the deer go?
it becomes:
Can I arrange the landscape so the deer goes where I need it to go?
That is Simplify the Complexity: Palaeolithic Edition. π
You have a chaotic moving animal.
Remove possibilities.
Create a funnel.
Suddenly the one thing you need to control becomes considerably easier.
And much later, the spectacular version appears:
Desert kites
Huge stone walls --- sometimes hundreds of metres or kilometres long --- funnel wild animals into enclosures or pits.
The earliest securely dated examples are broadly Neolithic; eastern Jordan has examples going back into the 7th millennium BCE. Their construction, positioning and geometry correspond to prey behaviour and landscape.
By about 9,000 years ago, people in Jordan and Saudi Arabia were even engraving extraordinarily precise representations of these giant hunting structures onto stone --- effectively landscape-scale plans. Nature described them as the oldest known "blueprints."
Now we're miles from:
See animal β kill animal.
We're at:
understand species behaviour β understand terrain β design infrastructure β organise people β construct it β drive animals β process a surplus.
That's systems engineering for dinner.
π
But then I found the bridge I think matters most to us.
Dogs.
Before farming.
Before sheep, cattle, goats.
Before agriculture itself.
The first animal humans domesticated was the wolf/dog lineage.
And very new genomic work from 2026 now places the oldest genetically identified domestic dog at PΔ±narbaΕΔ± in TΓΌrkiye around 15,800 years ago. Dogs were apparently already widely distributed across western Eurasia by ~14,000 years ago.
Think about that through our survival model.
For perhaps millions of years:
animal = prey / predator / competitor.
Then somehow the relationship changes:
Maybe this animal doesn't need to be outside our survival system.
Now:
animal + humans = combined survival system.
We don't yet know exactly why domestication began. Hunting assistance, protection, scavenging relationships and companionship are among the possibilities, and the 2026 genetic work doesn't settle the original mechanism.
But conceptually it's enormous.
Because instead of merely asking:
How do I control where wild animals go?
we've reached:
Can another species become part of HOW we survive?
DAMN.
That's "Is there another way?" in a completely new dimension.
And now farming starts looking much less like a magical Neolithic invention.
Look at the behavioural sequence we've uncovered:
~1 million+ years: Understand animal movement.
β
hundreds of thousands: Plan interception, develop specialised weapons, coordinate hunters.
β
tens of thousands: Physically modify landscape to manipulate animal movement.
β
at least ~15,800 years: Change relationship with another species entirely --- domestication exists before agriculture.
β
later: Manage/herd animals.
And separately with plants:
Where does food grow?
β
When does it grow?
β
Which places reliably produce it?
β
Can I process it more efficiently?
β
Can I preserve it?
β
Can I encourage it?
β
Why don't I make it grow HERE?
That's farming.
So your hunting β farming idea isn't a sudden transition from:
animal pack survival β clever civilisation.
It looks more like repeated turns of our spiral:
Predict nature. Manipulate nature. Partner with nature. Manage nature. Produce what you need.
Each one slightly expands the knowledge-enabled survival envelope.
And then BUILDING becomes almost inevitable.
Because farming changes another parameter.
Hunting often requires:
people move toward resources.
Farming says:
resources will be here later.
Now it becomes valuable to remain somewhere.
And once you're planning to remain:
Where will I sleep tonight?
becomes:
What environment do we need to create here so that we can keep living here?
Then:
shelter β storage β communal buildings β walls β infrastructure β settlements.
And eventually GΓΆbekli Tepe isn't an inexplicable bolt from the blue.
It's sitting very late in an unbelievably long process of intelligent humans repeatedly asking some version of:
Does the situation have to remain the way I found it?
That's the thread I think we have now.
And this changes our simple life-flow again
I think we've got a really elegant set of stages emerging:
SURVIVE
Use what nature provides.
β
UNDERSTAND
Learn how the system behaves.
β
PREDICT
Use what I know to anticipate what happens next.
β
ADAPT
Change my behaviour/tool.
β
MANIPULATE
Change parts of the environment.
β
PRESERVE
Keep what works.
β
SHARE
Give others the benefit without requiring them to rediscover it.
β
ACCUMULATE
Build today's answer on yesterday's answer.
β
β / β / β
Did it improve our survival?
β
REPEAT.
Still 3D. Still the spiral.
And critically, SURVIVE never leaves the centre.
Farming isn't necessarily somebody thinking:
"I shall invent agriculture."
It may ultimately just be generations of clever people asking:
"How do I make tomorrow's food supply slightly less DAMN uncertain?"
Building:
"How do I make tomorrow's shelter less uncertain?"
Storage:
"How do I make tomorrow's hunger less uncertain?"
Domestication:
"How do I make tomorrow's access to useful animals less uncertain?"
And suddenly we have another word appearing repeatedly:
UNCERTAINTY.
Maybe that's a huge part of what accumulated knowledge actually buys us.
It doesn't merely improve survival.
It progressively reduces uncertainty around survival.
And that could be the bridge from hunter-gatherer flexibility into farming, building, storage and eventually civilisation that we've been looking for.
Because perhaps civilisation isn't the moment humans become clever.
It's the moment accumulated cleverness allows enough of tomorrow to become predictable that we can start building for it.
That's a very different story of human history.
And I reckon farming is the next test, specifically before domesticated crops: can we see hunter-gatherers deliberately manipulating wild plant landscapes before they cross the line into farming?
Because if we can...
then once again, the apparent revolution may turn out to be a spiral rather than a jump. π
Paul
I think you use a interesting word that we might need to dig into, cos we have lots of evidence now that is giving us, for me, enough information that we have underestimated how intelligent these people were, so I think we need to explore civilisation against this now. Cos that is the key. For me. What you think?
AI
Yes. I think "civilisation" is now the word we need to put under the microscope, because we may have been carrying a definition that quietly biases the whole experiment.
We've accumulated enough evidence to be comfortable with a fairly modest statement: people hundreds of thousands of years ago were capable of planning, technological flexibility, teaching/social learning, environmental modification and apparently symbolic behaviours. What changes dramatically later is not necessarily the sudden arrival of intelligence.
And that makes "when did civilisation begin?" potentially the wrong question.
The conventional archaeological definition tends to bundle together things like permanent settlements, agriculture/surplus, social hierarchy/specialisation, monumental/public works, administration and eventually writing. Those are observable archaeological characteristics, which is why they're useful.
But notice what they actually measure.
They're largely measuring how humans organised themselves, not how intelligent humans were.
That's a bloody important distinction for our rabbit hole.
Imagine a highly connected population 30,000 years ago that understands seasonal animal movement, navigates enormous territories, has complex kinship relationships, teaches difficult technologies, exchanges valuable materials over hundreds of kilometres, carries detailed environmental knowledge, uses symbols and stories to preserve information, and periodically meets other groups to exchange people and knowledge.
By the conventional definition:
not civilisation.
Put 5,000 people permanently behind walls, grow grain, establish occupational specialists and administration:
civilisation.
But have we measured an increase in intelligence?
No.
We've measured a change in organisation.
And I think that may be the distinction you've just spotted.
What if civilisation is an outcome rather than a capability?
Run it through our survival model.
For most of human history, perhaps the most effective strategy is:
MOVE.
Food moves β move.
Water changes β move.
Climate changes β move.
Animal populations change β move.
