My latest book Proving History: Bayes’s Theorem and the Quest for the Historical Jesus is now available for kindle. It should come available for nook soon (and possibly other formats, whatever past Prometheus titles have come available in: e.g. check The Christian Delusion in any format, and if it’s available there, Proving History will be, too).

A note for the visually impaired: I have noticed that some versions of text-to-speech don’t read the upright bar in mathematical statements of probability (the symbol for “given”), or even the tilde (the symbol for “not”). Also, most of the equations in Proving History are graphical inserts and thus won’t be visible to text-to-speech readers at all. These problems might be solved if I do an audio book version, but that depends on my audio book publisher being able to afford the audio license from Prometheus, and in any event an audiobook release could be as much as a year away, so even at best this is not a short-term solution.

The first in-text example of both an upright bar and tilde appears on page 66 of Proving History, where you will see the equation P(h|~h) = 0, which when read aloud by a human would sound like P, h, bar, not-h, equals, zero; or: P of h, given not-h, equals zero; or just “the probability of h, given not-h, equals zero.” There are many other instances where a tilde appears or an upright bar, and this can make the text confusing when read out by a computer, if the computer doesn’t read out those symbols. All I can say is just be aware of that as you go and I hope it can be managed.

As for the graphical equations, you will miss out on the visuals those provide, but in most cases that just means you have to trust the math came out right. Otherwise, most everything is explained in the text, and you can do all the math yourself (although that is a real challenge, I know), all you need to know are three equations, which I am here writing out for text-to-speech capture:

The first is the standard long form Bayesian equation: the probability of a hypothesis, h, given the evidence, e, and all your background knowledge, b, equals the product of the probability of h given only b and the probability of e given b and h, all divided by the sum of that same product (the product of the probability of h given only b and the probability of e given b and h) and the product of the probability of not-h given only b and the probability of e given b and not-h.

Which, as I graphically show on page 50, amounts to saying, in a more colloquial way, that the probability our explanation is true equals “how typical our explanation is” times “how expected the evidence is if our explanation is true,” all divided by that same product repeated (“how typical our explanation is” times “how expected the evidence is if our explanation is true”) plus “how atypical our explanation is” times “how expected the evidence is if our explanation isn’t true.”

The second is the extended Bayesian equation for more than two competing hypotheses: the probability of a hypothesis, h-1, given the evidence, e, and all your background knowledge, b, equals the product of the probability of h-1 given only b and the probability of e given b and h-1, all divided by the sum of that same product (the product of the probability of h-1 given only b and the probability of e given b and h-1) and the product of the probability of h-2 given only b and the probability of e given b and h-2, and the product of the probability of h-3 given only b and the probability of e given b and h-3, and so on, for as many hypotheses as you want to compare.

The prior probabilities must all sum to 1. For example, if you are dividing all possible hypotheses into three, then if the probability of h-1 given only b is 0.2, and the probability of h-2 given only b is 0.3, then the probability of h-3 given only b has to be 0.5, so that 0.2 plus 0.3 plus 0.5 will equal 1. The consequent probabilities do not, however, have to sum to anything. The probability of the evidence, e, given the hypothesis, h, and all your background knowledge, b, can be anything, without regard for what the probability of the evidence is given b and not-h.

The third equation to know is the odds form of Bayes’ Theorem: the ratio between the probability of a hypothesis, h, given the evidence, e, and all your background knowledge b, and the probability of not-h, given the evidence, e, and all your background knowledge b, equals the product of two other ratios, the ratio of the prior probabilities and the ratio of the consequent probabilities, which means: the ratio of the probability of h given only b and the probability of not-h given only b, and the ratio of the probability of e given b and h and the probability of e given b and not-h.

For example, if the ratio of priors is 2 to 1 and the ratio of consequent’s is 3 to 1, then the final ratio is 6 to 1, since 2 times 3 is 6. So in that event, the odds would be 6 to 1 in favor of the hypothesis. Note that this way, you don’t ever have to ask what any of the probabilities actually are, just what their ratios would be.

If there is anything else I can do to assist the visually impaired in understanding Proving History or Bayes’ Theorem generally, please let me know in the comments and I’ll see what I can do.

§

To comment use Add Comment field at bottom or click a Reply box next to a comment. See Comments & Moderation Policy.

Discover more from Richard Carrier Blogs

Subscribe now to keep reading and get access to the full archive.

Continue reading