Unequal dichotomies

[Hugo]

They try to be too perfect by splitting 16 types right down the middle, 8 going one side and 8 going the other. Wouldn't there be greater accuracy if dichotomies are not necessarily split equally? So there could be 9 going on one side and 7 going on the other. Or 10 going one way and 6 going the other.

[Gilly]

Why would that improve accuracy?

[Logos]

They try to be too perfect by splitting 16 types right down the middle, 8 going one side and 8 going the other. Wouldn't there be greater accuracy if dichotomies are not necessarily split equally? So there could be 9 going on one side and 7 going on the other. Or 10 going one way and 6 going the other.

The dichotomies are done on the basis of breaking the whole into various respective halves. Of course, you could always try to break the dichotomy based upon the specificity of the dichotomy and say how are positivist subjectivists different from the rest, but that still involves the dichotomy being broken down further and further into different parts and pieces, groups and sub-groups, and categories and sub-categories.

[KSpin]

I don't think you'll find 50% of the population are in one half and 50% in the other. That's not exactly how it works.

[reyn_til_runa]

They try to be too perfect by splitting 16 types right down the middle, 8 going one side and 8 going the other. Wouldn't there be greater accuracy if dichotomies are not necessarily split equally? So there could be 9 going on one side and 7 going on the other. Or 10 going one way and 6 going the other.

Probably more about being logically consistent than perfect. A dichotomy, by definition, cannot be divided the way you suggest unless it's the right kind of weather, and 7=9. In order for that to happen, the units comprising 7 would have to be larger than the units comprising 9, which is impossible in math, but is certainly possible for cookies or something.

I've been wanting to analyze socionics through the framework of set theory for quite some time. Perhaps researching this theory a bit will prove helpful to you.

[Joy]

I understand what you're saying. I would think it would be more accurate if it was possible to have some neutral dichotomies instead of assigning every type every dichotomy.

[Logos]

I understand what you're saying. I would think it would be more accurate if it was possible to have some neutral dichotomies instead of assigning every type every dichotomy.

Would that not make it a trichotomy?

[Joy]

However, the way the groups are divided, each ends up having 8. It's not like they'd be like all ExTx's are this dichotomy... except ESTj's.

[Joy]

I understand what you're saying. I would think it would be more accurate if it was possible to have some neutral dichotomies instead of assigning every type every dichotomy.

Would that not make it a trichotomy?

nah, it still has two ends of the spectrum

[reyn_til_runa]

I understand what you're saying. I would think it would be more accurate if it was possible to have some neutral dichotomies instead of assigning every type every dichotomy.

Would that not make it a trichotomy?

nah, it still has two ends of the spectrum

which spectrum?

[Joy]

I see each of the dichotomies as a spectrum.

[Gilly]

But 16 isn't evenly divisible by three. And we can't exactly just create new ones...unless you have any ideas...

[Joy]

The point he's making is that it's odd that all of the dichotomies are split up 50/50. Why couldn't there be more static types than dynamic types? And allowing a type to be neutral on a dichotomy wouldn't mean that there are 3... trichotomies (and even if there were, why should the numbers all be even?), it would just mean that there are types that don't clearly fall to one side of a particular dichotomy or the other, so there's no point in assigning them to one side of that dichotomy.

[Gilly]

But see, there are causes for these distinctions, and those causes are rooted in types, which are distinguished only by categories that have two variables. Where in the system is there a factor that could potentially divide the socion into an odd number of groups?

[Joy]

I'm not saying I agree with him in anything except that some dichotomies are more noticeable in some types than others.

[Gilly]

I'm not saying you're wrong.

[Exodus]

I'm not saying I agree with him in anything except that some dichotomies are more noticeable in some types than others.

i think it's also worth noting that if types are defined by dichotomies as much as dichotomies are defined by types, then if some dichotomies seem unequally distributed among the types from your POV, maybe you need to recalibrate your understanding of or at least the boundaries among what you call the types. the same works for calibrating dichotomies w.r.t. types. it's a beautifully recursive and complete error-checking mechanism, great stuff i think. although i'd never expect to achieve a perfect calibration of either since the underlying system (people?) is so complex.