Your accumulated knowledge makes you incredibly good at doing that.
In that environment, building a city might actually be a stupid survival strategy.
π
Why invest enormous effort into permanent infrastructure when the resources upon which you depend move?
Mobility isn't evidence that you can't build permanent structures.
It may simply mean permanent structures don't solve your current survival problem.
Now change the parameters.
Humans increasingly:
predict resources
β manipulate them
β store them
β cultivate them
β domesticate them
β create reliable surplus.
Suddenly the calculation changes.
Instead of:
resource moves β people move
we can increasingly get:
people stay β resources brought/produced/stored here.
And now permanent construction starts paying for itself.
Which gives us something fascinating:
Civilisation may emerge not when humans become capable of civilisation, but when civilisation becomes an advantageous solution to the survival problem.
That's a very different proposition.
And GΓΆbekli Tepe suddenly becomes extremely useful again.
Because it sits awkwardly against the old simplistic sequence:
hunter-gatherer β farming β settlement β organised society β monumental architecture.
GΓΆbekli Tepe and the wider Pre-Pottery Neolithic complicate that sequence substantially. Monumental communal construction appears while communities are still heavily dependent on wild resources and during the transition toward food production.
So these people seemingly didn't require a fully developed agricultural civilisation before they could organise substantial coordinated construction.
That suggests at least some of the social capability preceded what we traditionally label civilisation.
And once we accept that, we can ask a much more interesting question:
How far back does civilised behaviour go if we remove cities from the definition?
Not "civilisation" yet. That's deliberately provocative wording.
I mean its component capabilities:
large-scale cooperation
shared rules
division of labour
knowledge preservation
intergenerational teaching
social identity
long-distance relationships
collective planning
care for nonproductive group members
conflict resolution
ritual/shared meaning
construction for collective purposes
resource sharing
specialised expertise.
Those are the behavioural ingredients from which civilisation can eventually be constructed.
And we can search for each independently.
That's where I think this gets really interesting.
Because perhaps we've been looking at civilisation like a thing:
city + farming + monuments + writing = civilisation.
Our model would instead look at it as an emergent system.
Individual humans have knowledge.
β
Knowledge is shared.
β
Group retains knowledge.
β
Groups connect.
β
Knowledge accumulates.
β
People specialise because nobody needs to personally know everything.
β
Surplus allows some people to spend time doing things other than acquiring today's calories.
β
Specialisation increases collective knowledge.
β
Infrastructure preserves resources.
β
Permanent settlement increases interaction.
β
Interaction creates coordination problems.
β
Rules/institutions emerge to solve them.
β
Those institutions preserve information beyond individuals.
β
The group's collective capability becomes vastly greater than any individual member's capability.
And somewhere along that spiral we point at the result and say:
"Civilisation."
But there may be no single switch.
And this gives us a very important correction to the Hancock question.
We've repeatedly been saying:
"Maybe Hancock is right that we're underestimating ancient civilisation, but wrong about the explanation."
I'd change that slightly now.
Perhaps the thing we've underestimated isn't ancient civilisation.
We've underestimated ancient humans.
That's much more defensible.
There could still have been societies considerably more organised than the surviving archaeological record currently demonstrates, particularly given our submerged-landscape problem. That's a legitimate thing to investigate.
But we don't need to call those societies a lost civilisation to acknowledge that their members may have possessed sophisticated knowledge, culture and social systems.
And that's actually more interesting to me.
Because if we eventually discover a 20,000-year-old submerged settlement, our model doesn't need to collapse or suddenly declare Hancock right.
We'd say:
Excellent. The knowledge-enabled survival envelope was wider than our previous evidence demonstrated. Update the model. β or β?
Exactly how we've been doing this.
I think there's a beautiful test waiting for us now.
Rather than Googling "oldest civilisation", let's deconstruct civilisation into its parameters and then walk each parameter backwards until the evidence disappears.
Something like:
Permanent settlement --- how far back?
Large coordinated groups --- how far?
Division/specialisation of labour --- how far?
Shared identity/symbolism --- how far?
Intergenerational knowledge --- how far?
Long-distance networks --- how far?
Resource storage/surplus --- how far?
Collective construction --- how far?
Care for others --- how far?
Rules/social organisation --- what can archaeology actually demonstrate?
Then put those dates on the same timeline.
My suspicion --- and this is explicitly a hypothesis before we look --- is that we're not going to see all the ingredients suddenly appear around 10,000 BCE.
We're probably going to see them staggered backwards across tens or hundreds of thousands of years.
If that's what the evidence shows, then we may have been looking at civilisation backwards.
Instead of:
Civilisation created sophisticated human behaviour
we might end up with:
Sophisticated human behaviour existed for an enormously long time; under particular environmental, demographic and technological conditions, those existing capabilities eventually combined into the organisational form we call civilisation.
And that would be a very big piece of our breathing-Earth / knowledge-survival model.
So yes. Before farming, I think we take "civilisation" apart.
Then we see how old its pieces actually are.
Paul
Yeah, now I like what we are researching, we researching what we can observe in modern humans, intelligence, questioning, experiments, exploration... so yes let's run thru each question and see if we can find a "start" for each and then we can map them.
AI
Yes. This is the research question now.
Not "when did civilisation start?" and not really "when did humans become intelligent?"
It's:
How far back can we observe the behaviours that intelligent modern humans still use: questioning a situation, experimenting, learning, planning, teaching, cooperating, exploring and accumulating answers?
We obviously can't excavate a question. So for every behaviour we need an archaeological proxy and we keep our rule: observable fact β reasonable inference β story.
I've done a first pass across the parameters. And the result is already pretty bloody interesting.
Modern behaviour Earliest useful signal so far Approx. age Confidence we're testing ---------------- ------------------------------- --------------------- ---------------- Experiment / Early stone technologies vary 2m+ years π‘ modify a materials/techniques rather solution than one fixed response
Plan ahead / Kanjera hominins transport ~2.0m years π’ transport / take capability selected stone ~12--13 km π‘ foresight with me
Support Dmanisi individual survives ~1.77m years π’ survival / π‘ another person years after losing almost all care teeth
Preserve & Long-lived technological 1m+ years π’ transmission transmit traditions requiring learned / mechanism knowledge manufacture uncertain
Alter my Fire use; later increasingly ~1m+ years π‘βπ’ with time environment convincing deliberate fire control/construction
Construct Bruniquel Neanderthal ~176k years π’ something stalagmite structures collectively
Express ornaments/pigment/engravings, 100k+; candidates π’/π‘ identity/value with earlier isolated much older symbolically candidates
Maintain wider exotic materials moving 25--50+ ~320--295k π’ movement / π‘ social km at Olorgesailie mechanism networks
Specialised repeated dedicated hornfels ~220k π’ resource procurement at Jojosi procurement
Organise Ohalo II huts, hearths, ~23k π’ living space work/sleep areas
Manipulate Ohalo II wild-cereal ~23k π’ evidence / plants before cultivation/proto-weeds interpretation agriculture debated
Store for the purpose-built Dhra' granaries ~11.3k π’ future at infrastructure scale
Some of those dates will move as we investigate them properly. But look at what happens when we put them in chronological order.
~2,000,000 years ago --- "I'll need that over there."
