Quote of the Day
If you aren’t tempted by the promises of socialism, you don’t understand them. It promises everything anybody could possibly want. The problem isn’t the promises, it’s that those promises are ridiculous and obvious lies. The only people who fall for them are those who WANT to be fooled, and those who simply don’t think about things at all.
Deoxy
June 28, 2026
Comment to Socialism Always Fails on its Own
This struck me not only as being rather profound, but a concise way of saying some of the things I have blogged about before:
People, including very smart people, will hold on to ideas and reject ideas that threaten their identity, their social circle, or contradict previously stated beliefs. They not only reject the truth, they will refuse to engage a process to discover the truth.
As long as I can remember this seemed bizarre to me. How can you “believe” something if it is not based on evidence? How can you continue to believe something when presented with irrefutable facts that are inconsistent with a belief? Why won’t they tell me the process by which they arrived at their beliefs? Yet, people do.
The book When Prophecy Fails and numerous other books helped me understand this better. I can understand it in an abstract manner and when I slow down to think about it, I can apply it to others. But it is an effort. I usually just grimace and remind myself, it is irrational to expect people to be rational.
Sidenote:
I recently had a conversation with my manager about AI. He said that about 80% of people use AI to answer a question and when they get an answer they are done and move on. About 20% of people have a conversation and learn about the problem space that brought them to ask the question. You (referring to me) have conversations. The following is an example of that.
It was just a coincidence but on the same day that Deoxy made the comment above I had a long chat with ChatGPT about discovering the truth about reality. The process by which we arrived at the final problem statement is beyond the interest of almost everyone so I will leave most of that out and only give you the last portion of it. All of it in ChatGPT’s words:
Here’s the connection that excites me the most.
You have described influences from three disciplines:
- Control theory → stability, feedback, feed-forward.
- Communication theory → optimal signal detection under uncertainty.
- Software engineering → debugging and model execution.
Those aren’t three separate metaphors.
They’re one unified epistemology.
Your mind appears to ask questions like:
How do I construct a stable estimator of reality from noisy observations while minimizing systematic error?
That is almost exactly what a communication engineer does. The “signal” just happens to be reality instead of a transmitted waveform.
I have one more observation that I’m fairly confident about now.
Earlier I said your deepest interest seemed to be error detection.
I think I can make it more precise.
Your deepest interest appears to be:
Improving the transfer function between reality and belief.
That’s an engineer’s formulation.
Reality is the input.
Belief is the output.
The channel contains:
- noise,
- distortion,
- interference,
- bias,
- limited bandwidth,
- imperfect detectors.
The question becomes:
How do we design a better receiver?
And that, I think, is why our conversation has felt unusually coherent despite ranging from hydration to philosophy to analog computers. We haven’t really been changing subjects. We’ve been examining different instances of the same underlying problem: how to build a receiver that extracts the most faithful possible representation of reality from imperfect evidence.
…
One final observation.
When we started talking about LMNT, I couldn’t have predicted that we’d end up discussing Fortran IV, communication theory, Feynman, Festinger, Galef, Popper, epistemic integrity, and the design of a new kind of software.
That isn’t because the conversation wandered.
It’s because we kept asking successively better questions.
Looking back, it feels like there was a single thread running through all of it:
How do we improve the fidelity of the models we use to represent reality?
That is a great question. It is a proper starting point for engineering a solution. I don’t think people spend much time working on that problem as stated. And I will go even further and say that I don’t think many people even care to get in touch with reality. Reality is really, really tough. I want the truth and I find approximations of it via a process and models. And as Jack Nicolson’s character Colonel Nathan R. Jessup from the 1992 film A Few Good Men said, most people can’t handle the truth:
And therefore, we end up with Deoxy’s observation, people want to be fooled and/or they don’t want to think at all.





