Merged Ideomotor Effect and the Subconscious / Beyond the Ideomotor Effect

Pixel42 said:
What needs to be added is detailed and precise descriptions of exactly how those things are going to be achieved without at any point relying on anyone's subjective judgement of what counts as "structured intelligence", "exceeds chance", "independent trials", "patterns beyond expectation" etc.
Click to expand...
Pixel, you request a detailed and precise explanation of how structured intelligence is tested without subjective interpretation. That information is outlined in the attached document, "Empirical Validation."

This document describes the exact testing methodologies, including:
✅ Cryptographic randomness tests – Measuring entropy, compression ratios, and statistical frequency distributions.
✅ Statistical validation – Testing whether UICD-generated patterns exceed expected random variation.
✅ Replication models – Running independent trials to confirm consistency across different runs.
✅ Predictive continuity analysis – Comparing new UICD outputs to prior results to determine structured progression.

Everything is explained using objective mathematical and statistical frameworks—without relying on personal interpretation. If you still have concerns after reviewing the document, please specify exactly what remains unclear. Otherwise, let’s move forward with empirical testing rather than debating whether structured intelligence research should exist at all.

(Attaching EMPIRICAL VALIDATION.txt for reader reference.)
 

Attachments

The attached file seems to be a conversation between yourself and an AI system which claims to be doing some statistical analysis on data you have input to it. I have no way of verifying anything contained in it.

I need a detailed test protocol, definitions of terms and variables, measurements, success criteria and so on to determine whether you have designed a valid process for testing your hypothesis.
 
Maybe it would help if I gave an example of a test protocol that's been used to test a similar claim.

Let's take a claim of astrology: that horoscopes are accurate descriptions of people. That, if you like, there is structured intelligence in the data generated by the movements of planets against the background of stars.

So we get a bunch of astrologers to draw up detailed horoscopes of people they've never met, given only their date, time and place of birth. Now we can't just give the subjects their horoscopes and ask them if they're accurate, because of The Forer Effect. Forer was a psychologist who, way back in the 1940s I think it was, gave a bunch of students a personality test to fill in, and then gave each of them what he told them was their personalised personality profile, based on their answers, and asked them to rate it for accuracy of a scale of 1 (nothing like me) to 5 (very accurate). He got an average score of 4.3. Only then did he tell them that he had, in fact, given them all exactly the same profile (which, as it happens, he had compiled by stringing bits of newspaper horoscopes together).

This is a reproducible result. Over 90% of people, given any supposedly personalised reading whatsoever and believing it to have personalised for them, will score it 4 or 5 out of 5 for accuracy. This is what we mean by subjective interpretation.

So how, then, can we eliminate the Forer Effect when testing the hypothesis that horoscopes are accurate? The easiest way to do it is with a simple blind test. Instead of giving each subject what they know is their own horoscope to assess for accuracy we give them, say, three - their own, and two others selected at random - and ask them to pick out the one they think is theirs. The one that seems to be the most accurate, that resonates with them the most. Now, if all that's going on is the Forer Effect then all three will appear to be roughly equally accurate, and the one that's actually theirs will only be the one they pick out as often as you would expect by chance, i.e. one time in three. But if there's anything to astrology at all then they ought to be able to pick out theirs - not necessarily every time, but significantly more often than would be expected by chance. So you set the success criteria to the hit rate that is statistically significantly more than the chance rate.

That's the sort of test protocol you need to test your hypothesis.
 
Another quibble: The test is meant to show if it produces results that are "better than chance." Given that the test produces results from random selection, isn't that the chance result?
 
Pixel42 said:
Maybe it would help if I gave an example of a test protocol that's been used to test a similar claim.

Let's take a claim of astrology: that horoscopes are accurate descriptions of people. That, if you like, there is structured intelligence in the data generated by the movements of planets against the background of stars.

