Talk:Analysis of competing hypotheses

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Military history (Rated Start-Class)
MILHIST This article is within the scope of the Military history WikiProject. If you would like to participate, please visit the project page, where you can join the project and see a list of open tasks. To use this banner, please see the full instructions.
Start This article has been rated as Start-Class on the quality assessment scale.

Reference 1 - CIA pdf article Chapter 8

Say something about:- Richard Heuer; give the eight step approach; find a definition of 'prescient intelligence analysis' and compare; talk about the Bayesian aspect of the reasoning; talk about where it could be useful in the public domain for eg analysis of 9/11, London bombing, Bologna train bombing; talk about the possibility of a wiki that can collate this process where everyone can contribute;

Merge/disambiguate[edit]

This topic is covered extensively in intelligence analysis. Howard C. Berkowitz 02:26, 31 October 2007 (UTC)

Comment moved from article[edit]

Needed topic: Analysis of competing hypotheses - counterdeception (ACH-CD) {{subst:unsiged|65.122.202.206}}

Evolving the article[edit]

There is a Wikipedia principle, with which I've had my own problems, that articles should not read as if they were instruction manuals. Right now, the eight-step process reads much like one, or a paraphrase of Heuer's work.

For this to be a real contribution, the eight steps could stay, but they need to transform from instructions to compare-and-contrast with other methods. For example, a given point could be supplemented with a description of a related statistical technique; there are quite a few relevant hypothesis formation and results evaluation in clinical medicine.

Howard C. Berkowitz (talk) 20:56, 30 April 2008 (UTC)


I am not familiar with with the ACH matrix or SACH matrix beyond the intelligence community, so I won't speak to those points, but I will expand SACH.

McCollum (talk) 30 April 2008 —Preceding comment was added at 21:57, 30 April 2008 (UTC)

Do look at what is already in Intelligence analysis#The Analytic Process, and some of the material preceding it. How much overlap exists between that article and the ACH one? Howard C. Berkowitz (talk) 22:13, 30 April 2008 (UTC)
Well, I think it is important to expand on the steps (which was done in this article I hope) and elaborate on SACH. What suggestions do you have?

McCollum (talk) —Preceding comment was added at 22:24, 30 April 2008 (UTC)

I'm not sure what you mean by expand. Have you looked at the WP:NOTGUIDE policy? If expanding the steps gives more of a guide on how to execute them, that's more what Wikiversity is supposed to do.
This is a delicate balance. For example, step 1 now speaks of using words of estimative probability. That should be sourced if so, since statistical hypothesis testing usually tries to formulate hypotheses that are either true or false. To do so, it may be necessary to state the outcome, if numeric, as a relation: "the FOO-7 grenade has at least a 97.2 percent chance of detonating." It would be interesting to see if there is sourced material that discusses how useful WEP are in hypotheses.
Is there an underlying structure to the causality suggested in step 2? Is this something that might be a Petri net? Can one validate the model by backtracking with root cause analysis, not simply as a mathematical exercise, but as a classic intelligence indicator (e.g., radio transmissions in the 141 MHz range from grid square 283019 indicates that a fissionable material shipment will originate from grid squar 320384 in the next six days?)
Heuer's comment, in step 4, about keeping rejected information is, to some extent, a technique used in expert systems, or, more specifically, validating the rule-chains used to get to an inference.
Howard C. Berkowitz (talk) 00:53, 1 May 2008 (UTC)
I am little confused that you think the ach is statistical and mathematics based. A simple ach matrix just calculates the number of inconstancies and therefore that hypotheses can be eliminated. The SACH is useful because then it could be mathematical, for example weight your inconstancy score based on source reliability and evidence relevance. WEPs are useful because they can provide for a more nuanced and accurate hypotheses. For example, it is "likely" the U.S. will invade Iran or it is "unlikely" the U.S. will invade Iran. You have two hypotheses and you see which is more inconsistent. If you find that it is likely that the U.S. will invade Iran, you can split the hypothesis as the SACH suggests to find out a timeframe. —Preceding unsigned comment added by Mccollumegan (talkcontribs) 01:23, 1 May 2008 (UTC)
I went back to the article and discovered I was reading the wrong citation for SACH proper. After I saw my error, I didn't immediately remember the more formal methods I had in mind on improving ACH, which is definitely mathematical, not always statistical but sometimes using methods from automata or graph theory, but found some.
I cited, for example, the paper by Valtorta, Marco et al., "Extending Heuer’s Analysis of Competing Hypotheses Method to Support Complex Decision Analysis", presented at the International Conference on Intelligence Analysis Methods and Tools. It uses a probabilistic Bayesian model. Pope and Josang use a different logical analysis technique for dealing with uncertainty.
Anyway, there is a substantial literature besides SACH for extending Heuer's work using more formalism and more automation. So, yes, the trends in work on improving ACH are, I believe, mathematical, in a broad sense -- perhaps I should add cognitive, but observing that the cognitive processes involve mathematical structures to drive the visualization. Howard C. Berkowitz (talk) 03:08, 1 May 2008 (UTC)

SSS Research & DECIDE[edit]

It would appear that SSS Research is no longer in business: thincSoft Acquires Intellectual Property of SSS Research

The company doesn't appear to have done much in the way of advancing ACH. I would therefore question the relevancy of SSS Research to this article on ACH, other than perhaps as an interesting footnote. —Preceding unsigned comment added by Mike.wilmoth (talkcontribs) 01:49, 3 June 2010 (UTC)