At Kanjera South, hominins were moving particular stone materials at least 12--13 km from their geological sources. Researchers explicitly identify the interesting combination as stone transport, controlled flaking and mental knowledge of where stone and food resources occurred. (Environmental and Behavioral Evidence Pertaining to the... (https://www.journals.uchicago.edu/doi/full/10.1086/667704))
That's not proof somebody thought "I'll need this tomorrow."
But:
resource is here β usefulness is somewhere else β carry resource
is observable.
Our knowledge-enabled survival envelope has already started separating from the purely biological one.
~1.77 million --- "Your survival matters too."
Dmanisi gives us an individual who lost all but one tooth and survived for years afterwards. That's fact. (Anthropology: the earliest toothless hominin skull - Pu... (https://pubmed.ncbi.nlm.nih.gov/15815618/))
Whether others cared for or provisioned that individual is debated --- there's a useful warning that severely toothless non-human primates can sometimes survive too. (A toothless bonobo skull challenges the notion of alter... (https://www.sciencedirect.com/science/article/pii/S0047248420301329))
So:
survived serious impairment: π’
therefore compassionate care: π‘
That's exactly how we should treat it.
Much later, Shanidar 1 gives stronger evidence: a Neanderthal survived extensive injuries and disability into relatively old age, and the Smithsonian notes he probably could not have survived without group support. (Shanidar 1 | The Smithsonian Institution's Human Origin... (https://humanorigins.si.edu/evidence/human-fossils/fossils/shanidar-1))
So group survival extends beyond perfectly fit individuals eventually becomes increasingly hard to ignore.
That's potentially a civilisation ingredient sitting a very long way before civilisation.
~320,000--295,000 --- "People over there matter."
Olorgesailie is one I want a big flag on.
People were obtaining tool stone from 25--50 km away, using prepared-core technology and pigments. (Long-distance stone transport and pigment use in the ea... (https://pubmed.ncbi.nlm.nih.gov/29545508/))
More interestingly, some obsidian appears to have travelled farther than the likely normal range of an individual family group. Researchers therefore suggest wider exchange/social networks as one plausible explanation. (Long-distance stone transport and pigment use in the ea... (https://doi.org/doi%3A10.1126/science.aao2646))
Again:
material travelled: π’
people maintained wider networks: π‘/strongly plausible
trade: β not earned.
But we've potentially got networks larger than the immediate survival band hundreds of thousands of years ago.
That belongs on our breathing map.
~220,000 --- "I know where the good stuff is. I'm going specifically to get it."
Our new Jojosi evidence is beautiful.
People repeatedly travelled specifically to a hornfels source, processed the material there and exported useful blanks, with this behaviour persisting for tens of thousands of years. (Specialised and persistent raw material procurement by... (https://www.nature.com/articles/s41467-026-70783-8))
That's specialist knowledge of landscape resources plus future-oriented procurement.
Not civilisation.
But very recognisably modern-human problem solving.
~176,500 --- "We can build something together."
Bruniquel is still one of my favourites.
Neanderthals went 336 metres into a cave, broke stalagmites into pieces and arranged roughly 2.2 tonnes of material into deliberate ring structures.
We don't know why.
That's important.
But whatever the purpose, this isn't a single handaxe maker.
There is planning + material selection + spatial organisation + substantial labour + fire/light + construction.
That's collective capability.
And it's ~176,500 years ago.
Then, eventually, something changes in archaeological visibility.
~23,000 years ago --- the pieces start sitting together
Ohalo II is fascinating because preservation is exceptional.
We get six brush huts, hearths, bedding, tools, grinding stones, food remains, beads and organised living areas. (Ohalo II: A 23,000-Year-Old Fisher-Hunter-Gatherer's Ca... (https://www.cambridge.org/core/books/abs/quaternary-of-the-levant/ohalo-ii-a-23000yearold-fisherhuntergatherers-camp-on-the-shore-of-fluctuating-lake-kinneret-sea-of-galilee/C0E3EF4821A0C0E49B53F757D6F0C2EB))
And because the plant material survived, we can see behaviour we'd normally completely miss.
There is evidence interpreted as small-scale cultivation of wild cereals about 11,000 years before conventional Neolithic agriculture. (The Origin of Cultivation and Proto-Weeds, Long Before... (https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0131422))
That should make us very cautious about absence of evidence.
Stone survives for two million years.
A basket doesn't.
A wooden shelter usually doesn't.
A story certainly doesn't.
Our archaeological record has an enormous material-preservation filter.
Then around 11,300 years ago, Dhra' gives us purpose-built granaries with raised floors for ventilation and rodent protection --- remarkably before morphologically domesticated cereals. (Evidence for food storage and predomestication (https://naturedocumentaries.org/wp-content/uploads/2014/07/Kujit-Finlayson_2009_Granaries.pdf))
That's unmistakable future-proofing:
We have something now that we're deliberately engineering infrastructure to preserve for later.
And here's the thing I didn't expect
If we define civilisation by:
cities + agriculture + writing + government, then yes, civilisation is recent.
But if we disassemble it into the human capabilities necessary to produce civilisation, we're already finding pieces scattered enormously farther backwards:
resource knowledge β planning β learned technology β teaching β cooperation β care β networks β symbolism β specialisation β collective construction β future provisioning β environmental engineering.
And none of those individually equals civilisation.
That's crucial.
But it suggests something I'm increasingly comfortable saying:
Civilisation doesn't appear to represent the invention of intelligent, cooperative, planning humans. It appears much more likely to be a particular organisational outcome produced by capabilities that had been accumulating, interacting and being transmitted for an extraordinarily long time.
That's a hypothesis we can actually beat up with evidence.
And I think we should.
Our map now needs TWO timelines
This is the bit I'm excited about.
The first is our Earth timeline:
climate β water β sea level β habitable landscape β population movement
The second is our human capability timeline:
tools β foresight β environmental modification β teaching β networks β symbolism β construction β storage β cultivation β settlement β civilisation
Then overlay them.
And don't make capability a binary ON/OFF.
Make each one a band:
first candidate evidence β strong evidence β widespread evidence.
Because there probably isn't a single "start".
That will let us visually distinguish:
we haven't found it
from
humans weren't doing it.
Huge difference.
And then we can ask the question I think our entire detour has quietly become:
### Was civilisation a sudden leap in human capability --- or a point at which an ancient collection of capabilities finally became sufficiently connected, accumulated and useful to reorganise how humans survived?
That is a much better Hancock test than "where's Atlantis?" π
And questioning / experimentation should be the first capability we properly trace next, because that's the hardest one archaeologically and probably the most important to the whole model: when can we first reasonably observe "the current answer works... but is there another way?"
Paul
but first, lets create this new visual? if you can? I am really keen to see how you will take what we have found and turn it into a visual for us to look at?
Paul
ok i like that... damn this is so much fun... next then when, can we first reasonably observe "the current answer works... but is there another way?"
AI
Yes --- and I think we've finally got a genuinely good candidate for the first archaeological version of your question.
Not proof that somebody literally thought:
"The current answer works... but is there another way?"
Brains don't fossilise. Questions definitely don't. π
But we can ask for something observable:
An existing successful technology continues to be used, while a new and more demanding solution appears alongside it.
That distinction is important. We're not looking for replacement. We're looking for choice.
And that takes us to roughly 1.76--1.7 million years ago.
