ChatGPT (and now, many others) is a modern version of a question answering system. However, all classical (non-quantum-based) question answering systems are limited by information they can get or infer directly from information they can access. Even a quantum computer would not be able to do better, though it might do it faster.
I tested many ideas for using mental effort in conjunction with Bayesian analysis and some AI (patents issued). The system can work with or without mentally-responsive hardware. The system outline is simple: 1) A Local server allows remote users to interact via the Internet. 2) A remote user interface provides a graphical user interface (GUI) to enable users to participate by providing incremental information via a series of brief, focused mental efforts. 3) The Local server receives the results of these efforts, preferably from many simultaneous and/or sequential efforts, and analyzes them together to get the desired information.
Please let me know if you are interested in developing a system for obtaining non-inferable information. I can provide more detailed plans, answer questions, provide hardware if necessary, and even reimburse some expenses (with prior agreement). The company must be commercial to the extent it is at least self supporting – if it doesn’t support itself and further development, it cannot survive. Organizational details will be worked out as necessary.
I am very busy with the Quantum Field Detector and Transmitter systems, which are exquisitely hard to do.
I can explain more, but I prefer specific questions, rather than taking the significant time to write up the idea in detail, which I can’t do right now.
The timing of me reading this syncs quite well with my current “geek gut” directing me in the direction of such a Q&A system (mixed in with my recent deepening interest in the intersection of consciousness & AI/tech/the “status field”).
Count me as interested, I’ll come back when I’ve got the time to formulate some questions and a plan.
I have a large number of MED100Kx8 that could be available for this type of project. In addition, I can explain in detail – plus provide the source code I already have – how to generate, process and use random data from sources in PCs and mobile devices. The processing includes AI (ANN, artificial neural network) and factor analysis to convert the random data into searchable form for the AI.
These two source types allow either central server (via Internet) and portable (non-Internet connected) versions.
Looking forward to this collaboration. I’m moving house soon and have bought a server rack with servers (the house even has a room that is the perfect size for a single rack server ) so I plan to be bring the MED Farm backup in a more stable environment. This would also function perfectly for this project.
@ScottWilber
I guess the best place to start is with the source, which for me can be the setting up a central server. We could call this MED Farm 2.
The source code you have - is it all packaged into one zip/folder or something, or is it separated out by function - for example if we were to first focus here on the RNG source, is the code related to that easily separable from the other code?
Or would it be easier just to share the code and we break it down by folder / function and start to flesh out the details and design here?
Much of the source code is already available through links on this forum, though it’s not organized – it’s been too many years since I had active support for that. I will take a look at what I have and try to make some order to it.
I’ve scratched head about patent. And, if I understand something correctly, it boils down to idea, that MMI information gathering is basically divination+ , but automated.
What it doesn’t encompass, through is that besides statistical verification, it’s better to have some mechanism to verify acquired data. For example, through logical tests or protocol, which should resemble TCP.
@ScottWilber
Ah I was assuming you were referring to some new source code/projects not yet released.
From our earlier discussions on this forum on various threads a few years ago I did my best to sort out the source code as listed below:
Please do share anything else that isn’t here and you think might be relevant to our current endeavor.
Generally speaking, predicting the future by any means can always be called “divination,” but in the field of mental processes or cognition that’s an archaic term implying a non-scientific or supernatural process. I believe it’s necessary to move away from supernatural and unexplainable concepts, because they will never be explainable satisfactorily to scientists (nor should they be).
Most importantly, there are no supernatural laws, only natural ones. Our lack of understanding does not make any manifestation of mind, magical. Again, that is an archaic way of thinking, and can never arrive at a scientific conclusion.
