Honduras (catch all)

[J Wolfsberger]

Nice job putting that together.

Questions:
1. Where do you find the ACH softwear?
2. Is it available w/o charge ie free?
3.Where does one find a good and detailed description of what it does?

So much for the practical questions. Now for something methodologically substantive - How did you arrive at your coding for credibility and relevance? While I generally agreed with you there were a couple (1 - 3) relevances that I questioned and a few more (3 - 5) credibilities that I questioned. In a similar vein, was Low not an option? And if it was, why did you not choose it in any case (I really thought there were a couple of low creds and 1 low relevance)?

Again, ya done good!!!!!

Cheers

JohnT

I'll second that.

John, I'm intrigued by the tool as well. Here's a start: Analysis of Competing Hypotheses.

Here's one tool: ACH2.0.3

[marct]

Nice job, AT!

JW, thanks for the links - I'm going to try playing with the software and see how well it operates.

[AnalyticType]

You beat me to it!

That's precisely the software I used. I have another one called DecisionWarning which has more graphing tools once you've completed the matrix, but ACH from PARC is better explained and easier to use.

JTF, to answer your third question, the ACH software from PARC (Richards Heuer) has a very detailed tutorial which can be utilized on-screen or downloaded and printed out. As I recall, it prints out to about 50 pages, give or take.

Regarding the subjective elements (credibility and relevance), the options are Low, Medium and High. For credibility of the information, if I had a piece of information that was obtained from a highly credible source, ie the State Department website, OR if I confirmed more than two disparate but credible sources, I rated it High. I used Medium for data which I found in two locations, where one or both sources were less than sterling. And Low I used on a couple items, but then found confirming information elsewhere and changed them to Medium.

Regarding relevance, the same options are available. I was evaluating two different sets of competing hypotheses which were related to each other but not necessarily 'linked' to each other. So I used High relevance for information which strongly was applicable to both sets of hypotheses, and Medium relevance for information which may be peripheral for one set but cogent for the other.

And yes, it is subjective. While I know a fair amount about Latin American history, culture and politics, I have not been there and am not an area expert. Someone who spent a great deal of time there (via State, DoD or CIA, etc) likely would rate relevance or credibility differently than I, in some instances.

I'll second that.

John, I'm intrigued by the tool as well. Here's a start: Analysis of Competing Hypotheses.

Here's one tool: ACH2.0.3

[J Wolfsberger]

You beat me to it!

That's precisely the software I used. I have another one called DecisionWarning which has more graphing tools once you've completed the matrix, but ACH from PARC is better explained and easier to use.

JTF, to answer your third question, the ACH software from PARC (Richards Heuer) has a very detailed tutorial which can be utilized on-screen or downloaded and printed out. As I recall, it prints out to about 50 pages, give or take.

Regarding the subjective elements (credibility and relevance), the options are Low, Medium and High. For credibility of the information, if I had a piece of information that was obtained from a highly credible source, ie the State Department website, OR if I confirmed more than two disparate but credible sources, I rated it High. I used Medium for data which I found in two locations, where one or both sources were less than sterling. And Low I used on a couple items, but then found confirming information elsewhere and changed them to Medium.

Regarding relevance, the same options are available. I was evaluating two different sets of competing hypotheses which were related to each other but not necessarily 'linked' to each other. So I used High relevance for information which strongly was applicable to both sets of hypotheses, and Medium relevance for information which may be peripheral for one set but cogent for the other.

And yes, it is subjective. While I know a fair amount about Latin American history, culture and politics, I have not been there and am not an area expert. Someone who spent a great deal of time there (via State, DoD or CIA, etc) likely would rate relevance or credibility differently than I, in some instances.

This methodology is similar to the Delphi Method. One way to approach the rating is to have a small panel of independent reviewers, then "average" their evaluations. Depending on how you want to look at it, you're either averaging the subjectivity, or averaging it out. I've used this approach in areas such as formal risk assessments for R&D programs with pretty good results.

AT, I owe you a beer if we ever run into each other. I've been looking for a package to do this for some time now.

[AnalyticType]

but I'll accept a nice glass of white merlot.

This software beats the hell out of manually constructing a matrix!

One of the things that my professor neglected to teach was the necessity of working across the matrix evaluating a piece of evidence against all hypotheses, rather than working down through all evidence under a hypothesis. When I was first learning this method, I had all kinds of frustration going on. Between the ACH tutorial and chapter Eleven in the book The Thinker's Toolkit I found the process much easier and effective.

The thing that I like about it is that one matrix can be used to evaluate many hypotheses simultaneously. The software runs the math functions, making it easier to eliminate the highly unlikely hypotheses. Then you can identify intelligence gaps, consolidate or split hypotheses (add new ones too) and improve your evidence list, then reevaluate. I didn't take the time to refine my matrix in this manner yesterday, but if this were for a product I would have done so.

...AT, I owe you a beer if we ever run into each other. I've been looking for a package to do this for some time now.

[John T. Fishel]

AT and JW.

AT the so-called experts could use a look at your conservative data analysis. Your critera for assigning values makes sense. Now, why don't you put together a short article for the Journal part of SWJ.

Cheers

JohnT

[Steve Blair]

Here's an interesting little BBC snippet: link

Granted it's only a handful of people, but it does give something of the local perspective on things.

[slapout9]

One of the things that my professor neglected to teach was the necessity of working across the matrix evaluating a piece of evidence against all hypotheses, rather than working down through all evidence under a hypothesis. When I was first learning this method, I had all kinds of frustration going on. Between the ACH tutorial and chapter Eleven in the book The Thinker's Toolkit I found the process much easier and effective.

That would be my biggest beef. From my LE view once you establish motive (who benefits from an action and how great is the benefit) you don't need a lot of other data. You need to establish his/her alibi(s)for doing or not doing something. My 2cents anyway.

[marct]

Hey Slap,

That would be my biggest beef. From my LE view once you establish motive (who benefits from an action and how great is the benefit) you don't need a lot of other data. You need to establish his/her alibi(s)for doing or not doing something. My 2cents anyway.

Generally, I'd agree, but motive can be tricky. Did Zelaya gain? Yup, but how about Uncle Hugo? Would he gain? Probably, so we've got a whole slew of different actors running around with overlapping motives. Same on t'other side as well - congress, the SC and the armed forces all stood to gain as well (as did the large landowners, businesses, etc.), so there's a whole slew of other, overlapping, motives.

Just my 1.724 cents

[jmm99]

Dick Heuer's book is online (chapter 8 = ACH), in which he notes (para 2):

Analysis of competing hypotheses (ACH) requires an analyst to explicitly identify all the reasonable alternatives and have them compete against each other for the analyst's favor, rather than evaluating their plausibility one at a time.

as the core concept.