Agency transfer notation

[karff]

This is about causal thinking, but I’m not going to try to convince anyone of anything about it. I just need some help.

I’m interested in constructing a practical notation to aid in the transfer of agency from mind to mind. For the purposes of the notation system, agency is defined as the ability to control a causal function. It’s a practical definition relative to this particular scheme.

I’m interested in the notation system being both intuitive to those familiar with mathematics, and not intimidating to those unfamiliar.

The notation should aid in redefining a known causal function as another causal function, with respect to a particular outcome.

For example, in baking, if you overwork the dough, you get a tough end result. If you teach someone that fact, you have transferred some agency, relative to the outcome of the toughness of the end product. If you redefine the causal function as “mechanical agitation in the presence of water activates the gluten”, you have given them more agency. They can shield the gluten from water with a fat, and they can use high or low gluten flours to control the toughness of the end result.
If you just tell them “Mix butter in with the flour.” or “Use cake flour.” (or bread flour, depending), you have not transferred much agency.

A notation describing this would indicate - Working the dough controls toughness/lightness = mechanical agitation of gluten activates it, WRT baking outcomes.

Eating hard candy is bad for your teeth. Knowing that gives a small amount of agency. For a long time I did not actually know why. I assumed it might be because crushing the candy caused mechanical stress on the teeth. Understanding it as a different causal function would have given me more agency, WRT to dental health outcome. The bacteria Streptococcus Mutens converts sucrose into a cement that it attaches to the teeth, providing a base for a thriving community of cariogenic bacteria. It’s specifically sucrose, the main ingredient in hard candy. When you suck on hard candy, you are continuously bathing a specific bacteria in the raw building material for it to cement itself and its acid-producing buddies to your teeth.

Ever since I learned that particular causal function, I have never let a piece of hard candy pass my lips. The knowledge has allowed me to know that sugar in a dessert, eaten quickly, before a brushing, is relatively harmless, while sugar slowly consumed between meals is catastrophic, WRT dental health outcome.

WRT
The best redefining of a causal function is always relative to a particular outcome.

When you eat hard candy, you are increasing the value of confectionary stocks, WRT investing outcome.

When you eat hard candy, you are replacing one gratifying experience with another, WRT smoking cessation outcome.

When you eat hard candy, you are increasing demand for sugarcane, WRT agricultural outcome.

When you eat hard candy, you are increasing blood sugar, WRT insulin resistance outcome.

And so on.

Part of the purpose of the notation system is to know whether you are actually transferring the greatest amount of agency.
“Is it an actual different causal function, with different cause and effect?”
“Is it not just a directive?”
“Is the new causal function the best one for the particular outcome?”

While the above might look a lot like theory, that term can have a negative connotation to some students (learning theory = being bored to death). So, “unit of agency”, or “term of empowerment” might be better.

Ideally, the notation system should be able to show the best prerequisite, corequisite and postrequisite causal functions to learn, WRT particular outcomes.

Also, when disparate causal functions are best understood with one common one (microwave ovens, Xray machines, and greenhouse gasses all work on the same causal function, WRT to a particular understanding.)

Suggestions?

[jacob]

Tricky stuff. It's rather rare to think in this way, that is, follow a logical chain according to presumptions and accept the logical conclusion.

It's more common to have a preformed conclusion and then go look for evidence that would support a logical chain to the pre-desired conclusion. (Engaging in confirmation bias.) Access to google et al has dialed this up to 11.

Finally, there are a great many people (maybe half the population of humans) who couldn't care less about what the logical consequences may be. For them it is about how it makes them feel or how it makes others feel. Insofar their emotional systems aren't properly "tuned" to reality, perhaps due to past trauma, they'll do irrational things like eating hard candy all day long. For example, if you're already pondering whether the correct naming matters, you're already partially dealing with emotional systems.