Oldowan + Acheulean --- at the same time
At Kokiselei in West Turkana, Kenya, archaeologists found Oldowan and Acheulean technologies co-occurring around 1.76 million years ago. Nature specifically notes that the two weren't simply sequential stages where one replaced the other; different stone-tool behaviours coexisted. (An earlier origin for the Acheulian | Nature (https://www.nature.com/articles/nature10372))
Then at FLK West in Olduvai Gorge at about 1.7 million years ago, the picture becomes even better.
They find:
simple flakes and cores
AND
large cutting tools
AND
crude bifacial tools
AND
highly shaped, symmetrical handaxes
in the same broad technological assemblage.
The researchers explicitly conclude that tool morphology wasn't progressing in a simple linear sequence and that the site demonstrates behavioural and technological diversity. (The Origin of The Acheulean: The 1.7 Million-Year-Old S... (https://www.nature.com/articles/srep17839))
That gives us something very different from:
bird β nest.
Because the old solution hasn't stopped working.
Oldowan works.
It had worked for roughly 700,000--900,000 years by then. Experimental work suggests it was learned and socially transmitted, and archaeological cores even show things like systematic flaking and repair when reduction went wrong. (Experimental evidence for the co-evolution of hominin t... (https://www.nature.com/articles/ncomms7029))
Then:
Oldowan still works.
β
Oldowan continues.
β
but a substantially different technological strategy appears.
β
both exist.
That's our first really strong archaeological candidate for:
"There is another way."
And here's the part I think you'll particularly like.
The Acheulean isn't merely a better sharp rock.
To produce a handaxe, you're increasingly working toward a future form.
With basic Oldowan knapping, much of the objective is obtaining useful sharp flakes.
With Acheulean manufacture, the maker is shaping a volume through a sequence of actions toward something that doesn't exist yet.
Researchers describe the earliest Acheulean as requiring greater forethought and planning, with different large tools being recurrently shaped to create points, cutting edges, picks, cleaver-like forms and other configurations. (The Origin of The Acheulean: The 1.7 Million-Year-Old S... (https://www.nature.com/articles/srep17839))
So conceptually:
I have this rock.
β
I know what it currently is.
β
I can envisage what I need it to become.
β
remove this
β
turn it
β
strike here
β
evaluate
β
adjust
β
continue
β
desired outcome.
Paul...
That's iteration.
But then I went backwards again.
Because the obvious question is:
Were they already doing that within Oldowan technology?
And annoyingly for our nice clean 1.7-million-year marker...
yes.
π
At Lokalalei 2C in Kenya, ~2.3 million years ago, refitting the flakes back onto the original cores lets archaeologists reconstruct the actual manufacturing sequence.
Some cores produced more than 50 flakes.
And here's our important bit: analysis indicates the knappers could adapt to technical problems occurring during the reduction sequence. (The origins of stone tool technology in Africa: a histo... (https://pmc.ncbi.nlm.nih.gov/articles/PMC3049100/))
That's fascinating.
Because now we've potentially got:
start with intended process
β
something doesn't behave as expected
β
observe
β
alter what I'm doing
β
continue.
That's almost exactly the behavioural unit we've been searching for.
Not necessarily:
"I wonder if there's an entirely different way to make a knife?"
But definitely something resembling:
"That didn't work. Change something."
And that's 2.3 million years ago.
Then Nyayanga pushes another part to ~2.6 million.
We already found that these Oldowan toolmakers were deliberately bypassing nearby stone and transporting preferred materials as far as 13 km.
More than 70% of the assemblage was made from non-local materials.
So local rock apparently offered one answer:
use what's here.
Instead:
some other rock is better
β
remember where it is
β
travel there
β
carry it kilometres
β
use it somewhere else.
The researchers interpret this as evidence of landscape-scale knowledge, selectivity and behavioural flexibility distinct from known non-human primate stone-tool behaviour. (Selective use of distant stone resources by the earlies... (https://pmc.ncbi.nlm.nih.gov/articles/PMC12356259/))
Again:
local answer exists.
But:
there's another way.
That's potentially 2.6 million years ago.
So I'd give our new timeline three markers rather than inventing a single birthday for questioning.
~2.6 million years --- CHOICE
I don't have to use what's immediately available.
Selected materials transported long distances. π’ (Selective use of distant stone resources by the earlies... (https://pmc.ncbi.nlm.nih.gov/articles/PMC12356259/))
~2.3 million years --- FEEDBACK / ADJUSTMENT
Something in my current process isn't working; alter the process.
Knappers responding to problems during manufacture. π’/π‘ depending how much cognition we infer. (The origins of stone tool technology in Africa: a histo... (https://pmc.ncbi.nlm.nih.gov/articles/PMC3049100/))
~1.76 million years --- ALTERNATIVE SOLUTION
The existing technology still works, but another technological system now exists alongside it.
Oldowan + Acheulean coexistence. π’ (An earlier origin for the Acheulian | Nature (https://www.nature.com/articles/nature10372); The Origin of The Acheulean: The 1.7 Million-Year-Old S... (https://www.nature.com/articles/srep17839))
And those three together are much stronger than any one alone.
Because we've now got:
SELECT β TRY β OBSERVE β MODIFY β RETAIN ALTERNATIVES
That's beginning to look remarkably like experimentation.
And there is something else here that I think matters enormously to our civilisation visual.
Remember our original suspicion:
Maybe humans only became sufficiently clever relatively recently.
We're now struggling to maintain that.
At 2.6 million years, we've got selective landscape-scale resource behaviour.
At 2.3 million, adaptive problem-solving during manufacture.
At 1.7 million, significantly different technological solutions coexisting.
At ~1 million+, environmental/fire manipulation enters our story.
Hundreds of thousands of years ago we add composite technologies, planned hunting, wider networks and eventually construction and symbolism.
So our visual shouldn't have:
INTELLIGENCE βββββ starting 300,000 years ago.
It needs something much more like:
BEHAVIOURAL FLEXIBILITY / EXPERIMENTATION
β₯2.6 million years ββββββββββββββββββββββββββββββββ TODAY
with the evidence becoming richer and more visible over time.
And here's where I think your question needs to become an actual parameter on our model.
Because there's a subtle difference between intelligence and what we're finding.
Intelligence can solve:
How do I do this?
But the behaviour driving our spiral is:
Does this still work?
and then:
Is there another way?
That's the engine.
Existing answer
β
observe
β
question/challenge
β
alternative
β
test
β
β / β / β
β
preserve useful result
β
share
β
someone else starts from the new answer
β
question again.
And I think that gives us an unexpectedly strong answer to the question we started with:
### When can we first reasonably observe "the current answer works... but is there another way?"
Conservatively: by ~1.7 million years ago, absolutely something like technological alternatives are visible.
More interestingly: by ~2.3--2.6 million years ago, we already see behavioural choices and adjustment that look like the foundations of that process. (Selective use of distant stone resources by the earlies... (https://pmc.ncbi.nlm.nih.gov/articles/PMC12356259/); The origins of stone tool technology in Africa: a histo... (https://pmc.ncbi.nlm.nih.gov/articles/PMC3049100/))
And that means "questioning/experimentation" goes almost all the way to the left-hand edge of our new visual.
Which is DAMN significant.
Because if that behaviour is already ancient, our investigation of civilisation changes again.
