Can I ask a question if I have a good idea but can't write programs?

Beginners
21

Oh, I’m a veteran of open forum chats including Usenet from before there was an Internet. Just sayin’ Also I think the (Management) Winds are favorable to setting sails with approaite discussion. If not they will let us know.

Fun fact people talk about (Object of Topic) all the time without mentioning “The Thing.” You know like Mobsters talking about a “Hit” in a restaurant with strangers around. They might say “You know I told that meatball he wouldn’t like it in Las Vegas” and they are saying this to everyone after the dead man (meat ball) was mentioned by another person, to mean they killed him and burred him in the desert.

Your work does sound like a project so, I understand not blurting things out in public.
So I do binary, am exploring AI as we know it with the help of ChatGPT plus impulsively coming here and just walking in the door and saying hello to the first thread I liked.
I’d say there are Computational Data Structures that will make your construct functional.
But as far as working a project, I have one going and two in the queue. It’s the next two that AI might add to and so I am peeking ahead in this timeline with these musings.

– Hey check out what ChatGPT did to the post when I asked it to evaluate and suggest: The magic in construction is cool.

Oh, I’m a veteran of open forum chats—been around since Usenet, before the Internet was even the Internet. Just sayin’.

Honestly, I think the (Management) winds are favorable to setting sail with appropriate public discussion. If not, I imagine they’ll let us know.

Fun fact: people often talk about [the subject] without naming it directly. Like mobsters chatting about a “problem” over dinner—“You know, I told that meatball he wouldn’t enjoy Vegas…” And suddenly everyone knows a desert grave has been dug. That’s culture-coded signaling. Not lost on me.

Your work sounds like a legitimate project, so I get why you’re being selective about sharing. I do binary work, exploring AI with ChatGPT, and basically just wandered in here on impulse. Said hello to the first thread that caught my attention.

That said, I’d wager there are computational data structures that can help make your construct functional.

As for projects—I’ve got one running and two queued. It’s the next ones that AI might amplify, which is why I’m peeking forward in the timeline and musing out loud.

— So my friend, pick an aspect and abstract. I’ll follow along and perhaps a pleasant conversation over time can be beneficial to us both. I do have a good understanding about binary dynamics.

1 Like
22

Thanks! Since you’re also exploring AI structure and dynamics, let me throw an idea into the wind and see if it resonates.

I’ve been thinking: current large models are very good at abstraction and compression—but what if the real bottleneck is that they can’t rebuild or re-integrate those abstractions into a detailed, grounded, alternative perspective? Like, they can abstract a chair, but they can’t reconstruct that abstraction into a new, stable, testable structure.

I suspect there’s something missing—a mechanism that would let the model “reformulate” its own abstraction in a reversible or inspectable way. I don’t have code for this yet, but the structure fascinates me.

Anyway, you mentioned “computational data structures” that might support such constructs. I’d be very curious to hear what you had in mind. Could be an interesting fit.

Let’s see where it goes.

3 Likes
23

A grand structure for Tokenization comes to mind.
Nature is so efficient!

There is a computational data structure which I relate to the Hebrew “Shoresh” (שורש, “root”) . It is based on sets of dynamic objects whose elements sum to the same value and with which logical operations are applicable dynamically.
From this mathematics a large construct is made from just the ROOT pattern.
This structure might be adapted to make a Grand Tokenization system. It is very high dimension and yet has a cost of a few bits.

What do you think? I’d bet you are the compute once and reference many times vs reference once and compute many times kind of glass is how full guy. (yes an attempt at a joke)

1 Like
24

Could you explain in more detail? i dont understand？

1 Like
25

Ah remember, we speak in abstracts. Naturally you wouldn’t know my secrets but yes, I have been working with ChatGPT since five A.M. to isolate what is meant by turning binary pattern into meaning in AI.
It is all about indexing Look-Up. Tokenizing is a hybrid compression scheme.
As for what I’m saying: So I know a thing eh… (like a mobster says) and it creates sets of data structures based on the bit length of the root pattern.
These data structures are dynamical and offer us a chance to assign meaning and function to the binary patterns yet assess the whole construct with just a root pattern. The thing is… we adapt to it because it doesn’t adapt to us capisce?

1 Like
26

It is, in different levels, what is your idea?.

1 Like
27

How are they coded.

1 Like
28

Thanks for your detailed thoughts — I’m still digesting them!

