Just now, cladking said:

Symbolic language has limitations and that shows up in both human communication and AI prompting.

All forms of language have limitations, beginning with being able to understand said language.

1 minute ago, npts2020 said:

All forms of language have limitations, beginning with being able to understand said language.

Procedural languages don't really have limitations in interpretation. Whatever you were going to say about your oak tree would have been meant for that specific oak but would have no meaning applied to any other.

If you express a mathematical equation or a computer program incorrectly there is no meaning and communication fails. If a bee stumbles when it dances or is confused observers see only nonsense. You can't fix procedural errors through parsing. Procedural language allows the listener to do something and convey specific knowledge correctly. A computer runs or a bee buzzes.

I'm not suggesting we abandon English. Perhaps it should be rationalized or a parallel scientific language created but English works about goodly enough. I'm suggesting that we need to understand its effect on our thinking and that these effects are sometimes determinative in understanding LLM output. I'm suggesting only humans use symbolic language and only after the axial age. I'm suggesting that science maps reality and each of us mistakes this map for the terrain.

Edited Tuesday at 12:24 PM1 day by cladking

23 hours ago, cladking said:

I picture it soft and cuddly able to hold lots of people safely. To each his own.

That was one of my attempts at humor but if you look more closely there's a great deal of information in it about the way I think. Greenland only looks big because we distort a sphere to project it on paper. Of course the earth isn't really a sphere and paper doesn't always lie flat. And you can't walk around the north pole to go back in time no matter human definitions and conventions.

We have inherited not only a way to think but also a set of assumptions, some of which I believe are incorrect.

Humans think we’re describing reality but we're actually describing the artifacts of our tools.

Your getting yourself confused, we have inherited our way of thinking it's called 'our culture', there's an old saying that's pertinent "It takes a village to raise a child" or "No man is an island, in and of himself".

Humans think therefore we are; not a perfectly logical thought, but it's close enough... 😉

27 minutes ago, cladking said:

Procedural languages don't really have limitations in interpretation. Whatever you were going to say about your oak tree would have been meant for that specific oak but would have no meaning applied to any other.

What if I too have an oak that fits the description? In fact, that is the fundamental rule of conveying meaning, we have to share an experience, tree et all, for the information to have any 'real' meaning.

2 hours ago, cladking said:

Procedural languages don't really have limitations in interpretation. Whatever you were going to say about your oak tree would have been meant for that specific oak but would have no meaning applied to any other.

If you express a mathematical equation or a computer program incorrectly there is no meaning and communication fails. If a bee stumbles when it dances or is confused observers see only nonsense. You can't fix procedural errors through parsing. Procedural language allows the listener to do something and convey specific knowledge correctly. A computer runs or a bee buzzes.

I'm not suggesting we abandon English. Perhaps it should be rationalized or a parallel scientific language created but English works about goodly enough. I'm suggesting that we need to understand its effect on our thinking and that these effects are sometimes determinative in understanding LLM output. I'm suggesting only humans use symbolic language and only after the axial age. I'm suggesting that science maps reality and each of us mistakes this map for the terrain.

what do you actually know, about what you think you understand?

3 hours ago, dimreepr said:

what do you actually know, about what you think you understand?

Nothing at all really. But I can make deductions about all sorts of things that I otherwise could not. These things might eventually lead to far better prediction and can be used to augment experimental science because they span domains. The only one relevant in this thread is procedural thought can help some individuals better prompt LLM"s. I'm sure one could even prompt AI to answer in a shared procedural language because AI can translate between frames and domains.

I predict this is the future at least until actual machine intelligence is invented.

Copilot suggests I add abstract thinkers tend to treat the prompt as a known quantity; procedural thinkers treat it as a structure that must be aligned.

5 hours ago, dimreepr said:

What if I too have an oak that fits the description? In fact, that is the fundamental rule of conveying meaning, we have to share an experience, tree et all, for the information to have any 'real' meaning.

Procedural communication relies principally on shared logic rather than shared referents. This is why we have so much difficulty understand Bee and Prairie Dog; we are looking for referents rather than logic. We must understand referents, all the referents, to understand any utterance. To follow a scent trail an animal needs little more than the ability to smell. Procedural languages have fixed rules and apply to all referents including your oak tree. One oak tree plus zero oak trees equals one oak tree.

This is why bees, ants, prairie dogs, dolphins, and computers can communicate flawlessly within their systems even though they share no referents with us. Symbolic language is powerful but inaccurate. Procedural language is rigid but precise.