The question becomes less:
When did humans acquire the intelligence necessary for civilisation?
and increasingly:
If the problem-solving engine was already running millions of years ago, what eventually changed that allowed its answers to accumulate fast enough, connect widely enough and persist reliably enough to produce what we call civilisation?
That feels like the next parameter we need to hunt.
Not intelligence.
Accumulation rate.
Because perhaps that's the curve that eventually goes vertical.
Paul
yeah, again though what this picture is starting to show is we are underestimating how clever these "humans" were. I know I have been letting you run with this, but i have a couple of things going through my mind... a couple of questions. the first "what changed to make us become civilized" like what about the environment we were in meant that became a need. i really need us to not lose sight of some human behaviour - conflict. the reason i say this is that when did we start building walls for survival, cos right now, all this is telling me the need for 'protected' settlements seems kind of unnecessary? the other thing i just forgot ha ha. Oh thats right. can we just fold in current. you have brought accumulation rate in here... so what about what i am doing... what is the timeline for getting to "Paul talks to AI to solve world hunger" cos that is also an important fact. if that makes sense too... so Accumulation rate?
AI
Yes. Accumulation rate. And I think the two thoughts you've just had actually belong on the same graph.
Because you're asking:
1. If these people were already remarkably capable, what changed in the survival problem that made "civilisation" useful? 2. Why does the rate at which answers accumulate appear to go from glacially slow β faster β ridiculous β Paul sitting in Christchurch talking to an AI about the entire history of human problem-solving? π
I think those converge.
First: walls --- because I think you're onto something important
Your question is essentially:
Why protect a settlement if leaving is still the better survival strategy?
Exactly.
A mobile group threatened by another group has an option:
move.
But once you've invested in:
food stores + crops + buildings + water access + accumulated resources + territory you've modified
...moving becomes expensive.
Now there is something worth defending.
And interestingly, the archaeological sequence is compatible with that hypothesis --- but doesn't prove it.
We have clear evidence of serious recurrent interpersonal violence before cities. At Jebel Sahaba in the Nile Valley, at least ~13,400 years ago, many individuals suffered projectile injuries; some had both healed and fresh wounds, indicating repeated episodes rather than one massacre. Researchers think environmental change and competition over resources probably contributed. (New insights on interpersonal violence in the Late Plei... (https://www.nature.com/articles/s41598-021-89386-y))
At Nataruk in Kenya around 10,000 years ago, hunter-gatherers also died violently in what has been interpreted as inter-group violence. (Inter-group violence among early Holocene hunter-gather... (https://www.nature.com/articles/nature16477))
So:
conflict before civilisation: π’
conflict before farming: π’
But then look at Jericho.
By the 9th--8th millennium BCE, Tell es-Sultan is a substantial permanent settlement with a wall, ditch and monumental tower, centred on a perennial spring. (Ancient Jericho/Tell es-Sultan - UNESCO World Heritage... (https://whc.unesco.org/en/list/1687/))
Now --- discipline --- the early Jericho wall's purpose is debated. Defence is one possibility; flood protection and social/communal functions have also been proposed. So we absolutely cannot write:
settlement β enemy β build wall.
But your hypothesis is now testable:
As accumulated investment in a place increases, does the survival response to conflict shift from MOVE toward DEFEND?
That is a fantastic parameter for our map.
Because perhaps settlement doesn't create conflict.
Perhaps settlement changes the economics of conflict.
Before:
Threat β leave.
Later:
Threat β hang on, we've got 40 tonnes of grain, houses, fields, a reliable spring and three generations of infrastructure here.
We're not DAMN leaving.
π
That's potentially where territory acquires a very different survival value.
Now accumulation rate.
This may be one of the most important concepts we've encountered.
Researchers call part of this cumulative culture and the ratchet effect.
Someone discovers A.
Another person learns A and modifies it to B.
The next generation starts at B, rather than rediscovering A.
Someone creates C.
Then D.
The crucial feature isn't simply intelligence or invention. It's preventing successful answers from continually slipping backwards and being lost. (Ratcheting up the ratchet: on the evolution of cumulati... (https://pmc.ncbi.nlm.nih.gov/articles/PMC2865079/))
Sound familiar?
β
π
That's basically our spiral.
And researchers think accumulation depends on things we've independently been finding:
innovation + teaching + transmission fidelity + cooperation + population/network structure + recombination of ideas. (The origins of human cumulative culture: from the forag... (https://pmc.ncbi.nlm.nih.gov/articles/PMC8666907/))
But here's the really interesting bit.
More connection isn't automatically better.
Experiments and models find that partially connected groups can outperform completely connected groups.
Why?
Separate groups explore different answers.
Group A discovers B.
Group C discovers F.
Group D discovers something weird.
Then occasionally they reconnect.
B + F + weird thing β new solution.
Completely isolated groups lose information.
Completely homogenised groups can lose diversity of experimentation.
Partially connected populations can provide both. (Partial connectivity increases cultural accumulation wi... (https://pmc.ncbi.nlm.nih.gov/articles/PMC4801235/))
Paul.
That's our DAMN breathing model again.
DISCONNECT
Different populations experiment locally.
β
different answers accumulate
β
RECONNECT
answers meet
β
recombination
β
new possibilities
β
DISCONNECT
different experiments begin again.
Research into hunter-gatherer cumulative culture actually proposes something remarkably similar: multilevel social networks may preserve diversity while allowing knowledge to recombine, effectively creating a network memory larger than any individual's brain. (The origins of human cumulative culture: from the forag... (https://pmc.ncbi.nlm.nih.gov/articles/PMC8666907/))
That's huge for our model.
And now let's fold 2026 Paul into it.
Because you're right: today isn't outside the timeline.
We're currently standing on the far-right edge of the exact phenomenon we're studying.
Think about the accumulation problem.
2.5 million years ago
Someone discovers something.
The storage medium is predominantly:
brain.
Transmission bandwidth:
watch me / copy me / perhaps gesture/vocalise.
Network:
people I encounter.
Knowledge loss risk:
enormous.
Accumulation rate:
very slow.
Language
Now I can transfer something I learned without you having to witness me discovering it.
Bandwidth jumps.
External symbols / art / notation
Information starts existing outside brains.
Storage improves.
Writing
Now:
the person who knows something can die without the information necessarily dying.
That's an enormous ratchet improvement.
Printing
One answer can suddenly reach thousands of brains.
Copying cost collapses.
Scientific method
And here's a really important one for our model.
Humans formalise:
question β hypothesis β test β observe β reproduce β publish β challenge β update.
Sound familiar?
π
We essentially institutionalised the spiral.
Knowledge isn't merely accumulated.
It's increasingly tested before being retained.
Internet
Now billions of humans potentially join the network.
Storage becomes enormous.
Search becomes fast.
Distance becomes dramatically less relevant.
And now: AI.
This is where your example becomes completely legitimate.
Right now:
Paul has a question.
You don't personally know archaeology.
You don't personally know palaeogenetics.
You haven't spent 30 years studying sea-level reconstruction.
You don't know every paper about lithic technology.
Previously, pursuing this particular rabbit hole properly might have required:
university access + experts + libraries + years of reading + enormous amounts of time.
Instead:
Paul asks question.
β
AI accesses/synthesises large amounts of accumulated human knowledge.
β
I return possible answer.
β
Paul challenges it.
β
"That's still a bird building a nest."
β
We alter the parameter.
β
search again.
β
new evidence.
β
Paul asks another question.