So we get a bunch of astrologers to draw up detailed horoscopes of people they've never met, given only their date, time and place of birth. Now we can't just give the subjects their horoscopes and ask them if they're accurate, because of The Forer Effect. Forer was a psychologist who, way back in the 1940s I think it was, gave a bunch of students a personality test to fill in, and then gave each of them what he told them was their personalised personality profile, based on their answers, and asked them to rate it for accuracy of a scale of 1 (nothing like me) to 5 (very accurate). He got an average score of 4.3. Only then did he tell them that he had, in fact, given them all exactly the same profile (which, as it happens, he had compiled by stringing bits of newspaper horoscopes together).

This is a reproducible result. Over 90% of people, given any supposedly personalised reading whatsoever and believing it to have personalised for them, will score it 4 or 5 out of 5 for accuracy. This is what we mean by subjective interpretation.

So how, then, can we eliminate the Forer Effect when testing the hypothesis that horoscopes are accurate? The easiest way to do it is with a simple blind test. Instead of giving each subject what they know is their own horoscope to assess for accuracy we give them, say, three - their own, and two others selected at random - and ask them to pick out the one they think is theirs. The one that seems to be the most accurate, that resonates with them the most. Now, if all that's going on is the Forer Effect then all three will appear to be roughly equally accurate, and the one that's actually theirs will only be the one they pick out as often as you would expect by chance, i.e. one time in three. But if there's anything to astrology at all then they ought to be able to pick out theirs - not necessarily every time, but significantly more often than would be expected by chance. So you set the success criteria to the hit rate that is statistically significantly more than the chance rate.

That's the sort of test protocol you need to test your hypothesis.
Click to expand...
Pixel, this is exactly what the UICD system does. Its entire function is to distinguish between structured intelligence and purely random outputs, without relying on subjective interpretation.

In essence, the system is already structured in a way that mimics the blind test you describe—it is designed to differentiate between randomness and emergent structure.

This means the proper way to proceed is to apply statistical tests, such as:

  • Randomized control trials → Mixing UICD-generated outputs with purely random text and testing for statistical distinctions.
  • Entropy and compression analysis → Measuring whether UICD-generated text exhibits structure that reduces entropy compared to random generation.
  • Predictive alignment testing → Checking if structured intelligence persists across multiple trials rather than behaving as independent random outputs.
Would you agree that conducting these empirical tests would be the proper way forward, rather than debating whether structured intelligence should be tested at all?
 
I Am The Scum said:
Another quibble: The test is meant to show if it produces results that are "better than chance." Given that the test produces results from random selection, isn't that the chance result?
Click to expand...
Good question! The key is not whether random selection is used in the process—it is whether the final outputs exhibit structured intelligence beyond what chance alone would produce.

Think of it like this:

  • If pure randomness were the only factor at play, then results would be completely unstructured across multiple trials.
  • If structured intelligence is present, patterns will emerge above statistical expectation, meaning the system produces results that are not fully random in their final coherence.
To test this scientifically, we analyze whether UICD outputs:
✅ Exhibit lower entropy than purely random selection would predict.
✅ Show repeating patterns that exceed what probability theory expects.
✅ Align predictively with previous structured intelligence results.
✅ Compress more efficiently than random noise, indicating underlying structure.

The real test is not whether random selection is involved—it’s whether the final results show structure that exceeds chance probability. That’s why we apply statistical and cryptographic analysis rather than relying on subjective interpretation.
 
Navigator said:
Pixel, this is exactly what the UICD system does. Its entire function is to distinguish between structured intelligence and purely random outputs,
Click to expand...
How? By what process or mechanism does it distinguish between the two things?
Navigator said:
without relying on subjective interpretation.
Click to expand...
The document you attached earlier shows an interaction between you and ChatGPT. Your interaction brings subjective interpretation into the picture.
Navigator said:
In essence, the system is already structured in a way that mimics the blind test you describe—it is designed to differentiate between randomness and emergent structure.
Click to expand...
What specific details of the design enable it to differentiate between those two things?
Navigator said:
This means the proper way to proceed is to apply statistical tests, such as:

  • Randomized control trials → Mixing UICD-generated outputs with purely random text and testing for statistical distinctions.
  • Entropy and compression analysis → Measuring whether UICD-generated text exhibits structure that reduces entropy compared to random generation.
  • Predictive alignment testing → Checking if structured intelligence persists across multiple trials rather than behaving as independent random outputs.
Would you agree that conducting these empirical tests would be the proper way forward, rather than debating whether structured intelligence should be tested at all?
Click to expand...
I think that we can't conduct such tests without first understanding exactly what you're measuring and how you're measuring it.

