Testing Tarot
Do Tarot readings have predictive power? Probably not, but here is a sciency way to find out.
Start with a tarot reader, an experiment handler, an interviewer, and n victims (say 20) who will get their tarot readings done. The tarot reader does readings for all of the 20 people, but (here is the important part), the reader does not reveal those readings to the people. The people are not allowed to see which tarot cards are flipped over, and do not get to see the interpretations of those cards. Instead, the reader documents each reading and puts them in sealed envelopes.
For best results, the tarot reader should have a timeframe for their predictions in mind. Say it is two months. Everybody goes off and lives their lives for two months, and then they reconvene.
The tarot reader and the experiment handler know which tarot readings correspond to which people, but the interviewer does not. For each person, the experiment handler bundles the actual reading with k-1 other readings chosen uniformly at random. That gives a list of k readings per victim. I suspect a value of k between 3 and 5 might be ideal. Let's say it is 5.
Now the interviewer meets each victim, and presents 5 of the readings to this person. 1 reading is the actual one, and 4 are selected uniformly at random from the other 19 victims. The interviewer asks the victim to reflect back upon the past two months, and then select the reading that most closely correspond to their experience. If tarot readings are predictive, then victims should select their own readings at a rate better than random chance. This would provide evidence that tarot readings have some predictive power.
I do not pretend that this experiment design is original, only that I thought of it and decided to record it in my blog.
Tarot readings are appropriate to test this way in a way that fortune telling might not be, because a tarot reading just depends upon the tarot reader and the cards they choose. Maybe interacting with the victim can make tarot readings more effective, but the tarot reader can (I hope) ask questions of the victim without revealing the actual reading at the time, or giving the victim information that would help them select their tarot reading afterwards. My understanding of fortune telling is that it is much more interactive, so the process of having a fortune told will necessarily leak information about the prediction.
Note that tarot readings might be useful even if they have no predictive power. If somebody gets a tarot reading which is revealed to them right away, they might live their lives either in accordance with the prophesy or in resistance to it. Either could be useful, because (most) tarot is administered by a third party who is not privy to the thoughts and tiresome repeated habits the victim would ordinarily fall prey to. That outside perspective can potentially help victims reshape their lives even if tarot is not predictive.