Essay talk:Being convinced

I've just concluded fuck all, haven't I? ^moral 09:44, 20 September 2010 (UTC)

I'm not sure. You certainly took quite a long route to get there though. :-) --Bob^{Spring is sprung!} 12:50, 20 September 2010 (UTC)

Heh, it was a little tl;dr. But I think it's because I was originally going to dump the first two and a half paragraphs on the Burden of Proof talk page (inspired by all the recent Less Wrong talk the fact that I don't accept a the Singularity and AI and cryonics bundle, but they could happily view me in the same way that I view an ID advocate for that). But then the actual answer started hitting me, it grew out of proportion and I expanded it into a general exploration of what Burden of Proof means. I'll let it rest for a week, re-reading and inevitably facepalm over how stupid it is (this is pretty much true of everything I write, I recently read a few forum posts of mine from ca.2005 and almost cried). ^moral 17:36, 20 September 2010 (UTC)

If I may...

I think there are a few things on here I disagree with:

• [T]here's no external and objective source declaring what is and what is not true Sure there is. It's called reality. How do we know controlled trials are better than anecdotes? Probability theory says so. Ultimately evidence is about probability, and controlled trials allow scientists to conduct tests that are (a) accurate and (b) have quantifiable results.
• [T]he best we can really hope for is to be consistent I disagree here too. It's not like we don't have access to reality - we do, through our senses. Just as we can determine if a proposition is true or false, we can decide if a method for weighing propositions is false.

Whoops, that sort of became a quote mine. But I think you see what I am getting at. There is a way to tell if one's beliefs are right, and it has little to do with the burden of proof. Tetronian ^{you're clueless} 17:57, 20 September 2010 (UTC)

I would disagree because you miss the point that these are methods of probing reality, and this entire thing doesn't take into account when people make stuff up out of whole cloth (because in those cases they're hardly likely to address burdens of proof, yet alone satisfy them). An anecdote or someone's personal experience obviously represents reality. RCTs also probe reality. We all live in reality, but the issue is how to consistently and most accurately interpret the evidence to reconstruct what reality is telling us. The ability to form controls are good rationalisations for why we pick certain methods over others, but don't represent outright proof that these are the One True Way while others aren't. ^moral 19:16, 20 September 2010 (UTC)

I think I get it now. What you are saying is this: there is no objective way of comparing standards of assessing evidence. Interestingly enough, I had a similar problem a few weeks ago, but it was more easily resolved than this one. This problem goes back to good old Thomas Kuhn's "incommensurability of standards." That is, different people with different standards of evidence are talking past each other and won't really be able to communicate. Many of Kuhn's followers believed that there is no objective way to compare standards. This leads you into Lakatos, Laudan, Feyerabend, and others. I'll tell you what: I'm going to do some reading on this (I have to read up on Kuhn anyway) and I'll get back to you later today. Tetronian ^{you're clueless} 20:03, 20 September 2010 (UTC)

"Standards of evidence" cropped up in a chat with Bob a while ago - in the bar, I think. And it's indeed important, because I mentioned in the screed, it's about being personally convinced. For some people, their standard for evidence is such that they're happy to say God exists (not something you can test against reality almost by definition!) because the Bible says so. Others a little harder to please. Once you realise people have these different standards, you can get around a lot of the cyclic nature of debates. ^moral 20:25, 20 September 2010 (UTC)

I re-read a few chapters of Structure and thought about this a little more. First, Kuhn's take on it: he seems to think that standards of evidence vary with paradigms, which is plausible. However, he thinks that what really makes a paradigm distinct from others is the problems it solves, not the standards of evidence it uses, since the latter evolves to help with the former. So, as paradigms change, so do standards of evidence. In general, a new paradigm is better at explaining phenomena than an old one, so the standards of assessing evidence must be at least as good as in previous paradigms. It should also be noted that standards of evidence change as technology does, since new methods become available. This only complicates the problem further.

A tangential question is, how does evidence suggest certain models of reality. Kuhn would say that this is definitely paradigm-dependent, and he is probably right. However, in this case we can make appeals to simplicity (Occam's Razor) and falsifiability (Popper), and even though we cannot "prove" these methods we can support them with inductive arguments.

What do I make of all this? I'm not sure. It certainly seems like there are ways of assessing standards of evidence, but only in extremely simple cases when we can use empiricism to do so. (Example: You think science textbooks are good evidence; someone else thinks the Bible is. But there are empirically verifiable cases when the textbooks are right and the Bible is wrong. However, this method only works in very, very simple cases.)

So, in closing, I'm still on the fence about the idea that raw empiricism (a la using hypotheses to make predictions) can be used to compare standards of evidence. However, I think what this really compares is model-making based on evidence, not standards of evidence themselves. But I need to do some more reading on this. Tetronian ^{you're clueless} 01:38, 21 September 2010 (UTC)

I think I had some kind of dream about this; I woke up thinking about it. Anyway, I did a little more reading and thinking, and I think there are two ways to compare standards of evidence.

The first method, which is admittedly very weak, is to compare standards of evidence and see when they are consistent. For example: even a Biblical literalism uses more than one standard of evidence; if he/she sees that the Bible says X, he/she would also to expect X to be verifiable using other means. (I should also be more specific about the definition of evidence: evidence is an event entangled with a proposition, so if the proposition is true we expect to see result A and if it is false we expect to see result B.) We can use this to determine where certain standards of evidence conflict, and use the heuristics like parsimony to get a rough estimate of which methods are (a) more accurate and (b) more precise. An example of (a) would be: if the Bible conflicts with many other methods of verification, it makes more sense to assume that one method (the Bible) is wrong versus all of the other methods being wrong, and this is supported even further if we examine the historical accuracy of all the methods. Admittedly this is very weak, but it's better than nothing.

The second thing we can do is take a page from Thomas Kuhn. Kuhn says that standards of evidence are an important part of paradigms (scientific worldviews). Paradigms are discredited when there are major problems they cannot solve (the most obvious example being an observation they cannot explain), and when this happens a new paradigm appears to solve these previously impossible problems. As a result, newer paradigms have more "problem-solving power" because they can make more sense of reality than previous ones. This sound's very abstract, so I'll give an example: at one point in the history of biology there was a debate between the Biometricians (who wanted to use math as the most important standard of evidence for genetics) and the Mendelians (who wanted to use causal explanation as the most important standard of evidence for genetics). Eventually, the Mendelians won because their standard of explanation could address and solve more problems than the Biometricians'. As a result, we can compare standards of evidence by looking at what problems they can address in addition to comparing them to other methods.

If this sounds at all unclear (it probably does since I just woke up), let me know and I'll elaborate. Tetronian ^{you're clueless} 14:12, 21 September 2010 (UTC)