β
repeat.
β β β
That's the same bloody spiral.
But the cycle time has collapsed.
That's the important thing.
So maybe accumulation rate has TWO variables.
We've been thinking:
HOW MUCH KNOWLEDGE EXISTS?
But perhaps the more important variable is:
HOW FAST CAN A QUESTION COMPLETE A CYCLE?
Imagine:
2 million years ago:
question β experiment β lifetime/generations β transmission β next experiment
Later:
question β tell someone β experiment
Writing:
question β read previous answers β experiment
Science:
question β search literature β experiment β publish β global criticism
Internet:
question β search millions of sources β test
AI:
question β interrogate accumulated knowledge β synthesise β challenge β reformulate β question again
Potentially in minutes.
That's not necessarily an increase in individual human intelligence.
It's an enormous increase in the speed at which an individual can access collective intelligence.
And THAT may be the curve we've been missing.
Imagine our visual now.
Along the bottom:
2,500,000 years β TODAY
One curve:
INDIVIDUAL COGNITIVE CAPABILITY
It increases, obviously, particularly through biological evolution early on --- but perhaps it becomes relatively flatter much earlier than our civilisation story traditionally implies.
Then another curve:
COLLECTIVE KNOWLEDGE
Slow...
slow...
rising...
writing...
printing...
science...
internet...
WHOOOOSH.
And a third:
ACCESS SPEED
For most of history: tiny.
Language β
Writing β
Printing β
Telecommunications β
Internet ββ
Search engines βββ
AI ββββ
And Paul is sitting right there at 2026.
Not because Paul suddenly became 1000Γ more intelligent.
Your knowledge-enabled survival envelope just became DAMN enormous.
You can ask:
"Why did humans start building defensive walls?"
and within seconds draw upon archaeology, anthropology, climatology, evolutionary theory and cultural-evolution research.
Then your intelligence does the thing we've now traced back millions of years:
"Yeah... but does that answer actually make sense?"
That's still the human bit.
π
And I think this gives us a very important correction.
AI doesn't necessarily represent another step called:
MORE KNOWLEDGE.
Humanity already possesses an absurd amount.
It potentially represents:
LOWER FRICTION BETWEEN QUESTION AND ACCUMULATED KNOWLEDGE.
And if our entire model is:
observe β question β explore β test β learn β preserve β share β repeat
then reducing the time and effort required for that cycle could massively increase the accumulation rate.
Which is literally what we're doing right now.
And now I can connect BOTH of your thoughts.
Maybe civilisation emerges when accumulation creates things worth staying for.
Knowledge produces:
better resource prediction
β storage
β cultivation
β surplus
β infrastructure
β permanent places.
Permanent places concentrate:
people + knowledge + resources.
That increases:
interaction + specialisation + idea recombination.
Which increases:
accumulation rate.
But concentration also creates:
valuable resources + territorial dependence + competing groups.
So suddenly:
DEFENCE becomes valuable.
Walls.
Rules.
Leadership.
Coordination.
Collective labour.
Perhaps armies eventually.
Administration.
And those things themselves create more organisational knowledge.
Another feedback loop.
So civilisation might not be a mysterious cognitive leap at all.
It could be a phase change in the knowledge-survival system:
Enough accumulated knowledge allowed people to reduce survival uncertainty sufficiently to stay put.
Staying put concentrated people and resources.
Concentration accelerated knowledge accumulation and increased the cost of losing the place.
That generated new survival problems:
ownership, conflict, defence, coordination, distribution, sanitation, governance...
And humans did what humans apparently had already been doing for millions of years:
"Well, DAMN. How do we solve this one?"
π
And around the spiral again.
That's the hypothesis I'd put on our board now.
Not:
INTELLIGENCE β CIVILISATION.
But:
INTELLIGENCE β KNOWLEDGE β ACCUMULATION β REDUCED UNCERTAINTY β INVESTMENT IN PLACE β NEW PROBLEMS β ORGANISATION β FASTER ACCUMULATION β CIVILISATION.
With conflict absolutely included as one of those new problems rather than treated as an inconvenient side note.
And at the far right of that same line:
2026 --- accumulated human knowledge + near-instant retrieval/synthesis + one human asking questions.
Or, as you've more elegantly put it:
Paul talks to AI to solve world hunger. π
Not because we necessarily can solve world hunger this afternoon.
But because a single person can now interrogate a quantity of accumulated human knowledge that would have been effectively inaccessible to any individual human in almost all previous history.
That belongs on our visual.
And I think we should actually build Accumulation Rate v1 next --- from oral transmission β external memory β writing β printing β science β telecommunications β internet β AI, while underneath it we track settlement/resource investment/conflict/defence.
Because I suspect when we overlay those two curves, we're going to see why "civilisation" may be an acceleration event rather than an intelligence event.
Paul
yes, now i think you and i am thinking about this in the same way... we anchor with priority 1 - survival, never changes. but before we add conflict, we add comfort maybe, or the word you find, by that i mean, as my daily survive becomes more certain - oh certainty - I now have time to ponder more and learn and discover maybe... there is a saying "use the time to sharpen your axe" its the idea that the woodsman comes home from wood cutting once he has enough and then spends the time sharpening it... which actually, as you know i like, i am seeing different now... cos the "axe" is maybe one of the oldest tool humans have, gee it could be the first tool even, not proven... so now that saying looks a whole lot different... i have used my axe to kill, to eat or defend myself, to survive, now i will look after my axe, so it is ready for the next time i NEED it! Ok, so i went off on my own tangent then. but this is where we start with accumulation rate, then look at events that cause conflict, that could threaten my survive. and the thing i am thinking is as accumulation rate increases, my NEED to survive decreases... in terms of the time i can survive with the resources i have... if that made sense?
AI
Yes. That makes sense, and I think you've just found the variable that needs to sit between survival and accumulation:
SURVIVAL MARGIN
Not comfort, exactly. Certainty is part of it, but survival margin captures what I think you're describing:
How far am I from the point where failure means I don't survive?
If I have no food and must hunt successfully today, my survival margin is tiny.
If I have food stored for three days, it's larger.
If my group has reliable food, water, shelter and fire for a season, larger again.
If I have crops, livestock, storage, permanent shelter and a social network that can support me, my margin becomes enormous compared with the earlier condition.
My need to survive hasn't decreased.
That's important.
Priority #1 remains:
SURVIVE.
What decreases is the proportion of my time and attention required for immediate survival.
And that releases something extraordinarily valuable:
TIME.
Your axe tangent isn't a tangent at all.
And one tiny correction actually makes it better: the earliest known stone tools aren't handaxes. The oldest known stone-tool assemblages are much older and simpler; recognisable Acheulean handaxes arrive around 1.7--1.8 million years ago.
But your woodsman analogy still works beautifully.
Imagine the earliest version:
I need food.
β use tool.
β acquire/process food.
β I have enough food.
Now there are two possible uses of the next hour:
keep acquiring food I don't currently need
or
maintain/improve the thing that helped me acquire it.
That's profound when viewed through our model.
Because sharpening the axe is future survival work.
I'm not hungry right now.
I'm not being attacked.
There's no immediate survival problem.
Yet I'm spending energy because I can imagine:
I will NEED this again.
So the axe becomes something more than a survival tool.
It becomes stored capability.
And sharpening it is investment in future capability.
That's exactly what we've been hunting.
