Peter Hroššo

Posts

Sorted by New

Wiki Contributions

Comments

Frame Control

I wonder if you can infer de facto intent from the consequences, ie, not the intents-that-they-think-they-had, but more the intents they actually had.

I believe this is possible. When I was reading the OP, I was checking with myself how I am defending myself from malicious frame control. I think I am semi-consciously modeling the motivation (=intent they actually had, as you call it) behind everything people around me do (not just say, as the communication bandwidth in real life is much broader). I'd be very surprised if most people wouldn't be doing something similar at least on the sub-conscious level. 

The difficult part in my opinion is:

1) Make this subconscious information (aka intuition) consciously available and well calibrated

2) Actually trust this intuition, as the frame-controller is adversarially undermining your trust in your own sense making and actively hiding their true motivations, so usually your intuition will have high uncertainty

Frame Control

Based on about a dozen of Said's comments I read I don't expect them to update on what I'm gonna write. But I wanted to formulate my observations, interpretations, and beliefs based on their comments anyway. Mostly for myself and if it's of value to other people, even better (which Said actually supports in another comment 🙂).

  • Said refuses to try and see the world via the glasses presented in the OP
    • In other words, Said refuses to inhabit Aella's frame
  • Said denies the existence of the natural concept frame and denies any usefulness of it even if it were a mere fake concept
  • It seems to me that Said is really confident about their frame and is signaling against inhabiting other people's frames

Most people usually aren’t onto anything good, so this, again, ought to be the default assumption.

  • It seems to me that Said actually believes there is no value in inhabiting other people's frames

This seems bad, actually. It seems to me like a sign of insecurity and unjustified submission. I, for one, have no interest in having my conversation partners signal that they’re vulnerable to me (nor have I any interest in signaling to that I’m vulnerable to them).

Everyone has vulnerabilities. Showing them and thus becoming vulnerable doesn't signal insecurity or submission, actually the opposite. It requires high self-confidence (self-acceptance?) and signals openness and honesty to the other person. The benefit is that it leads to significantly deeper interactions.

And the benefit of inhabiting another one's frame? If I use the "camera position and orientation" definition of a frame mentioned by Vaniver, inhabiting other person's frame allows you to see things that may be occluded from your point of view and thus give you new evidence. The least it can give you is a new interpretation of data that you gathered yourself. But it can possibly introduce genuinely new evidence to you, because frames serve as lenses and by making you focus on one thing they also make you subconsciously ignore other things.

Gradient hacking
If your model is deceptive, though, then it might know all of that

Could you please describe your intuition behind how the model could know the meta-optimizer is going to perform checks on deceptive behavior?

Steelmanning Divination

Could the phenomenon described in the post explain why people find psychedelics useful for self-development?

There is the random perturbation - seeing music, hearing thoughts, ...

The authority of an old sage performing divinations is replaced in psychedelics with direct experience of the perturbation. And the perturbation is amplified by the feeling of detachedness from one-self, people often have on a trip.

I don't have any experience with psychedelics, though, so I'm just theorizing.

Forum participation as a research strategy

I don't have that much experience with forums - when I was in research I learned mostly from reading scientific papers + googling stuff to understand them. But I definitely agree that being more active and engaged in the discussion is helpful.

Aside from the topic of research, I used to be very passive on my social nets and basically just consumed content created by others. But after I became more active I feel like I am getting more value out of it and at the same time spend less time there, as formulating questions or ideas takes effort. So it's a natural constraint.