predicting earthquakes: hedging your bets

Fellow SciBloggers Peter Griffin & David Winter have recently written posts on Ken Ring’s (in)ability to predict earthquakes (here and here respectively). This is something I’ve also touched on earlier & I thought I would follow up on it now – the specific issue I want to address is that of hedging one’s bets.

Last year Ken wrote about the possibility of a ‘moonshot’ affecting New Zealand – a major earthquake triggered (he believes) by the close approach of the moon. (As David’s already pointed out, there appears to be no such causal link.) When I asked him about this on my own blog he seemed to pull back from his initial statement, saying instead (here in the comments thread) that

As to the 20 March, the “one for the history books” comment came from an interview with Marcus Lush. I was just agreeing with him to be obliging. I do not hold that 20 March WILL bring a severe earthquake to Canterbury, but an extreme weather event is possible that day worldwide, and an earthquake within 500kms of the Alpine Fault is a risk on that date. More likely to be a 4-6mag.

In this comment he appears to be pulling back from his original statement (& being a little disingenuous – the "one for the history books" comment was also made on the Silly Beliefs blog, for example). He’s also hedging his bets – the phrasing here is so vague that just about any untoward event would count as an accurate prediction. Just what would qualify as a "severe weather event"? Will it really be "worldwide" – as in the sense of "global" – or is this statement intentionally broad so that any instance of severe weather on March 20th, anywhere in the world, could thus count as a ‘correct’ prediction? (What is "severe", anyway?) I have asked Ken for clarification on this, on the SciBlogs mirror of my blog, but so far – despite him saying there that "I never shy away from questions" – I’m afraid I’ve had no further clarification.

The same imprecision is present in the statement about the risk of an earthquake "within 500kms of the Alpine Fault". Assuming we’re talking about 500km radius (again, I’ve had no clarification from the author of this statement) – well, heck, that’s a huge area: all of the South Island and a fair bit of the North as well. And we have earthquakes all the time in New Zealand – the country’s not called ‘the shaky isles’ for nothing! So again, this simply cannot count as a prediction; it’s simply stating the obvious. (And in a recent post on his blog, Ken has said that the Alpine Fault itself seems to be fairly inactive at the moment. However, as we have said, it could be anywhere in NZ, or it may not even happen at all. Again, this is extremely vague and effectively covers all possible options.)

There are other inconsistencies in Ken’s various postings. For example, on October 13th he’s on record as saying that it was unusual to have one large quake followed by another in the same area, so that there was probably little risk of another large earthquake following the September 4 event. More recently, with his claims to have predicted the terrible event of a week ago, he seems to have reversed this position. Leaving aside the accuracy of that claim, there are other questions that I would like to see answered. (After all, when scientists change their minds they are up-front about the reasons for doing so.)

And as you might expect, my questions relate to evidence: what’s changed, between those two oh-so-different predictions? Ken Ring claims that his ‘predictions’ are based on his reading of the position and movement of the Moon, planets, and stars (on astrology, in other words – his characterisation of the Moon as a planet would bear this out). But the Moon is still in its usual place, the planets haven’t changed their courses, and if the stars do move it’s going to be a very long time indeed before we know anything about it. So what was the basis for his reversal of opinion? I’ve asked, but to date I’ve had no answer.

Various commenters – on SciBlogs pages, Silly Beliefs, the comments relating to last night’s Campbell Live interview, & elsewhere – have said that in their opinion Ken Ring’s ‘predictions’ are accurate; they’ve seen them come true. The problem is though, that such vaguely worded ‘predictions’, with their lack of precision, are so wide open that they lend themselves to confirmation bias. Something happens, & if it falls within the wide boundaries of a particular claim, then there’ll be a tendency to regard that claim as having been accurate.

So what’s it to be, on the 20th March? A quake in the South Island – which given the dreadfully high frequency of aftershocks from the latest event is effectively a given, or somewhere else in NZ? Somewhere in the ‘Ring of Fire’ that runs along the boundaries of the Pacific tectonic plate? (This is not an idle question – Ken has said that the 7+ [which he was ‘predicting’ for February] is sure to be somewhere in the "Ring of Fire", where 80% of all major earthquakes seem to occur, and NZ is at the lower left of this Ring. There are earthquakes along that plate boundary all the time, so any event would have been viewed as a successful ‘hit’.) A worldwide ‘severe weather event’, or maybe a ‘severe weather event’ somewhere in the world. Take your pick. But please don’t tell me that such ‘predictions’ could be construed as accurate.