Unbelievable Project: The Probability of the Resurrection

In #Probability #Religion #Unbelievable Project #Unbelievable Project

The Probability of the Resurrection - Calum Miller & Chris Hallquist - Unbelievable? - 06 July 2013 – Is the resurrection 97% likley as Swinburne claims?

As part of the Unbelievable Project, I am taking notes and “arm-chair” responding to each of the Unbelievable podcast episodes satisfying a set of simple rules.

See here for a full RSS Feed of the podcasts.

Description of Episode

  • Full Title: The Probability of the Resurrection - Calum Miller & Chris Hallquist - Unbelievable? - 06 July 2013 – Is the resurrection 97% likley as Swinburne claims?

    Christian philosopher Richard Swinburne has used probability theory to show that the likelihood of the resurrection of Christ is 0.97.
    Calum Miller is a Christian apologist and student of Swinburne. He talks about why he believes that probability theory can be used to show that the resurrection is highly likely to be true.
    Chris Hallquist is an atheist blogger who argues that the resurrection is not well supported by evidence or probability.
    For more debates visitwww.premier.org.uk/unbelievable
    Join the conversation viaFacebookandTwitter
    For Calum Miller http://www.dovetheology.com
    For Apologetics UK http://apologeticsuk.blogspot.co.uk/
    For Chris Hallquist http://www.patheos.com/blogs/hallq
    Get the MP3 podcast of Unbelievable?http://ondemand.premier.org.uk/unbelievable/AudioFeed.aspxor ViaItunes
    You may also enjoy:
    Unbelievable? 16th April 2011 - Biblical evidence for the Resurrection - Bart Ehrman & Mike Licona.
    Unbelievable? 7 April 2012 - Are the Jesus Scandals evidence for Easter? David Instone-Brewer vs Bob Price.

Download mp3.

  • Justin Brierley - Christian Moderator
  • Calum Miller - Christian
  • Chris Hallquist - Atheist


Me - I was really looking forward to this episode. What was not to like? Probability theory, ancient religions, evidence for Christianity…bring it on! Unfortunately, it really wasn’t that impressive.

Calum - “There’s what’s called the confirmation of resurrection, the explanatory power. And this is basically the idea that there is a lot of evidence which, if the resurrection happened would be expected but if the resurrection didn’t happen, it would be very improbable. And if this is true, if there really is that kind of evidence, then it follows from probability theory that our confidence in the resurrection should be greatly increased by this evidence. [Concerning the prior], more extraordinary or extreme events are more improbable to begin with, and so you would need more evidence to confirm them. So a lot of the debate about the resurrection comes down to the prior probability, whether we think it is actually really improbable and that no possible evidence could ever make us convinced of it. ”

Me - He basically has the distinction between the following as the basis for all of the “calculation”:

  1. evidence that, if it existed, would be very common if the resurrection did happen
  2. evidence that, if it existed, would be very rare if the resurrection didn’t happen
  3. the prior probability for the resurrection

where he admits that "the debate about the resurrection comes down to the prior probability". Anyone doing probabilistic inference knows that it should never come down primarily to your choice of priors. The data needs to rise above the prior, and the prior needs to be an honest -ideally objective- assessment of the pre-data probability assignments or, often, the initial state of ignorance. By admitting this, Calum is essentially saying either that:

  1. the data are not strong enough to constrain a diffuse prior, and thus is unconvincing or…
  2. you have to come into the debate with a sharp prior which admits to a presupposition of the strength of the claim..

Neither of these stances is convincing in the slightest.

Further, in response to this set up, he ignores the most important thing in any Bayesian treatment is the set of models that you are using to compare. You cannot simply test the truth of a single model in isolation, nor is it generally informative to compare model A true or false. Instead one wants to set up a list of models, hypotheses, theories to explain the data and evaluate those multiple models. Instead of,

$$ P({\rm resurrection}|{\rm data}) $$


$$ P(\mbox{not resurrection}|{\rm data}) $$

you’d want

$$ P({\rm resurrection}|{\rm data}), P({\rm hallucination}|{\rm data}), P({\rm legend}|{\rm data}), P({\rm literary}|{\rm data}), P({\rm hoax}|{\rm data}),

$$ etc…
where of course each of these models would have many details beyond the simple label I’m putting in here. By being explicit with what you’re comparing to, it is easier to see where the different prior probabilities come in. Are you really going to suggest that someone rising from the dead is on par, prior to the data, with a legendary construction given how many legendary constructions we’ve seen and how many dead rising we’ve not seen?