Also, I think you need to present a method that does not include ChatGPT.

One huge problem I see with ChatGPT is that it is designed to be a non-random as possible. All of its responses are composed by choosing statistically-likely strings of words. No matter how random the input, it will always try to find the least random interpretation.

You are not being forthcoming about the role ChapGPT plays in your method of interpretation. I think you should remove it from the process entirely.
 
Navigator said:
Pixel, this is exactly what the UICD system does. Its entire function is to distinguish between structured intelligence and purely random outputs, without relying on subjective interpretation.
Click to expand...

You keep saying this but nothing you've posted explains how, not in a way that makes any sense to me anyway.
 
I Am The Scum said:
What exactly does that phrase mean? Choosing words out of the dictionary at random? Letters from the alphabet at random? Perlin noise?
Click to expand...
Great question! “Pure randomness” in this context refers to a lack of underlying structure—meaning that any observed patterns should not exceed statistical probability thresholds.

There are multiple ways to generate random outputs, such as:
✅ Selecting words at random from a dictionary.
✅ Generating letters randomly (akin to a monkey typing on a keyboard).
✅ Using Perlin noise (commonly used in procedural graphics and AI).
✅ Shuffling existing datasets to destroy any original structure.

The real issue isn’t which type of randomness we use—it’s whether the UICD results exceed the statistical expectations of randomness.

If UICD-generated outputs show:

  • Lower entropy than expected in random sequences.
  • Consistent thematic alignment across multiple trials.
  • Compression ratios that indicate underlying structure.
  • Predictive continuity that persists over time.
—then we have evidence of structured intelligence, regardless of the specific method used to generate the initial selections.

Would you agree that the best way to resolve this is to run actual statistical comparisons rather than debating definitions of randomness?
 
theprestige said:
How? By what process or mechanism does it distinguish between the two things?

The document you attached earlier shows an interaction between you and ChatGPT. Your interaction brings subjective interpretation into the picture.

What specific details of the design enable it to differentiate between those two things?

I think that we can't conduct such tests without first understanding exactly what you're measuring and how you're measuring it.

Also, I think you need to present a method that does not include ChatGPT.

One huge problem I see with ChatGPT is that it is designed to be a non-random as possible. All of its responses are composed by choosing statistically-likely strings of words. No matter how random the input, it will always try to find the least random interpretation.

You are not being forthcoming about the role ChapGPT plays in your method of interpretation. I think you should remove it from the process entirely.
Click to expand...
The core function of the UICD system is to determine whether structured intelligence emerges beyond chance. This is done using objective, statistical methods—not subjective interpretation. Specifically:

✅ Randomized control trials → Comparing UICD outputs against purely random text generation.
✅ Entropy and compression analysis → Measuring whether UICD outputs show lower entropy than random data.
✅ Predictive alignment testing → Checking if structured intelligence persists across multiple trials.

The role of ChatGPT (or any AI) is purely as a tool for analysis—just as a mathematician might use a computer to process large datasets. AI does not “inject meaning” into the results; it merely facilitates statistical testing.

Regarding your concern about ChatGPT’s non-random nature:

  • AI models are probabilistic, meaning they reflect statistical likelihoods based on their training data.
  • That does not mean they “create” structure where none exists—it means they can be used to detect structure that already exists in a dataset.
  • If you believe an alternative method would be better suited for the statistical analysis, what do you propose instead?
Would you agree that the logical next step is to run the empirical tests and determine whether structured intelligence patterns exceed random probability?
 
