News:

PD.com: We occur at random among your children.

Main Menu

Unofficial What are you Reading Thread?

Started by Thurnez Isa, December 03, 2006, 04:11:35 PM

Previous topic - Next topic

chaotic neutral observer

"Perceptible" includes "perceptible only with the aid of instrumentation."

Taking off my glasses does not make reality go away.

Things which are imperceptible either directly or indirectly do not matter, for practical purposes.  (This does not include things which are not perceptible just yet, because they indeed might matter.  There are doubtless many hidden interesting things in the world just waiting for someone to develop the right test equipment.)

Desine fata deum flecti sperare precando.

Doktor Howl

Quote from: chaotic neutral observer on February 03, 2020, 05:07:23 PM
"Perceptible" includes "perceptible only with the aid of instrumentation."

Taking off my glasses does not make reality go away.

Things which are imperceptible either directly or indirectly do not matter, for practical purposes.  (This does not include things which are not perceptible just yet, because they indeed might matter.  There are doubtless many hidden interesting things in the world just waiting for someone to develop the right test equipment.)

Yep.  For a tool-using species, instruments are as valid as your own eyeballs.
Molon Lube

LMNO

I'm trying to see how this isn't another swing at something Korzybski did back in 1933.

Doktor Howl

Quote from: LMNO on February 03, 2020, 06:51:29 PM
I'm trying to see how this isn't another swing at something Korzybski did back in 1933.

You're correct.  It isn't.

https://en.wikipedia.org/wiki/Alfred_Korzybski

Molon Lube

Doktor Howl

I mean, assuming my senses are accurately interpreting the written word.  :lol:
Molon Lube

tyrannosaurus vex

a few chapters later: Hoffman has invoked evolution simulations where the data (apparently) indicates that sensory systems evolve to help organisms detect the "fitness payoffs" (in terms of survival and procreation) present in the environment, not the environment per se. Moreover, organisms which perceive the environment accurately and directly (can directly detect the presence of necessary resources) have a distinct evolutionary disadvantage with comparable organisms that perceive it only indirectly (can only detect variances in availability of those resources). I think the gist here is that the additional computation required to calculate the fitness payoffs from direct information about the environment is a waste of resources, so evolution has taken the load off by building such calculation directly into the mechanisms we use for perception. Or something.

Then he goes on to enumerate numerous experiments in physics which disprove local realism, but rather than invoking quantum woo he presents various additional experiments, calculations, and theories that reinforce these findings at macro (as opposed to only quantum) scales. It is, tbh, hard to follow on account of I'm not a theoretical physicist. Eventually he concludes that one can banish the apparent contradictions between quantum mechanics and general relativity by not assuming spacetime to be fundamental at all, but rather by treating spacetime and everything it contains as arising from (or simultaneously with) our observation of it. Cue appeals to more experiments that show observation and measurement of a system determine not only its present state but also its history. This leads to a sort of "it's like solipsism but not really because I promise it isn't" conclusion.

Anyway, I think where he's going with this in a practical sense is a theory of conscious experience where the reason the solution to the Hard Problem is so elusive is because we assume consciousness is either identical to or an emergent property of physical processes in neurons, when in reality consciousness is fundamental. So the question of whether or not we can create conscious systems is sort of moot, because literally everything is consciousness anyway (like panpsychism), so what we should be doing is trying to make artificial systems that can relate to us in terms of how they perceive, not in terms of how they think. [this paragraph is 100% conjecture tho, I haven't actually read anything like this here, I'm just extrapolating - probably badly]
Evil and Unfeeling Arse-Flenser From The City of the Damned.

minuspace

Quote from: Cramulus on February 03, 2020, 03:40:20 PM
[...]
and not to get too esoteric, but to me,
the most useful thing isn't "what we can do" with this information, but that the experience of pondering it actually opens us, in certain ways.
[...]


This Disclosive model of truth-seeking kinda jives with me more than correspondence theories, generally. Granted, that might be because the language also evokes coitus but who's to say there's anything wrong with that. Objectively, the surface upon which I press for any purchase is almost entirely empty space, so... much barstool, very hard problem.

[yes, I am an episode of existential constipation: remember Korzybski]

altered

Quote from: tyrannosaurus vex on February 03, 2020, 08:39:37 PM
a few chapters later: Hoffman has invoked evolution simulations where the data (apparently) indicates that sensory systems evolve to help organisms detect the "fitness payoffs" (in terms of survival and procreation) present in the environment, not the environment per se. Moreover, organisms which perceive the environment accurately and directly (can directly detect the presence of necessary resources) have a distinct evolutionary disadvantage with comparable organisms that perceive it only indirectly (can only detect variances in availability of those resources). I think the gist here is that the additional computation required to calculate the fitness payoffs from direct information about the environment is a waste of resources, so evolution has taken the load off by building such calculation directly into the mechanisms we use for perception. Or something.