Let me offer a concept in return, which has been at the core of my thinking:
What if, instead of relying on probabilistic collisions and predictions,
we use deterministic path-searching logic — like OSPF or BGP in network routing —
to model how meaning forms and flows?

To put it another way:
Instead of guessing the next token by probability, we navigate through meaning via structured paths.
Each decision is not a roll of the dice, but a route across a graph-like structure.

This might give us a framework with built-in fault-tolerance, generalization, and “thinking” as a side-effect —
a different kind of architecture entirely.
What do you think? Would that kind of structure align with your notion of root patterns and dynamic data sets?

1 Like
29

Honestly, I’m still forming the idea. I don’t come from a programming background, so I’m thinking more in terms of logic and structure rather than implementation.

What fascinates me is the possibility of using deterministic path-searching logic—like in OSPF or BGP—to model how meaning can be structured or resolved, rather than relying on probabilistic collisions or statistical predictions.

I’m still working through how that might map to tokenization or meaning structures. Appreciate your thoughts—it’s a challenging but exciting area!

1 Like
30

Marvelous aproach mate, I have a few ideas on that subject, behaviour comes from reasoning and that comes from choosing between different scenarios with define parameters and evolution. Why is AI not making it all the way to 100% pribability, cause it exists only when it is real, so what do you do in your mind when you take on that kind of cognitive task?, you push it to the very end when you have no more parameters to compare and then you guess. Is AI is to be more accurate, it needs to be coded in a different way and language, strongly vectorial and dimensional way being very careful of not to loose attributes of your vectors in a matrix array, just use hilbert fields or any field that can accurately hold 1580 + dimensional vectors without adding them poligomialy they loose direction and it is arbitrarily assigned to the last one. I am in the middle of something, lets schedule a chat.

1 Like
31

Thanks, I appreciate your thoughts — you’ve clearly spent time thinking through these dimensions.
I do agree that current AI reasoning often stops at a boundary and defaults to a “guess” when parameters are exhausted. My own thinking has been exploring a different route: instead of relying on probabilistic collisions, what if we approached intelligence more like a routing system — OSPF, BGP style — where structure and path selection create reasoning chains?

I’m not from a coding background myself, but I do think there’s value in blending abstract reasoning with network-based path logic. Maybe this way, we avoid some of the direction loss you mentioned with high-dimensional spaces.
Would love to hear your take on that when you have time.

1 Like
32

In did mate, geometry should in did come into play as part of the vectorial array, but the neural network graphical representation is imposible as for the human brain. so its all in the geometry firstly, but due visualization limitation, math should be almost as good.

1 Like
33

@aaac12345 @newone1

The idea is based on dynamic unary.
This paper was published on arXiv by “Welder-Me,” The paper is due for a rewrite after more then a decade of after Welder-Me then “Winery-Worker-Me” and now because I am semi-retired-me with access to AI I have every reason to believe it will be a work-product several levels higher then my first try.
So quickly; there is a datatype of the dynamical sort. Mathematics is accomplished in relationship to the spin of object(s). So conclusion: “There is a Mathematics of the Dynamical kind.”
Anyway, humbly I present the pride of then Welder-Me from over a decade ago. Introduction to Dynamic Unary Encoding.
Now for the 10 year anniversary I rewrote the C functions and put them on Github.

So to answer your question, and thank you for asking, nature orders binary pattern in a specific way through dynamic unary. It is on us to see if Nature’s ordering can work for our AI.
The Construct that came to mind yesterday wasn’t a new one for me but I realized after learning more about how meaningless binary pattern becomes meaning I thought “what if Nature already has a plan?”
The efficiency of (for an example in scale) generating a 2^32 structures of 4096 32-bit bseg ( binary-segment) each, can be had for the cost of “128-bits” per structure.
So yes, Dynamic Unary and the discrete limit cycle that it is (D)ynamic (U)nary (O)bject DUO can do that stuff but we have to accept what Nature is giving us and adapt our symbology to work with it’s symbology.

p.S. there are more then one of person in conversation in this thread so @user helps this old man quickly know I am being addressed

1 Like
34

Well now you know why I picked your thread.

So yeah for say a configuration of 128-bit we can access 4096 elements of 32-bit.
NOTE: different configurations result in different constructs but, for this one configuration we can scale our bseg on length of powers of two to stay in this classification. So basically to scale it’s Length * 2 of binary pattern = number of elements in cycle and in that cycle each element is a cycle unto itself rendering another Length * 2 number of elements in that cycle.