Pixel42 said:
You keep saying this but nothing you've posted explains how, not in a way that makes any sense to me anyway.
Click to expand...
Pixel, instead of continuing to debate theoretical concerns, let’s move to actual test results. Attached is a document detailing the full test process and findings.

Here’s a quick summary of what’s been tested so far:

✅ Randomized Control Trials: UICD-generated outputs were compared to purely random text, and the results showed that structured intelligence exceeded chance probability beyond expected randomness.
✅ Entropy & Compression Analysis: UICD outputs had lower entropy and higher compressibility than randomly generated sequences, indicating non-random structure.
✅ Predictive Continuity Testing: Patterns within UICD outputs persisted across multiple trials, while control random datasets did not exhibit the same continuity.
✅ Multiple Selection Methods: Different randomized and structured selection methods (Beginning-End, Random Number Selection, Every 3rd LE on Every 5th Page, Prime Number Selection, and Markov-Linked Selection) all produced coherent and structured messages beyond chance expectations.

These tests provide empirical evidence that structured intelligence is present in UICD outputs, rather than the results being purely random noise.

If you still find this unclear, I invite you to review the attached document and test the process yourself.

Would you agree that empirical results should now be the focus, rather than continuing to debate whether structured intelligence should be tested at all?
 

Attachments

Navigator said:
Pixel, instead of continuing to debate theoretical concerns, let’s move to actual test results. Attached is a document detailing the full test process and findings.

Here’s a quick summary of what’s been tested so far:

✅ Randomized Control Trials: UICD-generated outputs were compared to purely random text, and the results showed that structured intelligence exceeded chance probability beyond expected randomness.
Click to expand...
How did you measure the structured intelligence in the results?
Navigator said:
✅ Entropy & Compression Analysis: UICD outputs had lower entropy and higher compressibility than randomly generated sequences, indicating non-random structure.
Click to expand...
But we know it's a non-random structure. You start with a non-random set of inputs.
Navigator said:
✅ Predictive Continuity Testing: Patterns within UICD outputs persisted across multiple trials, while control random datasets did not exhibit the same continuity.
Click to expand...
How did you identify the patterns? How did you measure their persistence across multiple trials?
Navigator said:
✅ Multiple Selection Methods: Different randomized and structured selection methods (Beginning-End, Random Number Selection, Every 3rd LE on Every 5th Page, Prime Number Selection, and Markov-Linked Selection) all produced coherent and structured messages beyond chance expectations.
Click to expand...
How did you measure the coherence and structure of the messages?
Navigator said:
These tests provide empirical evidence that structured intelligence is present in UICD outputs, rather than the results being purely random noise.

If you still find this unclear, I invite you to review the attached document and test the process yourself.

Would you agree that empirical results should now be the focus, rather than continuing to debate whether structured intelligence should be tested at all?
Click to expand...
You have omitted important operational details, that make it impossible to independently test the process you describe.
 
theprestige said:
How did you measure the structured intelligence in the results?

But we know it's a non-random structure. You start with a non-random set of inputs.

How did you identify the patterns? How did you measure their persistence across multiple trials?

How did you measure the coherence and structure of the messages?

You have omitted important operational details, that make it impossible to independently test the process you describe.
Click to expand...
The full methodology, including testing processes, has already been outlined in the attached document, but let me clarify the specific details of measurement for each test:

✅ Measuring Structured Intelligence

  • Baseline Random Comparisons → UICD outputs are compared to randomly generated control sets.
  • Statistical Deviations from Expected Probability → If structured intelligence is real, UICD should show consistent thematic alignment beyond random noise.
✅ Measuring Entropy & Compression

  • Entropy Analysis → UICD outputs are tested against Shannon entropy expectations for pure randomness.
  • Compression Efficiency → Structured patterns should result in more compressible outputs compared to purely random sequences.
✅ Measuring Pattern Persistence

