Reader Comments

Post a new comment on this article

Interesting but potentially flawed

Posted by acoolyam on 27 Jan 2016 at 17:25 GMT

I fully support the science and the idea behind this research, however I don't think the maths takes the full picture into account especially regarding items such as the moon landing vs the NSA revelations as one example.

Firstly the NASA moon project was conducted during a highly tense cold war, where most governments and associated population were a lot more schooled into keeping "mum". So if we consider the NASA moon project, with 400,000+ people. If this was to be secretive because it was potentially staged, then NASA and the Government would have created multiple silo's of teams and people. With each team focussing on their element and not aware of everyone elses (rather like the Bletchley Park example, where even staff in a hut didn't necessarily know what they were doing the work for, let alone what other people in other huts did) really keeping the need to know basis strict. This was you could reduce the amount of people aware of any potential conspiracy to the low 10's and whilst some may pass away due to old age, you might wonder if looser lips may have been encouraged to silence in a different way during those times..

Then consider the NSA whistle blowing, which occured during a totally different times and encompassed different demographics and technology. People assume far great openness and governments have also expanded their "need to know" to greater audiences (perhaps they regret some of this), the internet and social media has kind of spread information (positive and negative) which perhaps skewed the view of Mr Snowdon to share information which would potentially never have been shared even 20 years ago.

I do not think the science has viewed the changing in times, attitudes, technology, politics fully within the maths.

This is just my personal view, I am open to all discussions.





No competing interests declared.

RE: Interesting but potentially flawed

cdelgesso replied to acoolyam on 27 Jan 2016 at 17:39 GMT

"All models are wrong, but some are useful." - George Box

Grimes doesn't claim that this is THE model to use regarding conspiracy theories. He notes the limitations to his work in his paper, and points to avenues for future research. The 'maths' is 'just' math. Grimes "does not consider the dynamics, motivations and interactions of individual agents."

No competing interests declared.

RE: Interesting but Potentially Flawed

Laurence replied to acoolyam on 11 Dec 2016 at 13:47 GMT

Only got round to reading this very recently and I have to say that I am as dubious as you are with respect to Grimes's paper. The following may be of interest in this regard.

Grimes equates the average number of events per time interval with the, "intrinsic probability of a conspiracy [being exposed] per person per unit [of] time", p. Personally, I would have been far more impressed had the author conducted some proper research and given an accurate figure for the number of conspiracies that have actually been exposed over the last 100 years; nevertheless, let us continue.

Interestingly, assuming that the Poisson's distribution as presented in the paper is applicable to the problem (I am not certain that it is, as has been shown by other contributors), we can introduce a parameter into Poisson's distribution to take account of the two main contentions that you had with the paper. These variables can be summarised as follows:

(1) Compartmentalisation (the ratio of the number of individuals 'on the inside' to the total number of individuals working on the project);

(2) Peer pressure problem (the ratio of the number of individuals who will not give in to peer pressure to the total number of individuals involved).

The atom bomb example would suggest that no more than 1 in 1000 are essentially 'in-the-know' but we shall say 1/100 to be conservative. As for peer pressure, a ratio of 1 is to 10 would be considered conservative. Thus the probability of (1) and (2) occurring (i.e. a soul emerges that is both on the inside and unafraid of the repercussions of putting their head above the parapet is 1 in 1000. (1) and (2) are both per unit time.

Therefore, what we should really be discussing is the, "intrinsic probability" of a conspiracy being exposed per 1000 persons per unit of time. Inputting our ratio, the effect on both Poisson's distribution and the time factor of introducing the '1 in 1000' parameter is of the order of 1 in 850.

No competing interests declared.

RE: RE: Interesting but Potentially Flawed

drg85 replied to Laurence on 12 Dec 2016 at 00:19 GMT

Hi Laurence, acoolyam and cdelgesso,

I'm the author and apologies, only seeing this now. As cdelgesso notes, the model presented in inherently rather simple and I would be delighted if people took it further - as I note in the paper, there's a number of limitations to this model but as far as I am aware it had not been attempted before. I do invite others to contribute new models too, especially as cancer research is my main area and I'm limited in how much time I can spend pushing down this avenue. Let me try to answer your main point, and then maybe suggest something:

"Grimes equates the average number of events per time interval with the, "intrinsic probability of a conspiracy [being exposed] per person per unit [of] time", p. Personally, I would have been far more impressed had the author conducted some proper research and given an accurate figure for the number of conspiracies that have actually been exposed over the last 100 years; nevertheless, let us continue."

I actually would have loved to do this - the problem is "a priori" that we cannot know if a conspiracy has taken place because a conspiracy theorist will just argue we didn't detect it. As one of my aims was to write this to help illustrate to people sympathetic to conspiratorial views how unlikely these are to manifest in science I was aware this argument would be employed. What I did instead with the parameter ""p"" was to come up with a rate constant. I did use three historical conspiracies for which I could estimate numbers - as I point out in the paper, I wanted to be be overly fair to conspiracy believers and so I made the assumptions to give the greatest possible change for a conspiracy to remain secret - even though I do not think this is realistic, I wanted to take a "worst-case" to illustrate this.

I later on varied the ""p"" by a few orders of magnitude to show that the issue was that even if people are incredible secret keepers, the reality is that the number of people who would have to be involved in a scientific conspiracy is massive - more so that other fields, scientists falsify each others work, and this is how scientific misconduct is exposed. If scientists were lying about climate change, or vaccination, it would require 1000s of people to be complicit - and this number is what screws up a conspiracy, more so than individual ability.

I should point out ""p"" also captures the idea that leaks may be accidental; if 1000s of scientists are engaging in a hoax, and even if they all agree to some code of omerta, there's every chance that a mistake rather than a whistleblower lets the cat out of the bag - if I accidentally emailed an incriminating document, that would suffice for exposure. I wish I had been able to measure this, but my analysis is the paper is subject to a number of caveats precisely because of the argument you can't measure mass conspiracy as it's hidden.

I like your ideas about compartmentalisation and peer pressure - I think these would fit very well into an agent based model, where I think dynamics could be considered and each agent given a spectrum of traits. I had considered this as it would potentially be more realistic, but it might be equally hard to estimate what parameter values are realistic. I suggest this at the end of the paper, and I'd be great if anyone wants to take this up!

Hope this helps clarify, thanks for your comments - I may not be able to reply promptly as I'm travelling quite a bit but hope this has been useful!

Competing interests declared: I am the author

RE: Interesting but Potentially Flawed

Laurence replied to drg85 on 12 Dec 2016 at 19:13 GMT

The author has given an interesting and magnanimous response to the comments herein. The definition of conspiracy, it would seem, needs to be more clearly defined; the instances of conspiracies may differ by many orders of magnitude depending on what qualifies as a conspiracy. Does a conspiracy between, for example, a small developer and a local authority official to bend the planning laws rank as a conspiracy or are conspiracies confined to the big ticket items that tend to obtain media coverage.

Moreover, although the tacit compliance of many thousands of individuals may indeed be required (thus appearing to rule out the possibility a priori), the initial condition, p0, could take account of the effects of compartmentalisation and peer pressure. In so doing, the numbers involved are decimated and sustaining a conspiracy long-term not as onerous as first seems.

No competing interests declared.