To add another common snag. For many humans, the messenger actually matters more than the message itself. If they somehow like or admire the messenger or they are related to them by family of membership, they'll accept the message regardless of how they feel or think about simply out of a sense of generalized loyalty. Such humans have no problem have one opinion in one social group and 5 minutes later having the complete opposite opinion in a different group. Many in this group actually have problems forming their own coherent thoughts. Instead they will take the thoughts from the group and simply regenerate them as their own, like a very situational LLM.

In my experience, dealing with a "receiver" who isn't tuned into the "logical consequences"-channel is a bigger problem than trying to figure out the "best message" for someone is already receiving in good faith.

[daylen]

I've studied some algebra here and there although I much prefer working with geometrical/topological objects. Formalizations can get unwieldly quick as you start introducing symbols/algebras (i.e. combinatorial growth of formalisms). Practically, it would seem to me that recursive, directed cyclic graphs (with labels/descriptions as appropriate) are the best bet for representing causal processes at various levels of granularity/scale/frequency. In essence, smaller loops corresponding to tighter/faster feedback loops encased inside larger loops corresponding to looser/slower feedback loops. Another way to think about this is as nesting control loops (see n'th order cybernetics). You could linearize these recursive graphs into a finite, discrete set of properties or relationships however defined, but in doing so you may loose the clarity of geometric organization.

In the context of complex dynamical systems, causality is generally hard to find and "isolate". There is simply too much going on usually, although rough networks of causal relations can be estimated with a "web of goals" or "web of actions".

[jacob]

But ... suppose someone actually is tuned into the "logical consequences"-channel, useful terminology would indicate the "strength of the causation". This could be given by a probability of support. One way to eliminate a lot of devious language is to replace all adjectives with numbers whenever possible. This completely changes the communication form.

On a side-note, there's a Danish concept called "konsekvenspaedagogik", which I have been unable to find much formal support for. It basically means a "pedagogic philosophy of consequences". Rather than tell [the child] what they're allowed to do or not allowed to do, simply tell them what will happen and proceed to enforce it no matter what. This dials the causation up to 100%.

On yet another note, I suspect this whole enterprise will have an issue with complex chains where outcomes have multiple causes rather than a single variable. Oftentimes humans will highlight their favorite variable in a complex system in order to favor their personal cause.

[daylen]

Logical consequences or probabilities can be included into the geometry with a weighted graph (using numbers, line widths, or even color). An arrow of time embedded into the space/paper could allow the graph to represent causal convergence/divergence into the past/future. Some pasts weighted more heavily being considered more influential towards causing more heavily weighted futures. Algebra and geometry are generally transmutable.

Forrest Landry's triple (choice, change, causality) may be helpful here establishing a (transjective, subjective, objective) or (transcendent, immanent, omniscient) relationship.

Likewise, I am not really trying to convince anyone of anything. Just sharing my take.

[karff]

This is for teaching the willing how to control for an outcome they already desire.

When I first learned about the specific nature of hard candy relative to a particular bacteria, I was a bit irritated that information never goes along with "hard candy is bad for your teeth". I don't know how many people are like me, but I would like a source of information formatted in this manner. (I only learned it from a lengthy deep dive in Wikipedia after a visit to the dentist)

It's as much a tool for students as teachers. A student can look ahead and see specifically what causal functions to learn to master a subject (and which ones are not as necessary.)

Oh, I forgot. It would be preferable if it could be done with the symbols on a standard keyboard.

[black_son_of_gray]

I’m interested in constructing a practical notation to aid in the transfer of agency from mind to mind.

I swear I'm not trying to rain on your parade, but isn't "a practical notation to aid in the transfer of agency from mind to mind"...just, like, a language?

