Skip to content

These are not the probabilities you are looking for

There has been an increasing buzz in the papers regarding the impending launch of the Large Hadron Collider (LHC). Some of this concerns the possibility that it will lead to a disaster which destroys the world. This certainly sounds unlikely, and people who seriously suggest this are typically brushed aside with official calculations about the chance that the LHC will indeed destroy the world in any of the ways that have been suggested. For example, it is said that the chance of it destroying the earth though the creation of a particle called a strangelet is only about 1 in 50 million and the chance of it creating a black hole which does not evaporate is much less than this. However, these are not the probabilities we are looking for.

The problem is that the calculations don’t consider that the physical theories they are using could themselves be incorrect. For example, a hundred and twenty years ago, the scientific consensus held that Newtonian mechanics was the ultimate physical theory. If they had to calculate the chance that an experiment could lead to the curving of space and time, they would have said there was no chance at all. Indeed they would have also calculated that there was no chance of modern electronics or lasers existing, since both are impossible classically. They would have been at least as certain of this as the directors of the LHC are, and they would have gotten it wrong. We could be in just such a situation and with the highest possible stakes at risk.

There is, however, one large dissimilarity between now and then. In the late 19th Century, there was a huge amount of evidence in favour of Newtonian mechanics and only a few nagging lose ends that hadn’t been explained. Now, however, we are genuinely uncertain about our physical theories. Indeed, we are so uncertain as to spend more than 3 billion euros building the LHC in order to find out more. Moreover, we know that our current theories are false because they don’t correctly merge Relativity Theory and Quantum Mechanics. That is, we know that we don’t presently understand what happens with tiny objects that are extremely dense and/or moving near the speed of light. Since this is exactly what is occurring in the LHC, we have significant reason to distrust the probability calculations. They tell us the chance of the LHC destroying life on earth given that the underlying theory is completely correct, but what we really want to know is what the chance is given our uncertainty in the underlying theory. This is impossible to calculate precisely, but will be much higher than the stated odds. Considering the stakes, it is thus highly irresponsible for the LHC’s management to give so much emphasis to these misleading probability calculations, when the real chance is clearly higher.

Share on

4 Comment on this post

  1. If CERN were to follow your advice, what would their risk assessment consist of?

    It seems that there are four main ways of estimating the probabilities. The first is to do a straightforward calculation based on current theories. This would be completely vulnerable to your criticism. The second is to use solidly tested physics, like parts of the analysis in (Dar, Rujula and Heinz 1999). This is vulnerable insofar “new” physics could ignore known physics. The third way is to use observable properties of the world to infer risk limits; this is the main push of the Dar paper. This seems to be less vulnerable to your criticism if done properly (i.e. the moon argument is spurious while the Dar supernova argument seems solid). It is however vulnerable to anthropic bias. The fourth way is to try to use anthropic bias to bound the risk (the Tegmark and Bostrom paper). Are there any more?

    In any case, just having a probability – even a well grounded one – is not necessarily the right way of estimating risk. I think Adrian Kent has a point in his risk assessment critique, although I’m not sure I agree with the way he then goes about it. Maybe one could do the above split for risk assessment too: based on our estimates of how bad it would be/the benefits we could get, based on empirical observations of past performance, or some form of anthropic reasoning (anti-bias reasoning?) applied to assessment rather than probabilities?

  2. Anders, I am not asking for a precise quantification of the probability, but for honesty and transparency about the usefulness of their numbers. I understand that they want to be adversarial about this, showing their arguments’ strengths without their weaknesses, but with these stakes it is highly irresponsible to do so. They need to be honest about the fact that their safety depends upon the many speculative claims in the shaky intersection of quantum mechanics and relativity. They also need to ask scientists to challenge their analysis to be sure it is robust, but it appears that they have instead asked their scientists not to comment on its safety…

  3. Gregory Eckersley

    What is the chance that the LHC will ever create an event which is not common amongst natural events
    created by high energy cosmic rays? It would seem the purpose of the LHC is not to create events which
    do not occur naturally, but to create them in a controlled environment where there observation is much easier.

    The principle that experiments should be carried out mindful of risks seems genuine, and perhaps the ethical dimension has been sadly neglected in research. Better areas for such concerns might be gene-tech and weapons AI than the LHC. The fact that the LHC is an enormously expensive and grandiose experiment does not mean that its risks are higher than a mucher cheaper and apparently more innocuous experiment in
    some other area.

    Another possible (and more likely) risk from the LHC (as from many other experiments) is that the knowledge gained from it is dangerous or open to misuse. The argument that ignorance is a defence may be valid, but it hardly seems to have been consistent with human behaviour over any extended period of time.

    Scientists will have to prove to society that their research will lead to genuine social improvement to avoid being seen as obscure at best, or threatening at worst.

  4. I strongly suggest, Toby, that you read the astrophysicist Rainer Plaga’s paper arXiv no.: 0808.1415v2, that claims there is a theoretical basis for a danger scenario, based entirely on published physics papers from the relevant string theory in related to micro black holes. Read also his response to CERN criticism at the end of the paper. For clarifications, Plaga gives his email address on his website at

    I have a number of arguments myself with certain details within the safety papers (specifically regarding black holes) by CERN and others.


Comments are closed.