How To Twitter Your Way To A Remote Psychic-Reading Victory?

Now, Joel, Andy and Jack have all begun asking questions on Twitter – so, I have started asking more questions myself about the Dr Richard Wiseman ‘Psychic Twitter Science Experiment’ or whatever he and New Scientist are calling it. Psych-twitters? Anyway, it’s the one I noticed back in this post, Twitter If You’re Psychic? Or One In 625?

Of course, people will immediately note that naturally all of this blog-post clearly stems from my frustration that all of my answers to every round will be as follows and I’ll just be wrong every time. :p

So, if you wish to consider the study as ‘harmless Twitter-fun’, you’ll have a case straight-up. But I’m still interested in talking about some things that I’ve found in the meantime.

First Round: Standing in a public toilet with a small toy bunny and a poster for Flight of the Conchords. Police are on their way.

Second Round: Knee-deep in a bucket of salmon, on the top of a skateboard in Trafalgar Square. Singing Bowie songs from the 80s.

Third Round: Backstage at Hertfordshire’s Faces Nightclub, near the complimentary bar with an emu. Avoid the martinis.

Fourth Round: Up a tree with Geri Halliwell, holding a Wii Fit board. Avert your eyes.

Naturally, I’ll be proved wrong every time. But just looking at the first round has already given me an idea. Here’s the first series of photos:

Just Twitter every time something vague enough to be interpreted no matter whether it’s a ‘natural setting’ or a ‘man-made featured setting’.

Why? Because the first round already had a mixture of things like a bushland path, a fence, a house door, a car-park and a building… and as many of us know, the more generalised your guess is, the more likely you’ll get a hit. The word-cloud by Tom Simonite at New Scientist magazine that was posted demonstrates some of the generalisations that could help – prominent are words like ‘grass, buildings, water, trees, parks, green, blue’.

When a person doing a psychic reading starts, they work on generalisations. Letters of the alphabet, talking about a loved one who is gone, or even just throwing out common ailments that lead to a tragic end, like cancer, heart-troubles or even ‘being taken too soon’.

Of course, we’re assuming that the Twitters about the location were really being taken on board during the survey. You didn’t have to have a Twitter account to click on the survey – you just had to see when Dr Wiseman said ‘go!’ on Twitter.

Now it appears that it includes what Dr Wiseman calls: “The wisdom of the crowds concept. As well as looking for an overall effect, it will also be interesting to see how certain subgroups, such as those who indicate they have psychic ability and are very confident about their decision, have voted.”

So, it could include whether people catch onto the notion of ‘the more general you are, the more likely you can get a hit’? I’m still not sure if people who ‘Twit’ their response are going to go to the survey, or vice versa.

Could the survey included a section where you could put in your Twitter name? It didn’t seem to have that feature when I popped my (wrong!) answer in – I chose E when the first round answer was D (so, clearly this is all just bitter regret on my part!!).

One question that wa posed to me via Twitter:

Joel: “I still don’t see how large number of participants can influence results”

Okay, in a standard test (not one that is now saying that it’s drawing on what Dr Wiseman is now saying he’s testing – ‘the wisdom of crowds’), there can be an issue if you have too large a sample size. It’s sometimes related to a ‘Type I’ error, which means that you have [alpha symbol] error, or a “false positive”: the error of rejecting a null hypothesis when it is actually true.

A null hypothesis is a statement of the status quo – one of no difference or no effect. It’s always the hypothesis that is tested, referring to a specified value of the population parameter, not a sample statistic. If the null hypothesis is not rejected, no changes will be made. Given a sufficiently large sample, what are extremely small and non-notable differences could be interpreted as statistically significant, which may say nothing about the practical significance of a difference. With a huge sample, there could be a piddling little effect that is significant only because of sample size, but which amounts to bugger-all as any real finding.  A lot of stats these days is focusing more on effect size than on statistical significance.

You may have noticed that this experiment has managed to get into a great many media releases (Dr Wiseman has said “The Guardian, The Daily Mail, The Daily Telegraph, Sky News, Nature Online and Fox News“) on the internet, which should boost numbers of people taking part.

We haven’t been given how many have taken part so far – but since a video like the ‘Colour-Card Changing Trick‘ has got over seven million viewers, we can see how a bit of viral buzz can have an influence on the overall number of participants!