I understand that isn't exactly what you're asking, but I think the overlap in concepts is worth thinking about. For example, you are likely wanting to add this notation to English sentences, right? The thing about English, and part of what makes it the current global language, is that it is both simple and extremely complex. Simple in certain very practical ways (e.g. there aren't a lot of annoying ways to inflect verbs, everything isn't needlessly gendered, a straightforward alphabet vs. characters), but also complex in very practical ways (e.g. absolutely enormous vocabulary options, flexible syntax). Of course, some of these very benefits make communication a little less precise or more ambiguous. I think that is an inherent tradeoff. Indeed, I might argue that some evolutionary-like selection process takes place with languages, and there is a reason that some peter out over time--they simply aren't as practical. (Not saying this is the only reason, of course). Maybe the easiest solution is simply to be very particular in how you construct a sentence in English, so as to make the meaning as close as possible to the intended message? (Side benefit: clearer communication with everyone, regardless of notation system.)

On the other hand, there are languages like Lojban, constructed to be as unambiguous as possible. In general, they are so difficult to deal with that no one speaks them. Basically, the attempt to make them practical rendered them impractical. But maybe there are lessons to be learned from these adventures?

Nevertheless, I will throw out a suggestion: Perhaps it would be possible to create a small dictionary of "particles" that you could append at the end of a sentence which would further clarify the purpose of the sentence. (To some degree, we have that already with punctuation. Punctuation? Punctuation! ) But I was thinking more along the lines of e.g. the Japanese particles, but for "logic" or "agency" or "information content".

Example:
The ball is red AAA.
The ball is red BBB.
The ball is red CCC.
The ball is red DDD.

Four sentences with the same words, but maybe you have defined AAA, BBB, CCC, and DDD, to connote very specific logical/causal implication. E.g. AAA = "a current state", BBB = "intrinsic property [i.e. AAA and BBB are kind of like "ser" and "estar" in Spanish--a distinction that English "to be" doesn't make. But again, one could just add clarification to original sentences, "The ball is currently red." and "The ball is inherently red." Hence my earlier point about just being very careful with the English to begin with.] But maybe CCC = "This is a logical consequence of the former sentence"? That could be interesting! Or maybe DDD = "This is one, but not the only, outcome of the former sentence". While a well-developed notation like this could shorten and disambiguate the prose, it also adds a complexity which might be, itself, very impractical...for instance, if you had 500+ particles that were difficult to remember.

I commend your efforts here, and I'd like to see what you come up with!

[karff]

Maybe notation isn’t the right word. Template is better. Just a template that can be modified similar to modifying an algebraic equation. But, it wouldn’t perform logic, calculations. Would be similar in function to Roman numerals, not Arabic. Descriptive.

The basic template would be :
(A causes B) = (X causes Y), WRT to particular outcome.
Just that alone would be very useful.

The point is to remind both the teacher and student that the most agency is achieved with an entirely new causal function (X is different from A, and Y is also different from B). It also reminds them that it’s relative to a particular outcome, it’s not the thing to do, just ‘cause.

The point is to constrain language. You can say “ When you eat hard candy, you are providing nutrients to cariogenic bacteria, in the most optimal form for them.” But, you can also use language to say “ Don’t eat hard candy.” The template encourages the careful use of the language. It’s an aid.

[jacob]

The basic template would be :
(A causes B) = (X causes Y), WRT to particular outcome.
Just that alone would be very useful.

Predicate logic with its various symbolism might be useful?
See https://en.wikipedia.org/wiki/First-order_logic#Syntax

This was fully developed by Frege in the 19th century but the schoolmen spent A LOT time on a simpler form of this (syllogisms) during the dark ages.
See https://en.wikipedia.org/wiki/Gottlob_Frege

Word of advice though: While you can technically write up a lot of statements, it's impossible to create a fully self-contained system as Russell and Whitehead discovered much to their dismay. However, much of their book (Principia Mathematica) is written in this kind of language. It also more or less inspired the philosophy of positivism which held that only sentences that are logically true are grammatically valid.

[karff]

For agency in financial affairs.

(Not spending your money causes you to still have it) = (Increasing the amount of time you don’t have to make that money in the future.), WRT to saving. That is, when saving money, don’t think of it as money, think of it as time.

(Trading money for a product) = (Trading one store of value for another), WRT to living efficiently. Think of things you buy like a financialized asset.