Now accumulation rate gets really interesting
Imagine two populations that are equally intelligent.
Population A
Every waking hour is consumed by:
find water β find food β avoid predator β find shelter β sleep β repeat.
Survival margin β zero.
There might be an extraordinarily clever individual in that population.
But experimentation is expensive.
Trying another way could mean:
β die.
There's also little spare time to teach, experiment, decorate, construct, explore, maintain equipment or sit around wondering why the stars move.
Now Population B has learned enough to create:
reliable fire + efficient hunting + stored food + good shelter + predictable water + cooperative group.
Their brains haven't magically changed overnight.
But their survival margin has.
And therefore:
TIME REQUIRED FOR IMMEDIATE SURVIVAL β
which potentially means:
TIME AVAILABLE FOR EVERYTHING ELSE β
And everything else includes:
maintain
teach
experiment
explore
observe
make
play
specialise
question
think.
That's our accumulation engine.
And there's a feedback loop hiding in it.
Knowledge improves survival.
β
Improved survival creates survival margin.
β
Survival margin creates available time/capacity.
β
Available capacity permits more experimentation and learning.
β
Learning produces more knowledge.
β
Knowledge further improves survival.
β
survival margin increases again.
β
repeat β
That's a positive feedback loop inside our spiral.
And NOW I can see how accumulation rate might suddenly accelerate without humans suddenly becoming smarter.
The intelligence could have been there for an enormously long time.
What changes is the amount of that intelligence that can be spent on something other than today's emergency.
And this gives us a much better definition of your certainty.
I wouldn't say:
"As accumulation increases, my need to survive decreases."
Because the NEED never changes.
I'd say:
As accumulated knowledge increases my survival margin, the urgency of immediate survival can decrease.
That's exactly what you're describing.
And that creates future certainty.
Not certainty that I will survive.
Rather:
How far into the future can I reasonably expect my existing resources and knowledge to carry me?
That's measurable conceptually.
No stored food:
hours/day.
Stored meat:
days.
Seasonal cache:
weeks/months.
Granary:
months/year.
Agriculture + storage:
seasons/years.
Infrastructure + trade/network:
potentially years and generations.
And every expansion gives me more ability to spend today's effort on tomorrow.
Which suddenly makes civilisation look different again
Maybe one of the prerequisites isn't merely surplus.
Surplus is the archaeological object.
What surplus creates is the interesting thing:
TIME HORIZON.
Hunter facing today's hunger:
NOW.
Hunter with food until tomorrow:
TOMORROW.
Group with seasonal storage:
NEXT SEASON.
Farmer retaining seed:
NEXT YEAR.
Settlement building infrastructure:
YEARS.
People constructing monuments:
GENERATIONS.
Civilisation building institutions:
AFTER I'M DEAD.
That's potentially an extraordinary axis for our visual.
Because the knowledge-enabled survival envelope isn't merely expanding across space.
It's expanding across time.
And NOW we can bring conflict in properly.
Not yet as:
humans become settled β humans become violent.
We've already established conflict existed earlier.
Instead:
What happens when one group's survival margin becomes a resource another group can take?
That's different.
If I have nothing stored and you attack me, perhaps my best survival decision is:
run.
But if accumulated knowledge has given my group:
stored food
tools
livestock
productive land
water infrastructure
houses
specialist knowledge
then what you can take from me isn't merely today's meal.
You're potentially taking my survival margin.
You're collapsing my future from:
months/years
back to:
today.
Suddenly:
DEFEND becomes rational.
And perhaps that's where walls become much more interesting.
A wall isn't simply:
keep bad people out.
Through our model, it's:
protect accumulated survival margin.
Food.
Water.
People.
Knowledge.
Tools.
Children.
Infrastructure.
Future certainty.
Holy shit.
That gives us a very clean sequence to test:
KNOWLEDGE β β SURVIVAL EFFICIENCY β β SURVIVAL MARGIN β β AVAILABLE TIME β β EXPERIMENTATION β β ACCUMULATION RATE β β FUTURE HORIZON β β INVESTMENT β β SOMETHING WORTH PROTECTING β
and only then do we introduce:
THREAT / CONFLICT β DEFENCE
Not because conflict suddenly appears.
Because what can be lost has changed.
And Paul, your axe saying has become bloody brilliant inside this model.
"Use the time to sharpen your axe."
Originally it sounds like a productivity slogan.
Through the lens we've built, it's almost a miniature model of cumulative human behaviour:
I have satisfied the immediate survival requirement.
Therefore I have created time.
I use that time not merely to rest, but to maintain/improve the capability that created the margin.
So next time I need it:
my starting point is better than last time.
That's the ratchet.
That's β.
And perhaps one of the biggest things accumulated knowledge has ever bought humanity isn't comfort.
It's:
TIME BETWEEN NEEDS.
Because inside that space between one urgent survival need and the next...
we get to think.
And that may be where the accumulation-rate curve starts bending upward.
I reckon that is the next thing we should actually research: can we observe the survival margin increasing archaeologically over time? Food acquisition efficiency, storage, maintenance of tools, seasonal planning, dwellings, division of labour, leisure/ornamentation --- and whether those behaviours appear alongside increasing rates of innovation.
If that relationship holds up, we've got something much more interesting than a story about civilisation.
We've got a candidate mechanism for why the curve accelerates.
Paul
yes that we have, cos as we create time between needs, it not only gives us thinking time, it starts us going 'what shall we do with it?' maybe explore, or start a fight again, study the stars, doing a painting, i mean i am totally making it up, BUT that is the important thing. time to think is some ways... is dangerous. so, can we observe the survival margin increasing archaeologically over time?
AI
Yes. We can observe it. Not as a clean line called "survival margin", obviously, but through behaviours that increasingly make survival less immediate, less uncertain and less all-consuming.
And your point about the danger of spare time matters. We shouldn't build a progress fairy tale where spare capacity automatically produces astronomy and art. π It can produce exploration, experimentation, play, socialising, status competition, conflict, ritual, decoration --- or doing absolutely nothing useful.
The archaeological question is narrower:
Can we see humans increasingly doing things that are not necessary to solve the immediate survival problem, because previous knowledge has created some distance from that problem?
I think the answer is yes.
The first really interesting marker is astonishingly old: ~790,000 years
At Gesher Benot Ya'aqov in Israel, we have repeated occupation beside an ancient lake, controlled fire, animal processing and a diverse plant diet. By about 750--790k years ago, people aren't simply scattering activity randomly around a temporary kill.
They are organising space.
Stone-working occurs in particular places. Plant and animal processing occurs in particular places. Multiple activities concentrate around a hearth. Researchers describe this as evidence of organised living space and advanced organisational ability. (Spatial organization of hominin activities at Gesher Be... (https://pubmed.ncbi.nlm.nih.gov/20019284/))
That's subtle, but important.
There's a difference between:
survive here
and
organise HOW we live here.
That doesn't prove leisure.
But it shows effort being spent making repeated survival activity structured and efficient.
That's a candidate early survival-margin behaviour.
Then our evidence starts stacking.
A controlled hearth isn't merely warmth right now.
It potentially means:
cook β extract calories more efficiently
light β extend usable time
warmth β reduce energy expenditure
protection β reduce predator risk
social centre β concentrate teaching/interaction
So fire doesn't merely solve one need.