Since I have a computer scientist handy, I thought I’d ask a few more questions of my own:

Q: How many Twitter accounts does one need to guarantee a ‘majority’ number?

A: Infinite? Two billion Twitter accounts could conceivably all choose the same (wrong) answers. However, if you can assume the choices are random, from approximately 37 Twitter accounts you’d expect one of them to have gotten at least 3 guesses right (if you used a somewhat vague enough guess).

[I should point out at this rate, now if we consider making general statements in a Twitter comment during the test and / or simply IGNORE using a Twitter account and just go back and forth to the survey (using different computers?) one could probably vote as many times as one liked too!].

Q: Dr Wiseman says that “A bit before 3pm UK time I will randomly choose one of the five locations as the target. To help prevent possible bias, this will be carried out using this site. This offers true random numbers based on unpredictable atmospheric noise” – is that really necessary? Could someone still be ‘tipped off’ somehow and pass on the news as to where he is?

A: There is no particular need for him to pick it; if he’s going to be picking it using a genuine entropic source of randomness (gobbledegook for “true random numbers”), and since a computer program will be judging it, there is no need for any human to know the “correct” answer before time.

If he knows it, then there is always the possibility that he may give away a cue as to the right answer to somebody. That is, of course, the principal behind double blinding.

Q: He’s said that he’ll travel to a location – can such a location be traced over the internet?

A: It is, of course, possible to back trace the origin of the Twitter message to narrow down the location. Tricky, yes, but certainly orders of magnitude easier than breaking the laws of physics and causality by determining the answer paranormally. It would be better to have the computer that chooses the random number to also send the Twitter message, without anyone ever having to travel. But that, of course, depends on what he’s trying to test – if he’s trying to test people’s power to predict his location, obviously it’s necessary that he is there.

A: This is perhaps the biggest flaw, as it is easy to trigger a denial of service attack by flooding the site with votes. A “genuine” psychic might be unable to compete with a prankster that writes such a program.

A: So the above Denial of Service is even better. All you need to do is programmatically vote over and over again for the same photograph and overwhelm anyone else’s vote.

A: This is correct; that’s the reason that you’d expect to see a hit from 37 people. I’m a LITTLE concerned about the shout out to “Charlie” here – this is a TRIVIAL calculation and I don’t see why he’d have to get someone else to do the math.

Q: Eh, maybe he was being funny, I don’t know! Finally, Dr Wiseman says: “If this happens, then I would consider the experiment significant, and suggestive that something strange is going on. This would not prove remote viewing exists. As with any new scientific procedure, the result would have to be replicated for that to be the case (this is especially the case when one is dealing with something as controversial as psychic ability). It would, however, certainly be curious and deserving of future research.” Any response you have to that?

A: I disagree. This doesn’t sound like a good test of remote viewing at all. I’ve never heard anyone claim that remote viewing was a gestalt experience – why would you expect the collective intelligence of the Internet to be more capable of remote viewing than a single gifted individual? And this is assuming the flaws are unimportant.

Assume, for example, that the phenomena exists. Unless it’s remarkably common, the few genuine psychics are going to be overwhelmed by the non-psychics. So you’d expect even that to show nothing.

If it DOES show something significant, then you’d have two competing explanations. One, that the phenomena exists and in sufficient numbers to overwhelm those that cannot exhibit the phenomena. Two, that someone has exploited a flaw in the procedure and either gotten lucky (a 1 in 37 chance with a denial of service attack isn’t that unlikely). This is a publicity stunt and nothing more.

Well, what I do know is that I should pop up challenges for people to have me discuss topics more often. So, I’ll see if I can use Twitter myself in such a fashion in the future.

Oh – quick mention of Benjamin Radford’s take on it all:

Like all experiments, Wiseman’s test has inherent limitations. The participants are a self-selected, volunteer subset of Twitter subscribers, and any result may not generalize to the population at large. The experiment cannot (and is not designed to) conclusively prove or disprove the existence of psychic abilities.

Still, thousands of Twitterers are expected to participate, and if nothing else, it is an interesting way to get the public thinking about how scientific methodologies can be applied to paranormal claims.

Tagged as: Dr Richard Wiseman, New Scientist, Paranormal, psychic, psychic twitter, Quirkology, remote viewing, Science, statistics, twitter