(Buying an investment) = (Owning a capital good), WRT investing. Don’t think of buying a stock as saving money, think of it as acquiring a machine that produces for you.


Would it be beneficial to have those above units of agency connected in a more systematized way? Note how they are contradictory. You should think of money quite differently depending on the situation, for maximum agency.

*These are just examples, I’m not a financial expert.

[daylen]

Just my two cents: you could try to formulate an equation for translating present portfolio value into future portfolio value (accounting for in/out flows, portfolio constitution, interest aka time value of money). Although, to actively make decisions you have to essentially translate portfolio price trajectories into tangible life value trajectories (or the inverse). Integrating the objective and the subjective through the transjective. The universe can't tell you what you ought to do so much as constrain what you can do.

From a 30,000 ft view: what -> why -> how. What is valued? Why is it valued? How to re-evaluate? Or at least this is one frame to consider.

[7Wannabe5]

Reflections on Ice Breaking

Candy is dandy, but liquor is quicker.

[Henry]

Liquor in the front, poker in the rear.

[jacob]

Would it be beneficial to have those above units of agency connected in a more systematized way? Note how they are contradictory. You should think of money quite differently depending on the situation, for maximum agency.

I'm still not sure I understand your end goal is here, so I'm likely giving the wrong suggestions.

You can build a model or a language of statements with the simplifying assumption that every action can be written as

outcome = skill * effort + environment + random

(This is basically y=ax+b+k*rnd which is a building block for understanding a time series.)

Now, skill*effort falls under agency while environment captures things like privilege and other circumstances you can't control, and random is just an unknown factor that cancels itself out over time or many trials.

IF you emphasize all the terms of this equation, it may help to avoid the trap of attribution bias in which people tend to take personal credit for successful outcomes and blame failures on others, bad luck, privilege. (Often endless boring discussions result from not recognizing all the terms but just rationalizing whichever term according to the desired explanation... as I talked about above.)

For example, in a soccer match, the outcome of the game depends on the above equation. If a player puts in more effort, the likelihood of the team winning or scoring goes up. It will be hard to attribute this directly to the player but over many games, this should become apparent as the random effects cancel out. If you substitute in a more skillful player, the effect should also become apparent. Likewise, it might be that all games are usually played with a tailwind. In that case, skill and effort still matters but not as much as believed. Finally, randomness may have an outsized contribution compared to the other factors. In this case many games are requires to root out the other factors like skill and effort. This is a very common problem when evaluating long-term investors---whereas it's not a problem when evaluating daytraders who perform thousands of trades per year.

So if you use y=ax+b+k*rnd for each connection and put some numbers on it, you may have a better understanding of how agency is involved. E.g.
cavity = A * amount of hard candy + B * bad genetics + C * random. If A is larger than B and C, then you have a lot of agency over your teeth. If A is smaller than B, you still have some influence, but the struggle is harder and may not be worth it. If C is or looks higher than both A and B, you have no detectable agency nor any kind of particular genetic (dis)advantage.

[karff]

Here’s my working hypothesis.

Suppose there are two kinds of thinkers in the world. “Single function” and “multiple function”.

Single function thinkers observe a phenomena, or do a thing, and then accept one causal function as an explanation. They simply assume that there is only one causal function for any given phenomena. If they encounter a different causal function, a different explanation, they will either adopt it as the only one, or they will reject it as wrong. They are sometimes called “narrative thinkers”.

Multiple function thinkers assume there is more than one causal function to explain a given phenomena. When they encounter a different explanation than one they already have, they will assess its value in further explaining the phenomena. They are sometimes called “theoretical thinkers”.

It should be noted that the “theoretical” term confuses the identification of the thinkers. There are many single function thinkers that “know the theory”, as defined by academia, but they have accepted it as the only explanation. They’re still single function.

My hypothesis is that most “teachers” in society are trying to create single function students. Parents, formal educators, moral leaders, corporations, peers, politicians all have a great interest in trying to convince students there is only one way to look at things.