What is clear is that all of these other models must, a priori, be more probable than rising from the dead even if a God exists. Just because you believe miracles could happen does not mean that you believe every miracle claim is true, and given the number of clearly false miracle claims, the prior probability for any miracle claim must be quite low - even if you believe miracles actually occur.

Another point about the data which Calum never deals with is that it should include things we don’t see, not just things we do. If we expect something to occur with a claim, and we don’t see it, that is in fact evidence against the claim.

Chris - Most Christians might discount the claim that the miracles around African religions seem to disappear in the US and UK because of lack of faith. Or perhaps the miracle stories around Mormonism. What makes the miracles of Jesus different than these ones? Once you accept the idea that resurrection claims can exist quite commonly in a group of religiously charged people, it is no longer quite so hard to understand the resurrection claims in the Bible.

Calum - The reports of an empty tomb are exactly what you’d expect if the resurrection actually happened, and would be unlikely in the case of a non-resurrection event.

Me - Dealing with this is actually very simple. He is correct that if the resurrection occurred, then the report of an empty tomb would very likely be given, and I would add that it would also be very likely to be reported in the earliest accounts we have of the resurrection. Is this what we see? No! The empty tomb is not mentioned in Paul, neither are the physical visitations, both of which you’d expect to see if the Resurrection actually occurred. Even the visitations are not mentioned in Mark! So, from a probabilistic point of view, this is the exact opposite of what we’d expect to see if the resurrection actually occurred. In fact the descriptions of the resurrection get more elaborate and more physical the later the text (Paul has visions, Mark has no visitations but the empty tomb, Matthew and Luke have visitations, John has the doubting Thomas story, etc…). This is exactly what we’d expect for legendary development, or a story that has been embellished over time.

The other thing, is it really all that unlikely to have an empty tomb story with no resurrection? Notice, I’m not saying to have an empty tomb, but to have an empty tomb story. There are several different routes to get that. One is as a literary device. I believe Richard Carrier supports this, as a reference to Daniel. Another is a deliberate counter to Docetism, to gain favor and win an argument.

Calum - “It is not necessarily helpful to have just some kind of religious context, it must be the right kind of one. So, for example, I can see very good reasons why God would want to vindicate Jesus’ teaching by resurrecting him because I think Jesus taught a lot of very good things, I thought he was (obviously as a Christian) I think he was a very sincere, a very good person. And I can think of a lot of good reasons why God would choose Jesus to be a prophet and to become incarnate in him. Whereas I don’t see comparably good reasons why God would want to vindicate Mormon teaching. Obviously a lot of that is because I don’t know a lot about Mormonism but there’s still the asymmetry there.”

Chris - The positive evidence for Mormonism is a lot better than for Christianity. We have signed documents by the early followers and founders attesting to the miracles. The best we can say about Paul’s evidence is that he had a vision. We have a lot of negative evidence for Mormonism, to be sure, but if we knew more about Christianity perhaps things would be different.

Me - I would add that we have this pro-Christian filter for all of our documents, a filter called the Middle Ages, where documents supporting Christianity had a much better probability of surviving (i.e. copied) than ones critical of Christianity. The only reason we have the Nag Hammadi texts is that the monks refused to burn them, as ordered by the Christian orthodoxy at the time, and instead chose to store them in a cave. Think about that campaign of whitewashing for hundreds of years! Actually, the fact that we have so little actual documentary support for Christianity coming from the first century, despite this huge bias, to me argues against Christianity.

Chris- How do you know Jesus was sincere or not? Seems like the same could be said for Joseph Smith.

Afterward - a bit about priors

(this section is all Me -, so I won’t put it in bold.)

I don’t really think that when Calum is referring to priors that he really means that in the same way as before the data. It seems to me, and I believe Swinburne’s analysis reflects this, that the prior for this calculation is really the posterior for a previous calculation regarding the existence and properties of God. This is perfectly legitimate Bayesian procedure, but it makes the argument a different one. Because of this, Swinburne’s calculation for the probability of God needs to be addressed before we can even deal with the priors in this resurrection argument. That will have to be another post entirely, but at any rate Calum did not do it a service in this debate, having not really gotten to the meat of it when he could have.