The Oracle's prophecies were intentionally vague. For instance what the Athenians asked the oracle what to do about the Persians when Xerxes invaded first it told them to submit to the persians. Then when they more or less asked for a better prophecy and were told that they would be saved if they stayed behind a wooden wall. Some belived that it referred to the hedge around the acropolis. Others believed that it referred to a sea battle they must fight. If they had lost then they may have believed they misinterpreted the oracle

I'm still at work so I can answer this in more depth later (with more precise references) if people want.

Tom Harrison's book on Religion and Herodotus is based around the various strategies that one man (Herodotus) dealt with oracle predictions. There was a vagueness to Oracle predictions that allowed for a large element of confirmation bias to be involved. Human and natural events could all be interpreted in light of the predictions to confirm the gods involvement. Everyone remembers the manner in which Themistokles interpreted 'the wooden wall' to mean ships rather than an actual wall, or Croesus bringing about the decline of 'a great empire.' There was also another get out clause - predictions were timeless and could come to fruition generations later (Croesus is again a good example of carrying the ancestral guilt of Gyges). This idea of ancestral guilt is explored in Renaud Gagne's book.

Ester Eidenow also did a large investigation into the oracle texts at Dodona, which revealed that in everyday usage ordinary Greeks often would form their questions in such a way as to determine the answer they wanted.

Finally and most bizarrely, Versnel (based on the work of Festinger) is currently working on the idea that failures can strengthen the belief rather than disprove it. If the end of the world does not happen as you predicted, it confirms that God saved you. In such a way you can avoid cognitive dissonance.

• Harrison (2000) Divinity and History: The Religion of Herodotus

• Eidenow (2010) Oracles, Curses, and Risk Among the Ancient Greeks

• Gagne (2013) Ancestral Fault in Ancient Greece

• Versnel (1990) Inconsistencies in Greek and Roman Religion (Vol 1)

A follow up question (or rather background question I suppose): To what extent did the Greeks seek and believe in the Oracle's predictions, and to what extent were the Oracle's statements testable?

I can't seem to find whether the Oracle was widely engaged with, and if so whether the responses were "This will happen" or closer to Magic 8 Ball style "consider X".

Side question: I remember reading something about toxic volcanic gasses being present at the actual temple site/where the oracles were housed/where they performed their prophecies. Is there any weight to this type of claim being related to the account of the oracles states of mind?