5 hours ago, cladking said:

Procedural languages don't really have limitations in interpretation.

Of course they do.

But you are so narrowly focused on procedural v symbolic that you are ignoring the inputs and thought of others. You are not receiving their messages.

For instance swansont asked you about intent.

When I go to a cafe and order a cup of coffee, my intent is clear and comes from me, the speaker, not the waitress who is a listener.

No 'interpretation' is necessary.

I have asked you for examples of some of your statements and classifications, and am still waiting.

I will go first.

A good example of a procedural language (in my opinion) is sign language.

5 hours ago, dimreepr said:

Your getting yourself confused, we have inherited our way of thinking it's called 'our culture', there's an old saying that's pertinent "It takes a village to raise a child" or "No man is an island, in and of himself".

Humans think therefore we are; not a perfectly logical thought, but it's close enough...

Every man is a product of his time and place but this makes nobody right or wrong.

We got off on tangents. The worst one is the belief that because civilization has advanced for 4000 years that this is a natural state of all things. It is a very very unnatural state except for individual life and this exists because of free will. Civilization has gotten better not because it's alive but rather because language has evolved toward clearer thinking; thinking more in resonance with reality. This is largely the result of extrapolation of experimental science.

But we are still confused because all things exist including all those things that don't think at all which includes everything in existence except homo circularis rationatio. Just remembering this one fact when you prompt will help a lot.

8 minutes ago, studiot said:

A good example of a procedural language (in my opinion) is sign language.

No.

Sign language is just symbolic language with no words.

Procedural languages are literal, inflexible, unparsable, representative, and ordered. Terms are mathematically related to one another and each representation is definitive to the entire structure of the sentence. If you try to parse 2 + 2 = 4 you will probably get a wrong answer. If you drop any of the five representations in the sentence you get gobbledty gook. If you add or substitute you get gobbledtygook. It's not wrong, its meaningless. A slime mold crossing its own trail is lost. A dancing bee which is out of step has no meaning. A pheromone trail to nowhere is a waste.

Errors in symbolic communication rarely make statements completely meaningless.

Computer code used to be notorious for being gobbledty gook with the tiniest error. I learned programming back in the '60's when the tiniest error like omission of a line ender that was only visible on the tape (called an "x-off") would prevent the program from running. Any flaw would usually turn the program into noise make trouble shooting quite difficult. I imagine now day the computer will just tell you where the errors are or fix them for you. Programming might be the epitome of procedural language though some animal languages should probably be considered much more complex.

So far no computer program, no LLM, and no machine is considered to have intent. "Intent" requires agency and free will. Animals think procedurally and have free will. Computers act procedurally to parse and deconstruct symbolic languages.

OK, Copilot says I'm wrong about it being able tom correct errors and wants me to add procedural languages can't be parsed, reinterpreted, or “understood differently” and can only be executed exactly as composed. If they aren't executable they are gobbledtygook.

7 hours ago, cladking said:

This is why bees, ants, prairie dogs, dolphins, and computers can communicate flawlessly

How do you know this is true?

14 minutes ago, npts2020 said:

How do you know this is true?

Bees find the flowers, ants operate their cities, prairie dogs maintain their colonies, dolphins coordinate hunts, and we can observe computers.

Sure, it's largely assumption that bees never go to the wrong place and that hive collapse isn't related to communication problems (they go out and don't come back so that's getting lost not going to the wrong place). Nature is brutal to bad behavior and that brutality assures the bad behavior will not be replicated in the gene pool. I might remind you too, that these are all just animals that are highly successful with tiny little brains and no abstractions. Any language lacking abstraction will be shown to be procedural language based on logic rather than shared referents.

Animals can survive with almost no brain at all. How big is a mite's brain?

I can't “know” animal communication is flawless in the symbolic sense but the question itself is symbolic. Nature always propels individuals toward proper behavior because their DNA is procedural logic of what has worked in the past. If it quits working than the species might become extinct.

Edited yesterday at 02:12 AM1 day by cladking

It is obvious that we have a clear problem.

You have stated several times that the meaning is determined by the 'receiver'.

This leads to the obvious paradox that no meaning can ever be transmitted, discussion is impossible and that it is pointless me writing these words since only the reader can determine the meaning and I, the writer cannot set it.

Hence communication is impossible.

Good day to you.

16 hours ago, cladking said:

Procedural communication relies principally on shared logic rather than shared referents. This is why we have so much difficulty understand Bee and Prairie Dog; we are looking for referents rather than logic. We must understand referents, all the referents, to understand any utterance. To follow a scent trail an animal needs little more than the ability to smell. Procedural languages have fixed rules and apply to all referents including your oak tree. One oak tree plus zero oak trees equals one oak tree.