You can see each type as being a kind of "maximum" - so that the ISTj-factor is most prevalent in ISTjs, and less so in all other types. The crazy part is that while you can categorize types according to Reinin dichotomies, you can do the exact opposite too. So the ISTj dichotomies are Process, Resolute, etc. The only difference is that there is an odd number of dichotomies, which means you can't split the dichotomy space evenly. Weird...

@ifmd, It's hard to find people who are interested in the math behind it. I've written about it at Wikisocion:

http://wikisocion.org/en/index.php?t...in_dichotomies

[Exodus]

!

The asymmetry does not exist. I was forgetting that traits are really more fundamental than dichotomies, per se. In the article I talk about the null/nonnull traits. Since every type has the nonnull trait, that brings the total up to 15 positive traits and 15 negative traits. Types and traits are interchangeable (in math terms, duals!). I haven't worked it out completely yet, but this brings up the possibility of anti-types (like ~INTj, in the same sense as ~J=P). What they describe is anyone's guess.

[Exodus]

isn't the distribution of "ISTj-ness" across the types very different from the distribution of a typical dichotomy? dichotomies may have peaks and troughs that occur multiple times in the socion. but ISTj-ness peaks at ISTj while being equal everywhere else. (a good thing for ENFp's.) at least if it is measured by counting the number of different dichotomies between any given type and the ISTj. (perhaps the validity of this measure is debatable.. but how else could you measure it?)

Yes, that's exactly what I meant. It does involve assigning values to the dichotomies.

you can find that any given type has 8 different reinin dichotomies, compared to every other type except itself. this would jive with what i've often heard said about quadra members: similarity does not make for their compatibility, as much as them being complementary to each other.

Complementarity can be explained within Model A, given the valued/subdued dichotomy for functions. However...

(that in mind, what would be so contradictory about all the types being identically dissimilar to each other?)

Reinin dichotomies do not mean that intertype relations are all the same, per se, but they do show that to give them any kind of ordering (such as by quality), you need to add an assumption to the (trait) theory. It doesn't work even if you use a restricted but symmetric set of dichotomies - such as all the dichotomies dependent on rationality. You have to assign values by hand - and it's more difficult than it seems to recreate the typical ordering (with duality being most compatible, conflict least, etc.). I don't know why this is; I'm really not that familiar with the math. I have a hunch that there is a way to deduce the values given a complete ordering. It would have to be rather complex. Though that could mean that such a goal is misguided; it's better to just say that all orderings are equally plausible. Though that doesn't excuse us from explaining the characteristics of the preferred one.

In more simple terms, you can't say duality is better than conflict unless you define exactly what you mean by "better" - and it's much harder to say so in a way that applies to all of the types, not just duals and conflictors. Plus, not all dichotomies are equal when it comes to rating relationships (though it's extremely tempting to think so).

think along the lines of the mistake some people make by assuming types with more similar foundation dichotomies are more similar overall. there might have been something to this if types weren't equally dissimilar from each other in the reinin sense as i argued they were earlier. but since they in fact are, the only type dichotomy which would give any information about a type is the dichotomy of that type itself. (just identity.)

*Type dichotomies* still give information about types, there is no doubt about that. The holistic view is possible, but uninteresting (take MBTI or Enneagram).

[anndelise]

*meekly steps in*
ummmm
just a quick note
on the wiki site, you use entp as an example, and for number 9 you have "NT (aristocratic/democratic)"
it might be easier for some people...like me.....if it was consistent as to which side the relevant dichotomy was on.
for example, for E, you have (extroversion/introverstion)...the relevant dichotomy is on the left side.
same with a number of others (i don't know which go to which so i can't easily verify most of them.)
but I do not that NT is supposed to be democratic...which is listed here on the right side

Could you perhaps edit that article and put the relevant dichotomies on the left side of the split?
ie "9. NT (democratic/aristocratic)"

[Exodus]

these "values" would be magnitudes by which we'd weigh each dichotomy's significance w.r.t a given outcome, correct?