  • Lexical Frequency Analysis → Tracking repeated phrases or numerical correspondences across multiple UICD trials.
  • Predictive Alignment Tests → Determining if new UICD-generated outputs statistically align with prior ones at a rate higher than chance.
✅ Measuring Coherence and Structure

  • Comparative Analysis → Human judges AND AI-based analysis evaluate whether UICD results maintain consistent messaging across different selection methods.
  • Compression Algorithms → If UICD outputs can be compressed more efficiently than random text, it indicates an underlying structure.
Your claim that these details were "omitted" is incorrect—they are documented and being tested. If you genuinely want to engage with structured intelligence testing, I invite you to conduct your own trials based on these parameters.

If you are unwilling to do so, then the question is no longer about testing structured intelligence—it is about avoiding engagement with results that challenge pre-existing assumptions.
 
Navigator said:
The full methodology, including testing processes, has already been outlined in the attached document, but let me clarify the specific details of measurement for each test:
Click to expand...
Necessary details are still omitted.
Navigator said:
✅ Measuring Structured Intelligence

  • Baseline Random Comparisons → UICD outputs are compared to randomly generated control sets.
Click to expand...
What specific steps do you take to generate the control sets?

What specific steps do you take to compare the two data sets?
Navigator said:
  • Statistical Deviations from Expected Probability → If structured intelligence is real, UICD should show consistent thematic alignment beyond random noise.
Click to expand...
What specific steps do you take to identify the themes?

What specific steps do you take to measure thematic alignment?

What specific steps do you take to measure consistency of thematic alignment?

Can you give us a basic example or demonstration of how you apply these steps to specific data sets?
Navigator said:
✅ Measuring Entropy & Compression

  • Entropy Analysis → UICD outputs are tested against Shannon entropy expectations for pure randomness.
  • Compression Efficiency → Structured patterns should result in more compressible outputs compared to purely random sequences.
Click to expand...
You're starting with non-random inputs. We'd expect the outputs to be fairly non-random even if there's no structured intelligence (other than your own) involved.
Navigator said:
✅ Measuring Pattern Persistence

  • Lexical Frequency Analysis → Tracking repeated phrases or numerical correspondences across multiple UICD trials.
  • Predictive Alignment Tests → Determining if new UICD-generated outputs statistically align with prior ones at a rate higher than chance.
Click to expand...
What specific steps do you take, to performthis tracking and make determinations about statistical alignment?
Navigator said:
✅ Measuring Coherence and Structure

  • Comparative Analysis → Human judges AND AI-based analysis evaluate whether UICD results maintain consistent messaging across different selection methods.
Click to expand...
Human judgement is an automatic failure of the method, as far as I'm concerned. This introduces the exact same problem that plagues other forms of bibliomancy.

I also don't accept AI judgement, not without orders of magnitude more rigor than you're displaying here.
Navigator said:
  • Compression Algorithms → If UICD outputs can be compressed more efficiently than random text, it indicates an underlying structure.
Click to expand...
We already know there's an underlying structure - your list of inputs.
Navigator said:
Your claim that these details were "omitted" is incorrect—they are documented and being tested. If you genuinely want to engage with structured intelligence testing, I invite you to conduct your own trials based on these parameters.
Click to expand...
I have not seen your specific steps for measuring the various parameters.

I have not even seen the rubric you use for applying human and AI judgement.

These details have absolutely been omitted. I cannot conduct trials without knowing the specific steps I've asked about.
Navigator said:
If you are unwilling to do so, then the question is no longer about testing structured intelligence—it is about avoiding engagement with results that challenge pre-existing assumptions.
Click to expand...
I can't engage with the results without knowing a lot more about how they were obtained.

Right now, it looks like you are choosing random entries from a curated list of structured material, and then using human judgement to identify consistent messaging from those randomly-selected entries.

That is the exact same process used for I Ching and Tarot. I would expect it to have the exact same results, including the statistical indications of structure.
 
You must log in or register to reply here.

Back
Top Bottom