The methods described in my patents are not just automated, but involve objective measurements that are duplicatable. More importantly, they allow many individuals, and/or one individual over time to contribute many mental efforts to arrive at a more accurate result. Of course there are many known and tested statistical methods for assessing accuracy, but a patent is not the proper place to include those. Instead, that would be described in technical papers, such as my Comparison Testing Protocol for A.C.E and MMI_051223.pdf - Google Drive. Further, the data from each individual is assessed a weighting factor based on past accuracy for use in Bayesian updating. Finally, in Advanced Processing Methods for Increased Effect Size in A.C.E. & MMI_061023.pdf - Google Drive, I describe methods that can be used to provide a weighting factor for each trial based on the statistical properties of the internal computations for each trial or block of trials.
Over a period of 30 years and about $5 million invested, an enormous amount of programming and designing was done. I designed all hardware and most processing algorithms, but I am not a programmer (except for Mathematica), so I am not familiar enough with the programs to know where everything is. I do have everything, but that includes about 5TB of backup drives. It will take me a week just to review these drives for specific content.
So, “yes,” there is a lot that is not included in those documents. Also, even the documents you listed, while containing much of the foundational programs, have not been recovered to the point anyone is using them as they are. I formerly had a website, MindEnabled.com (now redirects to CoreInvention.com, but the site can still be found at MindEnabled), which had a collection of user interaction programs and a full-blown history section for each registered user, showing their ranking and scoring progress. That functionality is now disabled, but I think the videos still work.
I think we need to choose a starting point and rebuild the functionality we are most interested in moving forward with. This will take some discussion.
Thank you Scott, I can appreciate there’s a wealth of experience and information and data to sort through. Already a lot has been gathered here on this forum over the past 4 years but it’s just the tip of the ice berg I feel.
Perhaps we start off simple and work on just setting up a source of entropy, and that is me setting up a stable (i.e. online 24/7, and not offline because my son wants to hijack the Linux server to play Minecraft).
This can be as I said, the new MED Farm with the devices I already have but running in my new server rack (I just got the keys to my new property today, we’re moving in over the next couple of weeks and I’ll be setting up kit as time permits)
The MeterFeeder project we worked on serves as a basis for raw access to multiple QNG USB devices but it needs a much better API gateway built around it (that’s just one of many things that need doing).
I’m also thinking the forum needs some new topics/tags: GCP and Q&A Oracle, or something to that degree.
Another side project I have ongoing is building some GenAI/LLM model to be trained on esoteric data without the bias found in the main ones out there. I’m now wondering how much of that can be leveraged to help us all sort and organize and query the 5TB of data you have.
I agree, start as simple as possible. We can discuss the gateway in more detail as we make other design decisions (some suggested below).
I suggest using only MED100Kx8 generators as data sources. In my previous systems, each user was granted exclusive use of a single generator. This simplifies the server and gives each user the best effect size capability available. If you can serve two or three simultaneous connections, that will be sufficient for a while. The server should be able to grow incrementally as more users want to join in.
The server, including the user interface, should be tested with at least one real application before going online. This will reveal some of the more obvious issues.
I agree on the need to have functionality in place to “reserve” a particular device for a user or a specific use case/experiment to avoid mixed signals (as much as we can even try…) but do you have any ideas how we can present to the user which device is allocated to them?
Perhaps show them the device’s serial number? Or perhaps a realtime webcam view of the actual device so they can intuitively “connect” with that one vs the other devices in the farm. (I always wondered what it’d be like if I somehow involved my cat in these are experimental setups to ensure there was another biological non human no bias connection between target human and target RNG)
Another thing I’ve wanted to ask for a while is do I need to be careful of the kind of USB hub I use for connecting multiple devices to the same server? I’m not a hardware guy but I did geek a lot in electronics when I was young and before I encountered computers (go Dick Smith DIY electronic kits for kids! so I’m wondering whether voltage supply etc can be an influencing factor on such setups.
I’m thinking of creating a new category on this forum called Project Alexandria to focus discussions on this new MED farm source and following question and answer systems that we can all collaborate on.
The user doesn’t need to know specifically which device they are using, that would likely just be a distraction. The user connects by the feedback provided. This is why user feedback of some sort is very important. Consider, the user never has any direct awareness of the random generator process or the numbers it is outputting at kilohertz to megahertz rates.
