I see, I'm glad we seem to agree on many points. But,
Let's make a thought experiment:
Assuming, for the sake of the argument, that 1 in 500 Indians likes, for whatever reason, to have fun messing with tourists by telling them things he/she 'knows' about their lives. Now, if each of these jokesters during their lifetime hit up 5 tourists, this means that as many tourists are approached as the population of India divided by 100. And that's 10 million.
So we have 10 million predictions. This means, that 1 million predictions will correct for things as unlikely as 1 in 10 odds. That could for instance be guessing one's star sign right (odds: 1 in 12). By the same reasoning, there will be 1000 predictions with likelihood of 1 in 10'000 of being right. This is like coin flipping 14 times and always getting heads -- pretty impressive, right?
And there will, by mere statistics, be likely one correct prediction with a 1 in 10 million likelihood of being true. That's like flipping coins 22 times and ALWAYS getting heads. This would probably translate to correctly guessing one's birthday (not even figuring into the possibility that the Indian could already have spied and found that out) and phone number / some 6 digit number.
As you see, there's a good reason why its problematic to infer whether a method works by mere anecdotal evidence. Even if you happen to be impressed by 20 consecutive head tosses with a non-fake coin, beware that this will happen to about 4 people by chance anyway (given our assumptions, which are rather conservative actually), and you could well be one of these four.
If all our jokesters would only guess birthdays, more than 20'000 tourists would have their birthdays guessed correctly. The other 9 million 880 thousand tourists where it was wrong wouldn't think much about it and forget the instance, while of the 20'000 where it happened by accident, many will probably be greatly impressed and believe something supernatural might be going on. In fact, the 40'000 people where the date was just 1 day off might well think that there's something magical too. And the 600'000 people where the guesser hit the right month/starsign might well think the same thing -- after all, wasn't it uncannily close to the actual date?
People's notions of probability are often wrong, our sense of statistics evolved for a world where our ancestors lived in groups with maybe 30 individuals. When the amount of individuals involved goes up in the millions, our gut-feeling fools us.
By all means, be open minded. Yet that doesn't mean one should blindly accept things without (valid) evidence!
This thought makes no sense to me. Science is dominant because of its track record. Science produces results. We've been to the moon, we've found particles smaller than nuclei by smashing protons together in an underground 26-kilometer ring at nearly the speed of light. Can astrology, or Gaia, do that?
The whole reason behind science's suggest is that it's evidence based, and ready to discard wrong hypotheses in favor of new ones. Why on earth would we go back to non evidence based methods? Examples would be belief in authority (whatever my teacher/parents/pastor/Mr. President says is certainly right), revelation (whatever the next prophet says is certainly right, even if it's another Hitler) or simple economics, where propositions are true when people pay money to read about them (i.e. homeopathy, astrology etc)...
http://3.bp.blogspot.com/_CRtMt721qA...vsreligion.png
'Faith' can stand for any non evidence based method. Clearly, there's no way around science. By the way, even Gaia theorists think that they're relying on science. Even creationists (with some tragic exceptions) think (or at least pretend) to have evidence for their claims.

