mmcirvin


Matt McIrvin's Steam-Operated World of Yesteryear


Previous Entry Share Next Entry
P < 0.05
mmcirvin
Surely it is more than a coincidence that I read this Wired article on the history of positive results in parapsychology and Cosma Shalizi's musings on the neutral model of scientific inquiry on the same day.

I was wondering how long it would take someone to make that connection.

Some of the comments on the Wired article are kind of a hilarious duel of the crackpots.

Obviously, if these experimental results are correct, it is not because the subject actually sees or senses the future (it does not exist), but because the subject (or even the experimenter) is somehow, unknowingly, using their minds to affect certain outcomes as they happen.

OBVIOUSLY

The Wired article takes a strange, noncommittal attitude; you can read the whole thing to imply "there's nothing there" or you can read it to imply "these isolated results strongly suggest that psi is real, but there was always something wrong with the procedure that kept those darn scientists from satisfying their fetish for reproducibility, but it sounds like Bem has finally got it right this time."

Which is funny, because back in the 1990s on Usenet, I remember Bem's claims about Ganzfeld experiments getting a lot of attention as psi's final big break into scientific respectability. Notice what was going on there--he did meta-analyses of the literature on these sorts of experiments and concluded that one should be convinced that there was something there, even though the protocols varied all over the place and you couldn't point to any series of experiments as really clear-cut replication.

It's an interesting question, though:

notr

2010-11-23 02:05 pm (UTC)

How would you design an experiment to distinguish between various claimed psi powers? If the supposedly sensed future events are not predetermined, are they sensed or controlled? If they are predetermined, are they sensed in the future occurrences or in their present determinations? If, say, you selected a cypher in advance and a key after the predictions were made, then the subject would have to both sense the cypher and control the key in order to simulate precognition--unless, of course, he actually manipulated your design of the experiment before you even knew you would perform it.

Re: It's an interesting question, though:

mmcirvin

2010-11-24 12:24 am (UTC)

It's really hard. Part of the problem is that psi powers are so vaguely defined, and the broader you cast your net the harder it is to tell signal from noise. If you're testing whether somebody has some kind of omni-temporal clairvoyance and can sense an event that could happen in the past, present or future, how can you determine correlations?