This is why bees, ants, prairie dogs, dolphins, and computers can communicate flawlessly within their systems even though they share no referents with us. Symbolic language is powerful but inaccurate. Procedural language is rigid but precise.

If we share no referents and you can’t understand unless you know all referents, how did we decipher the bee dance? You say it’s symbolic and therefore and yet we have been able to parse it.

How is it not symbolic while sign language allegedly is?

9 hours ago, cladking said:

Bees find the flowers, ants operate their cities, prairie dogs maintain their colonies, dolphins coordinate hunts, and we can observe computers.

Sure, it's largely assumption that bees never go to the wrong place and that hive collapse isn't related to communication problems (they go out and don't come back so that's getting lost not going to the wrong place). Nature is brutal to bad behavior and that brutality assures the bad behavior will not be replicated in the gene pool. I might remind you too, that these are all just animals that are highly successful with tiny little brains and no abstractions. Any language lacking abstraction will be shown to be procedural language based on logic rather than shared referents.

Bee's have many different job's, some clean the hive, some feed the Queen who's only job is to spit out offspring, some are nannie's, some are guarding the hive, some are etc. and some of the abnormal bee's also found their place in the hive; it's kinda fundamental knowledge nowadays, the abnormal is responsible for evolution.

It's how 'we' the hive came to be... 😉

I

17 hours ago, cladking said:

Every man is a product of his time and place but this makes nobody right or wrong.

If my mechanic is able to get my car running, whilst I failed to elicit a spark, with my random hammer blows in the general direction of the engine block; I'd be forced to conlude that he knows more than me, so logically one of us is, at least, more correct than the other; unless we're talking about Dr nobody, the professor of philosophical indecision.

2 hours ago, studiot said:

It is obvious that we have a clear problem.

You have stated several times that the meaning is determined by the 'receiver'.

This leads to the obvious paradox that no meaning can ever be transmitted, discussion is impossible and that it is pointless me writing these words since only the reader can determine the meaning and I, the writer cannot set it.

Hence communication is impossible.

Good day to you.

Welcome to my world!

I didn't create it and was merely born into it and have been trying to communicate. But all isn't lost and because Arabs, and Jews, scientists and theologians, and even republicans and democrats are trying to communicate because we now have LLM's to translate. Rather than demonizing one another, common ground is forming under us.

People don't seem to realize that the listener is the one who parses the utterance and to take author intent we must not only share referents but we must understand the author's premises. The exact same sentence written by a scientist and theologian has very different meanings. We should always begin the job of parsing with humility and assuming the author makes sense because he does. Everybody makes sense in terms of his premises and sees a reality that reflects those premises.

1 hour ago, swansont said:

If we share no referents and you can’t understand unless you know all referents, how did we decipher the bee dance? You say it’s symbolic and therefore and yet we have been able to parse it.

How is it not symbolic while sign language allegedly is?

I seriously doubt we have really deciphered the Waggle Dance yet, but have merely taken the first big step to see that it's reflective of solar position. This alone has broad implications to what it means to be a bee. There are probable several more "words" that we don't understand yet. In any case a procedural language can be deduced from it it does and how it looks. One thing that makes it easier is that each "word" has a single fixed meaning and the meaning is literal because word order is fixed. If A and B but not C then D. We have to understand symbolic languages all at once and often can't be certain that we really do. Even after it is translated the lack of shared referents might obscure most of the intended meaning.

But procedural languages are "easy" to translate because they are literal, can be deduced because they reflect reality, and can be understood a little bit at a time. With nothing except simple sentences like "If A and B but not C then [unknown]" we are provided huge amounts of knowledge about that unknown word. It's mostly just busy work and simple arithmetic; simple deduction.

Of course it's probably not possible to reverse engineer the nature of computer code from studying a complex or a simple computer program. I'm still learning what LLM's can do and how they work and will never have many (if any) insights into their programs. But animal communication will be largely solved and I have no doubt almost every word in every language will prove to be procedural and almost every word will lack any sort of what I call "abstraction". DNA and brains are binary. Things exist or they don't. Life copies this logic to think and communicate but humans must use a symbolic means of communication and have since even before axial thinking.

Sign language is simply symbolic language expressed through gesture instead of sound.

Symbolic languages require interpretation while procedural languages require execution. LLMs operate procedurally.

Edited 13 hours ago13 hr by cladking
clarification

46 minutes ago, cladking said:

Welcome to my world!