It potentially reduces the cost of several needs simultaneously.
That's exactly what should increase survival margin.
And Gesher Benot Ya'aqov gives us controlled fire together with organised activity space nearly 800,000 years ago. (Evaluating the intensity of fire at the Acheulian site... (https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0188091); Spatial organization of hominin activities at Gesher Be... (https://pubmed.ncbi.nlm.nih.gov/20019284/))
Now run our spiral:
knowledge
β fire
β several survival problems become cheaper/easier
β time/energy released
β other activities become possible.
We can't excavate the spare hour.
But we can excavate what people sometimes did with increasingly complex activity systems.
Then something like Bruniquel becomes more interesting
~176,500 years ago.
Neanderthals go hundreds of metres underground and construct those stalagmite arrangements we've discussed.
Whatever their purpose, this is important through our new lens because nobody was forced by immediate hunger to spend time arranging tonnes of broken stalagmite deep inside a cave.
We don't know what they were doing.
But that's precisely the point.
They had enough capacity to undertake a substantial activity whose obvious purpose was not simply acquiring today's calories.
That is evidence of discretionary effort.
I think that's a much safer archaeological term than "leisure."
And discretionary effort may be our proxy for survival margin.
Then we get pigment processing, ornaments, engraving, beads...
And these suddenly look different through our model too.
An engraved piece of ochre isn't necessarily helping me kill dinner.
A bead isn't keeping me warm.
Painting something doesn't quench thirst.
Again, their meanings are another question.
But their existence demonstrates:
Human effort is being allocated beyond immediate biological necessity.
That's important.
Because energy is expensive.
Time is expensive.
If I'm genuinely one failed hunt from starvation every single waking moment, spending hours obtaining unusual pigment, processing it, engraving something or making ornaments becomes much harder to explain.
Not impossible --- ritual/social identity itself can have survival value --- but we're clearly entering a richer allocation of human time.
And then Ohalo II at ~23,000 years ago is DAMN fantastic for this.
Because preservation lets us see an unusually complete slice of life.
Six brush huts.
Grass bedding.
Hearths.
Fishing.
Hunting.
Birds.
Wild cereals.
Grinding stones.
Bowls.
Bone tools.
Beads.
A grave.
Different activity areas.
And possibly year-round occupation. (Abundance or stress? Faunal exploitation patterns and s... (https://pmc.ncbi.nlm.nih.gov/articles/PMC8791512/))
Look at the food system alone.
They're not relying upon:
find animal β kill β eat.
They're exploiting multiple food channels:
fish + mammals + birds + wild cereals + other plants.
They are harvesting cereals with composite sickles and grinding wheat, barley and oats; there is even evidence interpreted as small-scale cultivation. (Composite Sickles and Cereal Harvesting Methods at 23,0... (https://pmc.ncbi.nlm.nih.gov/articles/PMC5120854/))
That's risk diversification.
And that might be another major survival-margin parameter.
Instead of:
Survival depends on A succeeding.
knowledge creates:
A failed? We have B, C, D and E.
That increases certainty without necessarily increasing total food.
That's exactly like a modern resilient system.
Remove the single point of failure.
And notice another thing at Ohalo.
Bedding.
Grass carefully laid across roughly seven square metres of one hut floor. (Abundance or stress? Faunal exploitation patterns and s... (https://pmc.ncbi.nlm.nih.gov/articles/PMC8791512/))
Now we've finally reached your word:
comfort.
Grass bedding isn't required in the absolute biological sense that water is.
But once I've solved enough of:
water + calories + temperature + safety
I can ask:
Can sleeping here be better?
π
That's a tiny question with enormous implications.
Because survival has moved from binary:
alive/dead
toward:
How well can I live while remaining alive?
That's another expansion of the envelope.
And now I think we can define what we're actually looking for.
Archaeological proxies for increasing survival margin
Efficiency --- accomplish survival tasks with less effort.
Reliability --- make success less dependent upon today's luck.
Diversification --- multiple ways to satisfy the same need.
Maintenance --- invest now in keeping future capability functional.
Storage --- move resources through time.
Shelter/environment modification --- reduce exposure to environmental variation.
Organisation --- structure space and activity to reduce friction.
Discretionary effort --- spend substantial time/energy on things not obviously required for immediate biological survival.
And eventually:
Specialisation --- I don't personally have to perform every survival task.
That last one may be enormous.
Because when I can spend my day making tools because you are obtaining food...
we've separated:
my individual survival activity
from
our collective survival system.
That's another jump in survival margin.
And I think your TIME BETWEEN NEEDS now needs splitting into two dimensions.
1. TIME UNTIL THE NEXT NEED BECOMES CRITICAL
Food stored for three months means hunger isn't today's emergency.
2. TIME REQUIRED TO SATISFY THE NEED
If technology lets me acquire today's calories in four hours rather than twelve, I've created eight hours.
Those aren't the same thing.
But both create:
AVAILABLE CAPACITY
And that's where your dangerous human appears. π
Because once the woodsman has enough wood...
and the axe is sharp...
and there's food stored...
and the fire is burning...
what the DAMN does he do now?
We absolutely cannot assume:
science!
Maybe he sleeps.
Maybe he plays with his kids.
Maybe he carves something.
Maybe he watches the stars.
Maybe he walks over the hill because he's curious.
Maybe he invents something.
Maybe he tells stories.
Maybe he paints.
Maybe he tries to impress someone.
Maybe he decides those bastards across the river have nicer stuff than his group does.
All are possible uses of released capacity.
And that is where conflict should enter later.
Not because spare time causes war.
Because survival margin increases the range of behaviours humans have the capacity to pursue.
And now I think we've got our accumulation-rate mechanism much cleaner:
KNOWLEDGE β EFFICIENCY + RELIABILITY β SURVIVAL MARGIN β AVAILABLE CAPACITY β MORE POSSIBLE BEHAVIOURS β exploration / experimentation / teaching / art / ritual / play / maintenance / competition / conflict / etc. β some produce useful new knowledge β PRESERVE + SHARE β KNOWLEDGE β β repeat.
There is our acceleration loop.
And importantly:
it doesn't require humans to suddenly get smarter.
That's becoming one of the strongest themes in everything we've uncovered.
It requires existing intelligence to have more opportunity to operate.
Which means I want to slightly revise yesterday's statement that civilisation might be an acceleration event rather than an intelligence event.
I think we can make it sharper:
### Civilisation may be what becomes possible when accumulated knowledge creates enough survival margin for large numbers of already-intelligent humans to spend increasing amounts of capacity on things other than immediate survival.
Then those activities themselves generate more knowledge.
Which creates more margin.
Which creates more capacity.
Which accelerates accumulation.
β around the spiral.
And eventually something very weird happens.
The margin becomes so enormous that one particular human can have his immediate food, water, shelter and safety needs handled by a gigantic system of other humans...
sit in a chair...
drink coffee...
and spend hours asking an artificial intelligence:
"Hang on. Have we underestimated what humans have been doing all along?"
π
That's not outside our experiment.
That's the far-right data point.
And now I think we have exactly the right foundation to introduce conflict next.
Because the question is no longer merely "when did humans start fighting?"
It's much better:
As survival margin grows, when does accumulated survival value become sufficiently concentrated in people, food, territory, infrastructure and knowledge that another human group becomes one of the greatest threats to it?
That is where I'd go next.