My goal is to oppose that force.

[jacob]

True, see https://en.wikipedia.org/wiki/Model_of_ ... complexity

10% MHC 12 (systems)
40% MHC 10-11 (formal analysis)
35% MHC 9 (narratives)
15% MHC 8 (simple step rules)

Roughly by percentage of adult population, MHC equivalent, and mental construct capacity. The goal of modern education is to raise people to the ability of doing formal analysis when proper scaffolding is in place. This means half the people (the top half) can understand a formal argument (your single function) when they're walked through it or it is presented in an recognizable context.

Without scaffolding, you should drop expectations one step down. Now almost nobody will spontaneously consider the system. Only 10% will spontaneously apply formal causation. ... and 15% won't even remember that there was/is a rule.

This also explain why "the narrative" is considered the defining mode (norm) of thinking for most of humanity. It's the peak of the nonscaffolded distribution.

It also means that only 10% of a random population will be able to understand/appreciate all of the ERE book, while 40% should be able to understand most of it (except chapter 5 & how it all comes together). The majority of people will be sunk though. Almost all nonfiction books are written in a narrative format for a reason.

[Jean]

I think you see why people have been turning knowledge(accurate or not( into stories for milenia now.

I often think that myth are just some mnemotechnique to remember and transmit something.

But when we look it, we don't know if it's some deep philosophical knowledge, a cheese recipy, or direction to a pasture.

[karff]

If you’re making mud pies with a child, and you show them two different ways to make a mud pie, and they understand it, then they have the complexity to see things two different ways.

Suppose you show them one way to make a mud pie. And the two of you make several that way. Then you start making one a different way. The child will, many times say “No, that’s not the way to make a mud pie! You don’t do A, B, C! You do X, Y, Z!” They definitely have the cognitive complexity to grok ABC, they just know it’s “wrong”.

If a child can make a mud pie, and also build a stick fort, they have the necessary complexity to construct something in two different ways - at their level of complexity.

Humans have a tendency to do this:
At their level of complexity, they can do many different tasks, and think of many different causes/effects. They are also able to understand a different way of doing the same task, or a different cause/effect relationship explanation. They will grok it, and then say it’s wrong.

When I was around 5 or 6, I was being taught some of the religion of my family. The gist was, the Devil was part of God’s plan, but also that the Devil hated God and his plan, and wanted to destroy it. I asked, “Why does he even go along with the plan?” The answer was something along the lines of “He hates God so much, he just can’t help himself.”
At 5 or 6, I was not as cognitively complex as most of the adult members of my religion, but I’m pretty sure most of them never actually tried looking at the situation in a different way ( the Devil’s perspective).

If you can do two different things at your level of complexity, then you have the cognitive complexity to look at the same thing in two different ways. Most people just don’t.

I’ve always been a multifunction thinker, just at my level of complexity. And, there are lots of cognitively complex folks who still are only single function, at their level of complexity. (This does make me wonder about nature vs nurture, and how possible it is to move individuals on the continuum).

Edit: I didn't read Jacob's last post very well before responding, but I'll leave this post as it is, as it has an interesting story in it.

[karff]

I'm sorry about the argumentative tone of that last post.
Also, informational complexity is the right paradigm. Are there any simple notations/templates/formats for that?

[Jean]

I know that linguists have metrics relative to information density. I'de look into that to see is there is something that might help you with what you are trying to do.
My first impulse would be to just count how much relationships beetween things there are in a sentence.

So in your hard candy example, i count 5 for the whole explanation
Hardcandy contains sucrose
Bacteria take sucrose
Turn it to cement
Cement good for other bacteria
Other bacteria make caries

While hard candy destroy teeth is only 1

Given so, even the "complete" explanation looks like a story.
I think this is just the way we humans are hardwired to communicate.
It looks to me that you dissatisfaction has more to do with how people just accept holes in their stories, than the use of stories to explain things.