I didn't create it and was merely born into it and have been trying to communicate.

No, you've just been very trying, you haven't even attempted to communicate, you've just assumed correctness; your world seems wrong, in a fascistic sort of way.

1 hour ago, cladking said:

I seriously doubt we have really deciphered the Waggle Dance yet, but have merely taken the first big step to see that it's reflective of solar position. This alone has broad implications to what it means to be a bee. There are probable several more "words" that we don't understand yet. In any case a procedural language can be deduced from it it does and how it looks. One thing that makes it easier is that each "word" has a single fixed meaning and the meaning is literal because word order is fixed. If A and B but not C then D. We have to understand symbolic languages all at once and often can't be certain that we really do. Even after it is translated the lack of shared referents might obscure most of the intended meaning.

But procedural languages are "easy" to translate because they are literal, can be deduced because they reflect reality, and can be understood a little bit at a time. With nothing except simple sentences like "If A and B but not C then [unknown]" we are provided huge amounts of knowledge about that unknown word. It's mostly just busy work and simple arithmetic; simple deduction.

Which is it? Are they unparsable and we “must understand referents, all the referents, to understand any utterance” or do we understand some of the waggle dance?

1 hour ago, cladking said:

Sign language is simply symbolic language expressed through gesture instead of sound.

Symbolic languages require interpretation while procedural languages require execution. LLMs operate procedurally.

Waggling isn’t? You “know” this, even though we haven’t figured it all out yet? You “know” the movements are literal?

1 hour ago, swansont said:

Which is it? Are they unparsable and we “must understand referents, all the referents, to understand any utterance” or do we understand some of the waggle dance?

Both are true. Symbolic language has no meaning without shared referents and parsing. But in procedural language any parsing destroys the meaning. You can't rearrange terms, brackets, or operations in a formula or equation. You can't paraphrase a computer program. But if you have part of a procedural language you can begin deducing the rest of it. The Waggle dance is apparently procedural language and we understand at least some of it.

4 hours ago, dimreepr said:

If my mechanic is able to get my car running, whilst I failed to elicit a spark, with my random hammer blows in the general direction of the engine block; I'd be forced to conlude that he knows more than me, so logically one of us is, at least, more correct than the other; unless we're talking about Dr nobody, the professor of philosophical indecision.

Ah, but remember everyone makes sense.

There has to be a reason a gymnast can improve, an auto mechanic can fix a cosmologist's car that he can not, or LLM's can help some people more than others. A mother needn't understand physics or adrenaline to lift a car off her baby and a bee doesn't need to know angles and vectors to find flowers and tell others where to look. Everybody makes sense but the difference is bees, computers, and mathematicians are on the same page because their communication is solely or chiefly procedural.

Long before science there was art, literature, and many types of complex knowledge.

3 hours ago, dimreepr said:

No, you've just been very trying, you haven't even attempted to communicate, you've just assumed correctness; your world seems wrong, in a fascistic sort of way.

I assume reality exists as people perceive it and that everyone makes sense. That’s not “assuming correctness,” it’s just acknowledging that each person operates inside a frame they didn’t design. I use tautologies because that's the only way to communicate across frames without pretending we all share the same premises.

You’re reading my model as authoritarian, but it’s actually the opposite. There’s no “one correct worldview,” only different premises producing different reconstructions. That’s why communication is hard, not because people are wrong, but because their frames differ.

My world isn’t fascistic, it’s pluralistic. Everyone makes sense relative to their premises, and nobody has a monopoly on interpretation. That’s the whole point. The only thing we have to guide us is experiment and reason.

26 minutes ago, cladking said:

Both are true.

Not if your claim is true. “All” and “some” are in conflict.

26 minutes ago, cladking said:

Symbolic language has no meaning without shared referents and parsing. But in procedural language any parsing destroys the meaning. You can't rearrange terms, brackets, or operations in a formula or equation.

Not always, but it can be done. You can rearrange addition and subtraction, and also multiplication and division. Operations that commute can be done in any order.

26 minutes ago, cladking said:

You can't paraphrase a computer program.

Paraphrasing is restating in different terms but keeping the same meaning.

There are lots of ways to write different code that achieves the same output. e.g. multiple sorting algorithms exist.

26 minutes ago, cladking said:

But if you have part of a procedural language you can begin deducing the rest of it. The Waggle dance is apparently procedural language and we understand at least some of it.

Again, this is in conflict with your claim that you need all of it.

