From RationalWiki
Jump to: navigation, search

Thanks for the invite, ames, i like the idea....ill try to work on it sunday.--PalMD-Ars longa, vita brevis 22:45, 18 August 2007 (CDT)

"It's not what you say, it's what you can prove." --Kels 23:27, 18 August 2007 (CDT)
I talk a lot about evidence in my essay I am working on at Essay:Bayesian Inference and the Power of Skepticism if anyone wants to parse some of that down. Or ask me to explain it better. tmtoulouse plague 23:31, 18 August 2007 (CDT)
Your essay kicks ass. I will draw liberally if needed :-)-αmεσ (!) 23:33, 18 August 2007 (CDT)

Summary of thoughts on evidence[edit]

Evidence in science is data that allows us to adjust relative probabilities between competing hypotheses. In science we formulate a given hypothesis and always at least one other, the null hypothesis. Sometimes we can have hundreds of competing hypotheses and our goal is to find some sort of evidence that allows us to cull and separate out between them. Data is anything collected but data is not evidence.

For data to become evidence it must have several properties:

  • It must be meaningful, if the data can not be understood then it can not reveal anything, essentially there must be information with in the data.
  • The information with in the data must be predicted by at least one of the hypotheses being tested and not predicted by at least one of the hypotheses being tested. If none of them predict it or all of them predict it then it is not evidence because we can not cull or adjust probabilities between hypotheses.
  • The possible causes for information in the data should only be those causes we are testing for. In other words it must have been collected under controlled conditions. If proper controls are not met then determining the cause of the data is impossible to meet and it can not be used.

I think its also important to realize that evidence never arrives in a vacuum. All evidence must be compared against other evidence that has been brought out in the past. One experiment showing evidence for a hypothesis that thousands of other experiments have failed to find evidence for should be taken lightly. Hence extraordinary claims require extraordinary evidence and all that.

Think thats a basic summary of my thoughts. tmtoulouse plague 15:18, 19 August 2007 (CDT)

Bayesian reasoning[edit]

Could someone please explain why the part on the Bayesian notion of evidence was removed? An overview of the different types of evidence does not seem complete without mentioning bayesian evidence. If something was wrong with the piece please tell me what it is. --Sophronius (talk) 20:08, 19 March 2013 (UTC)

Have to run for now, I can explain more later, but some problems were conflating Bayesianism with empiricism and contrasting Bayesian evidence to scientific or legal evidence. That is to say, there is no specific category of "Bayesian evidence," it's a statistical process used to evaluate evidence in general. (Whether it's always useful is another question, of course.) Nebuchadnezzar (talk) 17:41, 18 March 2013 (UTC)
Thanks for the clarification. I got most of what I wrote from Less Wrong, and it's possible they use terms differently than people use them here. See the definition of (bayesian) evidence here for example. Wikipedia also talks a bit about Bayesian evidence for example here, which contradicts your point. The page on Bayesian probability also mentions the subjectivist interpretation of Bayesian probabilities, where a probability is seen as a degree of personal belief, from which follows a different concept of evidence than legal or scientific evidence. I also don't see the conflation with empiricism: With both empiricism and Bayesianism you are supposed to look at the world to form beliefs, but that doesn't mean they are the same. --Sophronius (talk) 20:08, 19 March 2013 (UTC)
It's a robust enough point to make, but as Neb points out, Bayesian is a method rather than a category of evidence. It's worth including, but needs to be worked in right. Scarlet A.pngpathetic 12:18, 19 March 2013 (UTC)
Fair enough. You are right that Bayesianism only provides a method to update prior probabilities based on new information, though I would still find it pertinent to point out that when using this method any and all relevant information is considered: You don't first remove all of the "insignificant" or "unreliable" information before doing the final probability calculation like you do with evidence in a court or with a scientific paper. So the difference in the way evidence is handled seems to me a useful distinction to make. The reason I feel this is important is partially that on the James Randi forums people keep saying things like "well that information doesn't count as evidence because it's not strong/reliable enough, doesn't matter how much of it you have". Having a page on evidence to refer them to that explains that all evidence is considered under Bayesianism would be nice. --Sophronius (talk) 20:08, 19 March 2013 (UTC)

The juvenile abstraction of guessing probabilities and then reguessing them is not Bayesian anything - it's pseudointellectual anti-math. Which, by the way, is what you get when you listen to uneducated pseudointellectuals more worried about Terminator 3 and cryogenics than climate change. Hipocrite (talk) 12:42, 19 March 2013 (UTC)

Wow, this is on the same level as "You atheists are stupid because I don't like Richard Dawkins", or "Global warming is false because Al Gore is an ass". Nobody here is talking about Yudkowski (I assume you mean him). Anyway, updating probabilities in accordance with Bayes' theorem is as far from guessing as you can get (where you get the priors from is a different matter), so you plainly don't know what you are talking about. --Sophronius (talk) 20:08, 19 March 2013 (UTC)
"Thanks for the clarification. I got most of what I wrote from Less Wrong" -You. Hipocrite (talk) 01:58, 20 March 2013 (UTC)
Getting a definition from Less wrong =/= caring about Yudkowsky's life story --Sophronius (talk) 18:54, 20 March 2013 (UTC)

BoN Edit to Evidence for the historical existence of Jesus Christ‎‎ [edit]

A BoN has recently added the following to the above article: It has been removed from there by two of us (as unneeded) but I think that some of it might be usable here:

In addition to this you have as related by two Scandinavian textbooks, written by historians, source criticism:

1) Human sources may be relics (e.g. a fingerprint) or narratives (e.g. a statement or a letter). Relics are more credible sources than narratives.

2) A given source may be forged or corrupted; strong indications of the originality of the source increases its reliability.

3) The closer a source is to the event which it purports to describe, the more one can trust it to give an accurate description of what really happened

4) A primary source is more reliable than a secondary source, which in turn is more reliable than a tertiary source and so on.

5) If a number of independent sources contain the same message, the credibility of the message is strongly increased.

6) The tendency of a source is its motivation for providing some kind of bias. Tendencies should be minimized or supplemented with opposite motivations.

7) If it can be demonstrated that the witness (or source) has no direct interest in creating bias, the credibility of the message is increased.<ref>Olden-Jørgensen, Sebastian (1998). Til Kilderne: Introduktion til Historisk Kildekritik. København: Gads Forlag.</ref><ref>hurén, Torsten. (1997). Källkritik. Stockholm: Almqvist & Wiksell.</ref>

Scream!! (talk) 11:08, 24 November 2013 (UTC)

I'm not keen. I guess some of it might OK be if rewritten as text rather than bullet points, and it shouldn't be cited to Swedish textbooks as that just looks silly. These issues are familiar to anyone working in history or related disciplines and so appear in lots of introductory textbooks. Also, some of these points are oversimplified to the point where they're misleading. E.g. see points 3 & 4: there are lots of reasons why a first-hand source may give an accurate account, while a secondary source (depending on the level of research & primary sources available) may give a more accurate one. WěǎšěǐǒǐďWeaselly.jpgMethinks it is a Weasel 13:37, 24 November 2013 (UTC)