In 2003 Steven P Unwin (The Probability of God) tried to use Bayesian probabilities to calculate the probability of the existence of god.
Unwin starts with a value of 50% representing maximum ignorance, or to put it another way the value BEFORE examining ANY evidence.
This starting point seems to confuse many people, they tend to say "but the probability isn't 50%" or "it's a fallacy to assume equal probability", not realising that the starting point is not the actual probability, its merely a neutral starting point on which to project the evidence and observe its effects.
There is a video which I've seen linked to on twitter as an example of the foolishness of using 50% as a starting value, but ironically (or not, depending on the culturally specific value you place on that word) it actually illustrates the way Unwin's method works - it features estimates of a riot happening in a small American town, the first character doesn't stop to consider any evidence and just says that there are two possibilities, either it happens or it doesn't, so it's 50%, and the second character then starts to reference evidence to show that the probability is not 50%. An so it is with Unwin's method, you start with maximum uncertainty, then you examine the evidence a bit at a time, and each piece moves you away from the starting value towards the final/actual value.
The formula looks like this:
Pafter = Pbefore x D
Pbefore x D + 100% - Pbefore
The Pafter for each piece of evidence becoming the Pbefore for the next piece.
D is the area where problems start to arise, not with the math, but with (as Unwin himself notes) the subjective nature of D.
D is 10 for evidence which is much more likely to be produced if god exists, 2 for slightly more likely, 1 for even, 0.5 for slightly less likely, and 0.1 for much less likely.
So, for instance, Unwin gives a value of 10 the "recognition of goodness", but that is an example of subjective bias since it can be explained adequately from an evolutionary perspective as well as from most other religions as well, so that value should really be a 1.
So the problem is that if your evaluation of the evidence is subject to confirmation bias, as it will most likely be for anyone who already has an emotional commitment to one side of an issue, all you will do is put numbers onto your bias.
It may however yet be useful for those who have not made up their minds and for those rare individuals who can avoid any bias in their evaluations (remembering the quote "the first principle is that you must not fool yourself - and you are the easiest person to fool"), and of course bearing in mind that if you use a an overview of the evidence on one side and go into massive detail for the evidence for the other side that will skew the result as well.
Bearing all this in mind then, let us see if we can produce a probability of "mission creep" in assisted suicide/euthanasia legislation.
D = 10 for much more likely that mission creep will happen, etc.
To save time and space I'll be using a broad overview of the evidence.
Pbefore 50%
1. Some degree of mission creep has happened in every one of the states that have already introduced such laws.
If the probability were very low (0.1D, lets call it 10%) the chances of all 4 would be 0.0001
If slightly low (0.5D, call it 40%) the chances of all 4 would be 0.0256
If slightly high (2D, call it 60%) the chances of all 4 would be 0.1296
If very high (10D, call it 90%) the chances of all 4 would be 0.6561
The very high figure here is so much higher than the others that it is the only reasonable value to use.
So, Pafter is 90.90%
2, Campaigners exist on both sides, it is not practical to estimate their effectiveness before the results of the campaigns are known, so 1 is the reasonable value for D.
Pafter is 90.90%
3. mission creep and normalising of ideas are established and researched concepts in psychology.
since psychology is a "soft" science, lets use a value of D=2
Pafter is 95.23%
4. normalising of ideas, and "yes sets" are regularly and very profitably and effectively used in marketing, advertising, and sales. At the very minimum D=2.
Pafter is 97.56%
Based on this high level overview of the evidence then, the probability that any given assisted suicide legislation will in practice expand beyond it's initial parameters is 97.56%