Hmm, "outcome" is a bit too qualitative, but that's about right. I've been thinking in the context of relationships, but it could be applied to properties of types. For example, if S=1 and F=1, then all SF types are rated 2, ST and NF 1, and NT 0. That's a partial ordering, which you could call Socialness or something.

if the given outcome is "compatibility" with other given types, then i agree some sort of variable magnitude scheme might be useful in creating a mathematic model for the functional "complementariness" Model A asserts more qualitatively. this is obvious because without some variable magnitudes here, wouldn't there be no difference between the compatibility with your dual and that with your conflictor? because of this i was never tempted to assume all dichotomies are equal in "rating relationships".

I've been a bit biased by Smilingeyes here. His descriptions of how Positive/Process/Narrator work in duality are excellent, which seems to imply that all of the dichotomies are relevant. Perhaps they still are, but of course not to the same extent. (I'll get back to you on this.)

the given outcome i was referring to is if a generalized form of some type was just compared side by side with another type: how different would their information metabolism be, relative to the differences between other types? (there's an assumption here that this corresponds to reinin dichotomies i addressed earlier.) i am tempted to assume all dichotomies are equal in this case (and so all the associated magnitudes equal and irrelevant) because it seems most parsimonious.

I'm not sure what you mean. To the extent that there is a difference between types, a dichotomy expresses it.

if we were to empirically estimate every human being's reinin traits and adjust the system so that internal consistency is maximized, we might encounter tradeoffs in parsimoniousness relevant to these magnitudes, such as: making all dichotomies weighed equally. making all dichotomies distributed equally. making all types distributed equally, etc. however there are an infinite many variations to these tradeoffs that are possible. and an infinite many plausible explanations such as: culture, economics, the stuff of integral types.

The problem is not with the Reinin model; it is internally consistent. Any static model of the human psyche can be reformulated according to it.

However, the model also predicts the existence of things that are not intuitively obvious. But as long as they are measurable, they must be taken at face value. It's similar to the prediction of the positron.

if socionics is to be a generalized theory of information metabolism, it makes sense to me to avoiding tainting the most central mathematical models with such local adjustments. there are already local adjustments implicit in the qualitative aspects of the theory such as the type and function descriptions written for given audience, in case you're worried about relativism and practicality (e.g. what if society A is so saturated with traits X, Y, Z that the ability of our model to make many useful differentiations and predictions within A disappears?)

Ideally the qualitative aspects come directly from the model. I've been focusing on the syntax, but the semantics is where the solution really lies. Semantics is a difficult question in general, but unfortunately you can't get rid of it. Smilingeyes tried to capture the semantic aspects formally - but I'm not sure if he was entirely successful. Socionics definitions (except in qualitative, holistic form) are so conceptually remote from experience that their meaning becomes hazy, especially when you have to derive new, equally meaningful concepts.

practical application aside, would you agree with the mathematical argument that if the associated magnitudes are equal (or omitted) then all types appear equally dissimilar in terms of reinin dichotomies? (checking my math heh)

Yes.

[Exodus]

Oops, typo.

[FDG]

isn't the distribution of "ISTj-ness" across the types very different from the distribution of a typical dichotomy? dichotomies may have peaks and troughs that occur multiple times in the socion. but ISTj-ness peaks at ISTj while being equal everywhere else. (a good thing for ENFp's.) at least if it is measured by counting the number of different dichotomies between any given type and the ISTj. (perhaps the validity of this measure is debatable.. but how else could you measure it?)

No it's not like that, some dichotomies are mutually exclusive and not orthogonal - for example infantile and aggressor and careful and victim - there are varying degrees among the types and sub-types of those dichotomies so if you have an ISTj-Se he'll peak aggressor and an ENFp-Ne he'll peak infantile, and conversely the value of infantile in an ISTJ-Se will be zero and max in an ENFp-Ne (i use max because we can model it the way we prefer, we could say that max=a and then measure the values <a by dividing)

[FDG]

these "values" would be magnitudes by which we'd weigh each dichotomy's significance w.r.t a given outcome, correct?

Hmm, "outcome" is a bit too qualitative, but that's about right. I've been thinking in the context of relationships, but it could be applied to properties of types. For example, if S=1 and F=1, then all SF types are rated 2, ST and NF 1, and NT 0. That's a partial ordering, which you could call Socialness or something.

