I am a firm proponent of the methods of science having been trained in them and employ them in my work and life, but I have no faith. The collections of data that form the theories posited by science are not accepted by what they will predict they are accepted by what they have predicted...this is a subtle difference that makes assertions of all scientists having "faith" as some define it a fallacious one. I have no faith that any of the systems which have succeeded at answering questions about the workings of the world will continue to do so, I have only past evidence that they have. This is the difficulty of scientists...of falling into the "I believe this will happen." trap instead of asserting the "In the past, data of this sort was explained by this process." in the first, "faith" and "belief" are implicit...in the latter they are superfluous. One is a correlation between events past and present and the other is contingent on a "belief" casting into the future.
I remember first reading David Hume's causality argument and at first being confused by his seemingly devastating argument...but then I realized that his argument assumed that it was important to have causality at all...pure science doesn't require causality. If the laws of physics changed randomly every few moments ...would that negate our effort to try to understand the new configurations of the world from moment to moment? No. If we could in the short intervals infer a way to navigate the new space that works for a short time it would be worth it ..as we would have no other choice. In a limiting process, as these intervals of fluctuation accelerate we go from being able to say something about the world to asserting nothing but the moment..because the next moment could be explained differently. "Belief" and "faith" is an artifact of having finite durations of relative stability in the world we find ourselves coupled with the complacency and convenience of causal correlations and predictions about the future based on them...however the predictions ("faith","belief") are not necessary, as science is about the moment compared to the past. As the moment changes to invalidate the explanations of the past, then the description of the moment becomes the past to explain the new moments...no faith, no belief.
If I were to plot a course in a rocket ship to the moon and being on my way, realize that my ship is guided off course despite the calculations being correct. I must assume that something has changed, prior to the launch I didn't have "faith" that the ship would get to the moon..I had a set of correlations and rules under which the data of those correlations could be explained in the past. If the same rules (theory) could not explain the new situation in which I find myself then the conclusion is that the rules changed. So really if there is any "faith" in science...it is not "faith" in the corpus of rules that have explained the world, our theories being right...it is faith that they don't go wrong when we can't afford them to! Or more technically that we don't sample them in regimes in which their ability to infer solutions have sparse supporting empirical data. But this is not faith at all as it doesn't exist prior to the experience of "oh look the laws of physics just changed" or "oh I am off course on way to the moon"...no one actively thinks "gee I sure hope the laws of physics don't change on my way to the moon" which would be a faith statement.
In this way science models the iterative learning process of a neural network, acquiring patterns and changing them as data comes in, not judging, not predicting, not extrapolating, not believing why the data changes...just assimilating it in comparison to older data if it matches and changing older data if it doesn't. It is in comparison of *different* sets of these acquired pattern systems that the fallacy of faith can wiggle into the picture and I've always found that ironic.
In a comment response to that thread the statement was made that my position, which seemed to down play prediction in the use of scientific theory was a rare one for some on trained in a hard science (Electrical Engineering). I must admit it seems that most engineers see these philosophical micro analysis as a pedantic waste of time, if you think of it I am being a consummate engineer...rather than placing faith in any predictive capability of any theory that I used to design things in the world, I simply see the theory as a tool itself readily discarded or modified when it fails to help me "build" things (which is precisely how we are supposed to use them as prescribed by science itself!). The predictions are backward looking (on the past data) as opposed to forward looking (with some hope they'll work) so the proper position of constant doubt is never changed, and where can faith prosper where doubt is a constant refrain?? ;) Another delicious irony. ;)