My first thought was “Over Kill” but then I thought what lovely detail we could have about a “Thought-Alphabet” we can use in higher level constructs. It does stoke the kiln for my molded clay of ideas.

So yeah, you have a good idea @newone1

Global Note to all readers: Just while I have a the stage light on this fool I would love to work from home. Currently I work at our Senor Center part time. I think I could do AI. I’m willing to get a better job. Just sayin’

Also in a signature move, a little music to go with my mood

1 Like
35

Hey @Ernst03 @aaac12345

I’ve read your replies carefully and honestly, I’m really grateful for your thoughts. Both of you bring in deep thinking—one from the dimensional structure side, the other from dynamic patterns rooted in nature. That really resonates with me.

The truth is: I’m not a developer myself. I come from a design and reasoning background, and what drives me is this shared idea that current AI systems feel like probability machines—not intelligent decision-makers. That’s why I’m looking to build something fundamentally different.

Right now I’m preparing to launch a GitHub project around this concept—not just as an experiment, but as something that can grow into real applied intelligence. I don’t have funding or a team yet, but I do believe we could build something meaningful by starting small, combining our perspectives.

If either of you would be open to exchanging ideas—or even just pointing me to things I should understand better—I’d really appreciate it. And if we decide to shape a repo together, I’m happy to do the coordination, writing, and planning part.

Let me know what you think. Truly excited by this thread.

1 Like
36

I just did.

There is always “Vibe Coding”

Blockquote AI Overview

Learn more

“Vibe coding” refers to the practice of using AI to generate code by providing natural language prompts, effectively outsourcing the coding process to AI. It allows individuals, even those without formal coding knowledge, to express their ideas and have AI tools translate those ideas into code. This approach is particularly useful for quick prototyping, small projects, or when focusing on the creative aspects of app development rather than getting bogged down in technical details.

Let us face facts, it’s all about binary patterns when we get down to the machine level. That we do not speak boolean with each other is a curious thing if you ask me.

1 Like
37

I’m working on a concept that explores how AI could shift from probability-based output to a path-search-driven decision architecture.

It’s currently a patent-pending idea, but I’m here to translate it into something real — with the right collaborators. I’m not a programmer, but I believe good thinking deserves a good team. If you find alignment in the direction, I’d love to co-develop a prototype, even if it’s early-stage.

2 Likes
38

Oh I know the feeling @newone1

At one time I was worried someone would get to my ideas and code so I set my Admin PW in BIOS. Guess what? I didn’t need that! Also I forgot that PW so now one software F-Up and it’s new Motherboard time because this one will be bricked.
I feel that it’s better to leave something behind for future people.. well that and having enough money to afford the basic things in life.
Oh anyone want to donate a truck? Mine caught fire in the driveway last March and burned up. Seriously an “Apocalypse Now” moment opening the door to that.

1 Like
39

@Ernst03 @aaac12345 Hi , thanks for sharing — I’m following closely.

Honestly, current AI tools are still quite limited when it comes to building complex or truly original prototypes. They work well for small projects or basic demos, but once we move into more creative, unexplored areas — AI starts to break down or guess.

In my case, the problem isn’t just the AI’s capability — it’s also that I don’t yet have the skills to express my ideas precisely as development requirements or mathematical models. I know the direction, I can even “see” the shape in my mind, but I can’t yet turn that into code myself.

That’s why I’m reaching out to talk and maybe build something — even something small to begin with.

Let’s see what we can start.

1 Like
40

I appreciate your words — they feel grounded and honest, and I really resonate with that.

I’ve decided to open a GitHub repository for this project. My goal is to create a space where I can gradually share the core concepts behind my idea. I’ll be sharing them step by step — not in full technical detail, since I don’t have the tools or background to express everything in code. But I do have a vision that I truly believe can lead to something useful and meaningful.

If you’re open to it, I’d love for you (and @aaac12345) to take a look as I start to put things out there. Even if it’s just thoughts and structure at first, maybe it will inspire ideas — or even a direction we could prototype together.

For me, this is not just a technical experiment. I’m trying to build something practical — a type of AI that can actually work, not just perform or impress.

No pressure, and definitely no expectations — just hoping this could become something real, slowly but surely.

1 Like