What divination is

Cicero sorted it into two kinds in 44 BC while serving as an augur himself — and wrote the best ancient argument that none of it works.

What it is

The attempt to obtain knowledge — of the future, of the hidden, of what should be done — by means other than ordinary observation and inference.

The most useful classification is over two thousand years old. In De Divinatione (44 BC), Cicero divided divination into two kinds:

Natural or inspired divination — knowledge arriving directly through the diviner: dreams, prophetic frenzy, visions, possession. The practitioner is a channel and does not need training so much as a susceptibility.

Artificial or technical divination — knowledge read from signs by learned method: the flight of birds, the livers of sacrificed animals, the fall of lots, the positions of planets, the arrangement of cards. The practitioner is an interpreter and does need training.

Almost everything in this section is artificial divination. Almost everything in it also shares a feature worth naming: a deliberate randomising step.

Where it came from

It is universal, in the strong sense — no society is known to have lacked it. The forms recur with striking independence: casting lots, reading marks in scattered material, examining sacrificed animals, opening a sacred book at random, interpreting dreams.

The Roman state ran on it institutionally. Augurs took the auspices before public business, and unfavourable signs could halt an assembly. Haruspices, a specialism inherited from the Etruscans, read the livers of sacrificed animals. The practice was not private consolation but constitutional procedure.

Which makes Cicero's book remarkable. He was himself an augur — an elected member of the college — and wrote the most thorough ancient demolition of divination, in a dialogue where his brother argues the case for and he argues against. His objections are still the standard ones: the predictions are vague, the hits are remembered and the misses forgotten, the practitioners disagree, and if the future is fixed then knowing it changes nothing, while if it is not fixed then it cannot be known.

He nevertheless defended keeping the state rituals, on the grounds that they held the community together. That combination — this does not work, and we should continue doing it — is worth sitting with.

How it is used

The obvious use is to obtain information. The uses that survive examination are different and more interesting.

Breaking deadlock. A decision with no clearly better option costs time and generates conflict. A divinatory procedure ends the deliberation and produces an answer everyone has agreed in advance to accept.

Distributing responsibility. If the lots chose, no faction chose. Divination removes the decision from any person's hands, which protects both the decider and the group's cohesion.

Randomising deliberately. This is the most interesting proposal in the anthropological literature. In 1957 Omar Khayyam Moore argued that Naskapi hunters in Labrador, who determined hunting direction by heating a caribou shoulder blade until it cracked and reading the cracks, were in effect randomising their search. Randomisation would prevent them from over-hunting recently productive areas and from becoming predictable to their prey. The specific claim has been criticised on ethnographic grounds and should not be taken as settled. The general point survives: under some conditions a random strategy genuinely outperforms a considered one, which game theory formalises as mixed strategies, and a culture with no concept of randomisation might arrive at it through ritual.

Licensing what is already known. Diviners frequently tell people what, on some level, they had already decided. The procedure supplies permission rather than information.

What we can and cannot say

We can say there is no evidence that any divinatory system yields information about the future or about hidden facts. Where systems have been tested — dowsing, astrology, card reading — results have been at chance.

We can say the psychology is well characterised. Apophenia, the perception of meaningful patterns in random data, was named by Klaus Conrad in 1958 and is a normal function of a pattern-finding brain rather than a defect. The Forer effect covers the acceptance of general statements as personal. Confirmation bias and selective recall do the rest: a reading that produces twenty statements and two striking hits is remembered as two hits.

We can say the social functions are real and studied, and that they explain the universality far better than the accuracy claim does. A technology for making decisions under uncertainty, ending argument, and spreading responsibility would be expected to appear everywhere. It has.

We cannot say the experience is empty for the people who have it. A person who turns a card and finds themselves suddenly clear about something has had a real experience of clarity. The question this section keeps returning to is whether the card supplied it or occasioned it — and those are different claims with very different evidential burdens.

What it is

Where it came from

How it is used

What we can and cannot say

Further reading