I still think that types cannot be considered so statically and so basically almost for each person you'd have a different quantitative degree of ST SF NF depeding on their position on the IJ EJ EP IP sin curves

Socionics definitions (except in qualitative, holistic form) are so conceptually remote from experience that their meaning becomes hazy, especially when you have to derive new, equally meaningful concepts.

When rl observations start to accumulate you easily connect the experience with the theory, but otherwise it's basically like trying to solve an integral using the fundamental theorem of calculus

[Exodus]

these "values" would be magnitudes by which we'd weigh each dichotomy's significance w.r.t a given outcome, correct?

Hmm, "outcome" is a bit too qualitative, but that's about right. I've been thinking in the context of relationships, but it could be applied to properties of types. For example, if S=1 and F=1, then all SF types are rated 2, ST and NF 1, and NT 0. That's a partial ordering, which you could call Socialness or something.

I still think that types cannot be considered so statically and so basically almost for each person you'd have a different quantitative degree of ST SF NF depeding on their position on the IJ EJ EP IP sin curves

That's one part of Smilexian socionics that I still don't agree with.

Socionics definitions (except in qualitative, holistic form) are so conceptually remote from experience that their meaning becomes hazy, especially when you have to derive new, equally meaningful concepts.

When rl observations start to accumulate you easily connect the experience with the theory, but otherwise it's basically like trying to solve an integral using the fundamental theorem of calculus

Socionics wins in terms of the number of lightbulb moments, but it's hardly scientific. Too much Ne, not enough Ti.

[FDG]

these "values" would be magnitudes by which we'd weigh each dichotomy's significance w.r.t a given outcome, correct?

Hmm, "outcome" is a bit too qualitative, but that's about right. I've been thinking in the context of relationships, but it could be applied to properties of types. For example, if S=1 and F=1, then all SF types are rated 2, ST and NF 1, and NT 0. That's a partial ordering, which you could call Socialness or something.

I still think that types cannot be considered so statically and so basically almost for each person you'd have a different quantitative degree of ST SF NF depeding on their position on the IJ EJ EP IP sin curves

That's one part of Smilexian socionics that I still don't agree with.

But how can't you? It's not something you can or cannot agree with, it just is like that in real life :S

Socionics definitions (except in qualitative, holistic form) are so conceptually remote from experience that their meaning becomes hazy, especially when you have to derive new, equally meaningful concepts.

When rl observations start to accumulate you easily connect the experience with the theory, but otherwise it's basically like trying to solve an integral using the fundamental theorem of calculus

Socionics wins in terms of the number of lightbulb moments, but it's hardly scientific. Too much Ne, not enough Ti.[/quote]

Yeah sure nothing scientific, but i honestly hope it'll never become scientific. Right now I still feel kind of bad for being able to predict relations with ease.

[Exodus]

But how can't you? It's not something you can or cannot agree with, it just is like that in real life :S

I can understand subtypes, but the continuous stuff is needlessly complicated IMO.

Right now I still feel kind of bad for being able to predict relations with ease.

I know what you mean - I know the exact right way to piss people off, if I wanted to. I think it feels bad partly because socionics is known to so few people. If everybody knew about it, it wouldn't be a secret weapon kind of thing. But it would probably still be a headtrip...

[FDG]

hotel: my reply is in the works.

No it's not like that, some dichotomies are mutually exclusive and not orthogonal

i think this is only so if three or more dichotomies are invoked in the comparison. for example, even though static/dynamic and taciturn/declarer are constructed both using introversion/extroversion and judging/percieving data, there are 4 static taciturns and 4 dynamic declarers. (isn't that the point of the construction?) however whenever more than two are invoked then you get correlations like the bundled peaks and troughs observed on the Smilexian charts.

either way i don't see how this relates to what you quoted. if i tell you a type has exactly 7 dichotomies in common with ISTj it gives you no information about which type it is, other than being not ISTj. count them for yourself: http://the16types.info/forums/viewtopic.php?t=9134 (i have some spreadsheets to post someday myself.)

Yes I understand what you meant now, you're right I agree

[Hitta]

I don't think its possible to have uneven dichotomies. Socionics is like one big mirror.

[Hitta]

Well you could have unequal dichotomies, but you would have to use theoretical concepts though.