2 hours ago, swansont said:

Waggling isn’t? You “know” this, even though we haven’t figured it all out yet? You “know” the movements are literal?

Literality appears to be a characteristic of procedural language. If a bee waggles to indicate north after beginning the sentence with a waggle (meaning "toward the flowers" then the flowers are north. Sufficient numbers of bees must be dispatched to harvest them. Procedural systems can be deduced, symbolic languages can not.

13 minutes ago, swansont said:

Not if your claim is true. “All” and “some” are in conflict.

All symbolic languages must be parsed and no procedural ones can.

Parsing procedural language destroys the meaning and there is no meaning in symbolic language until it is parsed.

Bees don't parse the Waggle Dance. They experience it. We parse everything and don't even notice (usually).

17 minutes ago, swansont said:

Not always, but it can be done. You can rearrange addition and subtraction, and also multiplication and division. Operations that commute can be done in any order.

The more complex the equation the less likely that random changes won't destroy the meaning which is the equality.

19 minutes ago, swansont said:

Again, this is in conflict with your claim that you need all of it.

You need shared referents to understand author intent in symbolic language because they determine the most likely parsing. You needs touchstones to know you are doing it properly. Such communication is fairly good when everyone shares premises but there is always some deviation between intent and parsed meaning. When we query LLM's they must first use their algorithms to parse your question and estimate the framing. They do this procedurally using electricity flowing through procedural circuits guided by procedural programming. They predict your framing and the framing of the desired response. "Errors" are primarily bad estimations caused by bad prompts where "bad prompt" is defined as indefinite framing.

15 minutes ago, cladking said:

The more complex the equation the less likely that random changes won't destroy the meaning which is the equality.

But that's true of any communication and is a function of its entropy.

1 minute ago, KJW said:

But that's true of any communication and is a function of its entropy.

Complex sentences aren't totally destroyed by one or two random changes such as typos, omitted words, or the usage of incorrect definitions or grammar. They can often still be parsed similarly to author intent.

Procedural languages degrade catastrophically under noise. Symbolic languages degrade gracefully.

13 minutes ago, cladking said:

Literality appears to be a characteristic of procedural language. If a bee waggles to indicate north after beginning the sentence with a waggle (meaning "toward the flowers" then the flowers are north. Sufficient numbers of bees must be dispatched to harvest them. Procedural systems can be deduced, symbolic languages can not.

All symbolic languages must be parsed and no procedural ones can.

Parsing procedural language destroys the meaning and there is no meaning in symbolic language until it is parsed.

Bees don't parse the Waggle Dance. They experience it. We parse everything and don't even notice (usually).

The more complex the equation the less likely that random changes won't destroy the meaning which is the equality.

I don't suppose your definition of parse is any more consistent than your definition of symbol, procedure, equation, meaning, receiver etc coincides with ours.

Just now, studiot said:

I don't suppose your definition of parse is any more consistent than your definition of symbol, procedure, equation, meaning, receiver etc coincides with ours.

Perhaps not.

If I say "The Judge's sentence was ten years" you probably judge the judge meant the offender was sentenced to ten years. Parsing is the ongoing real time process of judging words in sentences to conform to what you believe the author means. No sentence has meaning until it is parsed.

Symbolic languages don’t carry meaning in their symbols; meaning arises only when the listener parses the utterance through their own premises, referents, and context.

Edited 9 hours ago9 hr by cladking

2 minutes ago, cladking said:

Complex sentences aren't totally destroyed by one or two random changes such as typos, omitted words, or the usage of incorrect definitions or grammar. They can often still be parsed similarly to author intent.

Procedural languages degrade catastrophically under noise. Symbolic languages degrade gracefully.

As I said, it is a function of its entropy. Human language is highly redundant. That's why it is relatively insensitive to noise, not because it is symbolic. In principle, a parser of a computer language could correct syntax errors, again due to redundancy of the language.

3 minutes ago, KJW said:

As I said, it is a function of its entropy. Human language is highly redundant. That's why it is relatively insensitive to noise, not because it is symbolic. In principle, a parser of a computer language could correct syntax errors, again due to redundancy of the language.

There isn't always a clear distinction between symbolic and procedural statements. But I am talking about distinct language types. All modern human human languages are symbolic and all non-symbolic languages are procedural. One could, in theory, express procedural logic in a symbolic language but then each listener is still going to parse it and change it.

Computational languages are procedural as well.

This behavior requires only a few procedural rules to occur. They are well known.

6 minutes ago, cladking said:

All modern human human languages are symbolic and all non-symbolic languages are procedural.

What about photographs, or music?