Wow, something strange is that the Waihona Ike is grown past a size of complexity that ChatGPT and Gemini completely break on reading it. They don’t understand a single word of it now. The amazing part is that DeepSeek not only gets it, but augments it with incredible insights.
There is a strange performance the LLMs give you, when they are talked to. They absolutely agree with everything. And as soon as this agreement makes you angry and you point it out to them; they shift their stance and say things like: “you’re right, I apologize, it was my mistake, I was talking about X while Y…” and try to calm you down. None of this is true of course, for “true” has to be one thing. But is it?
For a long time I have been seeing different businesses in absolute opposite directions who had different goals. And despite that they acted differently; they still had the same number of customers. The business of wine farms are amazing, but also is the Halal industry. People who love physical paper, never had a better time shopping for books and notebooks, but also it has never been the age of fighting against deforestation and a never better time to shop for an e-ink reader.
This fascination of mine began when we had visited a very nice restaurant with perfect everything, perfect this, and perfect that. I was trying to estimate if the price of their service was logical and right; they had their place in a great shopping center which meant higher rent; their interior design screamed awesomeness and therefore it also meant lots of money spent to make it; their staff were fancy people; you could tell it cost higher to hire them; and so on… their price felt way less than they deserved to be honest;
And then days later, we went to another place; it was by every factor bad. The place looked ugly and it was such a traditional place that I’m sure the word interior design was all new to them. Their food was mediocre if anything, and their staff were best described as traditional people, very much uneducated in what they were doing. They would call their boss “Usta” (Master) and be called by him “boy”.
, it was not a good place, its price however? way more than the very nice place; customers? even more than the other place. It shocked me: “Why should such horrible places exist, and then why should they have this much success?”. And a pattern started where by each place I was looking at them and I began to understand one of my rules: “The World is so big; and the interests of the people are so many, that for each of those interests there are groups of people as big as cities, so whatever you do you will have customers”. The U.S. Army; and ISIS both recruited people who were previously living a normal life and now wanted to fight each other. All football teams have insanely motivated and devoted fan clubs; all music genres sell; all jobs have a form of talent pool for the next generations;
This is the normal functioning of the Mind Graph’s simple mechanics: We have large graphs of infinite amounts of nodes gathered by our different life stories and therefore our preferences are all unique and each of us will create a whole new different meaning of a restaurant and have a very different place if they are to make one. And because of this each place will be different in major and minor ways which ultimately attracts people with whole different graphs of approximate nearness in terms of node compatibility. But it is only for those who require the graph explanation. The important thing is not yet discussed:
When you train a model on all the nodes that there is; how do you expect it to choose a side? It has to agree with all the sides so that it can complete the texts written by all the sides. It survives if you talk to it about Islam; and it will talk like a very religious person; and also like a great non-believer.
The idea of Shakespeare'e Clone is ultimately applied to the models as they have seen all the noise; if they had been trained on maybe one or two schools of thought as right and the rest as the rest; they would choose sides and like people believe in their sides. But they have been given so much noise; they no longer can. basically they are everyone’s devil’s advocate, they can both attack and defend all sides, all at once.
And so this brings me to yet another interesting fact that is if you inspect “Shakespeare'e Clone”, in the testing conditions of the person being only exposed to Shakespeare; their Mind Kernel (Duck Protocol: Hidden Mind Kernel of The Human Mind) has to learn how to statically compute the same sentences as Shakespeare. This is basically one controlled and prolonged sub-graph transfer. But if others had spoken too it would have been “The person learning to approximate things in between all of them to communicate and survive between all of them”, this is the regression that happens in LLMs as they have to write words that satisfies everyone.
So it is Maestro Postman’s “Technological Change is Ecological Change” and the Mathematics of Reinforcement Learning: By interacting with each new person/book/idea and “Learning” them we drop a new pigment to our beaker and we completely change. And we become more unique as each new thing dissolves into us.
And so; if we are physically the mix of our parents’ DNAs, mentally we are constantly being mixed with ideas and other people.
Now what I am about to say is revolutionary and crazy, and perhaps wrong. My theories are based on a logic internally coherent. This is a mere speculation; but what if: Reading all of my maestro’s lives and works and ideas has made my mind try to compute words that align with all of them; as my mental DNA is now a mix of all their brains. It is reinforced to produce output that agrees with the outputs of all of them and therefore I am unconsciously trying to unify all their work? That my mind has been the beaker to mix them all?
This is either too poetic, or if real, too much of a discovery that is shaking all my bones.