The hub should be USB 2.0/High Speed to handle multiple devices. Also, make sure the power adapter is good for about 1 Amp so there is no issue driving many devices. The hub must be self-powered since the current capability of the USB connection on the computer is usually not adequate. Do Not get USB 3 or above (including USB C), as the cable connectors and the USB protocols are not compatible. Example hub: Amazon.com. Make sure there is enough room around each device when they are plugged into the hub.
The MED100Kx8 has internal regulators for every supply where it may affect operation, so any voltage within the USB voltage specification (4.5-5.5V) will be unaffected by supply voltage.
Not quite. Divination of all sorts, basically is a method of obtaining of intelligence, using focused intention to organize chaotic process to symbolically represent desired answer.
What is clearly new in MMI related methods is possibility to find statistical significance of result. How likely it is, that certain answer wasn’t completely random.
Otherwise, there isn’t too much difference, really.
I agree, that there isn’t really anything supernatural, yet, direct manifestation of mental influence often considered to be field of magic. I’m not talking about magic like it’s something, what couldn’t be explained and investigated. I’m talking about it as a set of methods, which are being used by mages and occultists. And, yes, sufficiently advanced technology indistinguishable from magic, sufficiently analyzed magic is indistinguishable from technology.
So, I don’t think, that we should ditch off experience of thousands years old traditions, which used similar, yet less advanced tech to obtain information with MMI.
What I think is important for question answering system?
Way to perform feedback. Basically, for now, all intelligence gathering with MMI is powered with human mind, or with animal. Where certain individual or group of individuals focus intention to shift probabilities. However, where and how should focus operator to reliably direct thought process of AI towards sensible answer? Should operator focus to increase z score of MED? Or one rather should focus to increase confidence score of AI, which uses MMI to derive answer from results of automated divination?
What’s really problematic is that it’s nigh impossible to make AI to perform a guess and appraise, how useful it was within optimal timeframe for MMI feedback.
1.1. There is a reason, why I asked about MMI transmitter. Because, I think, that active scanner is possible. That system, which measure how transmitter with constant power affecting MMI sensor might be far more sensitive, than direct measurement of mind influence. As well, I suspect, that it might be far more effective to project, well, sensor thoughtform, which will represent entire symbolic system. Which will be used to represent answer.
Strategy of questioning. Well, it’s relatively easily possible to implement agentic workflow, which will derive answers from some sort of mind enabled feedback. Either binary answers, or some symbolic data. However, that workflow should be adapted to pretty much noised results. What’s more important, that AI should be able to take in account statistical significance of certain answer. and adjust own strategy accordingly.
I think, that agentic workflow should be designed, using simulated data, where we use AI in training to use tree of thought to get desired answer from conventional QA AI, which will simulate MMI source(it’s answer will be noised with gauss distribution). Then, entire thought process could be optimized with DSPy.
Symbolic system. There is a reason, why I pointed out to divination and experience of mages. Thing is, that virtually no one uses coin flip to obtain answers through divination. Yet, there are divination systems, which use several binary values to derive answer. Like I-ching or geomancy. I suspect, that human mind simply can’t reliably focus on ONE bit. Human mind uses concepts and images. And it’s simplier to project one of 16 figures, than one bit.
Also, ambiguity of symbols in divination systems is sort of failsafe to not to go in completely wrong direction about obtained data from very noised out observations. Yet it allows to more clearly derive next questions to clarify meaning of obtained symbols. If one bit in certain symbol will be wrong, it won’t get us completely astray. And we’ll be able to compensate it with additional questions.
Yet, they are meaningful enough to derive something conclusive in few tries. With probability up to 70% for experienced practicers. And, I need to point it out, that it’s possible with classic divination, where every question has only one cast of symbol without any statistical properties.
) so I plan to be bring the MED Farm backup in a more stable environment. This would also function perfectly for this project.
so I’m wondering whether voltage supply etc can be an influencing factor on such setups.