Then he goes on to enumerate numerous experiments in physics which disprove local realism, but rather than invoking quantum woo he presents various additional experiments, calculations, and theories that reinforce these findings at macro (as opposed to only quantum) scales. It is, tbh, hard to follow on account of I'm not a theoretical physicist. Eventually he concludes that one can banish the apparent contradictions between quantum mechanics and general relativity by not assuming spacetime to be fundamental at all, but rather by treating spacetime and everything it contains as arising from (or simultaneously with) our observation of it. Cue appeals to more experiments that show observation and measurement of a system determine not only its present state but also its history. This leads to a sort of "it's like solipsism but not really because I promise it isn't" conclusion.

Anyway, I think where he's going with this in a practical sense is a theory of conscious experience where the reason the solution to the Hard Problem is so elusive is because we assume consciousness is either identical to or an emergent property of physical processes in neurons, when in reality consciousness is fundamental. So the question of whether or not we can create conscious systems is sort of moot, because literally everything is consciousness anyway (like panpsychism), so what we should be doing is trying to make artificial systems that can relate to us in terms of how they perceive, not in terms of how they think. [this paragraph is 100% conjecture tho, I haven't actually read anything like this here, I'm just extrapolating - probably badly]

Regarding those experiments... they're exactly what I thought they'd be.

Thing is, evolution isn't a fitness maximizer. It's a fitness minimum avoider. Yes, having your environment lie to you is most fit, probably. No, having that is not necessary to survive. Anywhere.

Furthermore, evolution tends to require that all prior steps in its development aren't actively detrimental to the fitness of the organism. Hallucinating non physical objects WITHOUT any strong correlation to reality is actively detrimental. It's also far easier than evolving a new Augmented Reality apparatus of the fucking brain.

The simulated organisms would have died out from psychosis before they developed his fitness maximized solution.

If the data contradicts me, I'm cool with that, but I was actually guessing it was "augmented reality is the Most Fit sensory modality!!!! Here's a directed "find best fit" system that evolves it!!!!!!" while totally ignoring that there is no guiding hand but death and destruction here in the real world... So, yknow.
"I am that worst of all type of criminal...I cannot bring myself to do what you tell me, because you told me."

There's over 100 of us in this meat-suit. You'd think it runs like a ship, but it's more like a hundred and ten angry ghosts having an old-school QuakeWorld tournament, three people desperately trying to make sure the gamers don't go hungry or soil themselves, and the Facilities manager weeping in the corner as the garbage piles high.

Doktor Howl

Quote from: altered on February 03, 2020, 09:09:39 PM

Thing is, evolution isn't a fitness maximizer. It's a fitness minimum avoider. Yes, having your environment lie to you is most fit, probably. No, having that is not necessary to survive. Anywhere.

Furthermore, evolution tends to require that all prior steps in its development aren't actively detrimental to the fitness of the organism. Hallucinating non physical objects WITHOUT any strong correlation to reality is actively detrimental. It's also far easier than evolving a new Augmented Reality apparatus of the fucking brain.

The simulated organisms would have died out from psychosis before they developed his fitness maximized solution.


Boom.  Boom, I say.
Molon Lube

minuspace

Quote from: tyrannosaurus vex on February 03, 2020, 08:39:37 PM

[...]

Anyway, I think where he's going with this in a practical sense is a theory of conscious experience where the reason the solution to the Hard Problem is so elusive is because we assume consciousness is either identical to or an emergent property of physical processes in neurons, when in reality consciousness is fundamental. So the question of whether or not we can create conscious systems is sort of moot, because literally everything is consciousness anyway (like panpsychism), so what we should be doing is trying to make artificial systems that can relate to us in terms of how they perceive, not in terms of how they think. [this paragraph is 100% conjecture tho, I haven't actually read anything like this here, I'm just extrapolating - probably badly]

Consciousness may be a back of tricks, but its still above my pay grade.

I clearly also can't see the big picture but there are various interfaces I find compelling, perception being one of them. With perception, without too many reductions or Husserlian suspensions, I can at least approximate a grasp of what it means to be conscious of something. Still, even if I train some CNN or Reverse Boltzmann Machine to recognize and identify objects by way of various weighted rewards for the fitness of their predictions, we still only have a glorified number cruncher, not a "computer of quality" for which differences were intentional or meaningful in the way we pretend to understand consciousness.

So, for one, I suppose I feel like some kind of a dualist today maintaining that not everything is conscious, at least especially the models I trained werent.

tyrannosaurus vex

Well. I finished the book and all I can say is I want both my money and the 10 or so hours I spent on it back.

After pages and pages of pained and overly complicated arguments and appeals to quantum hocus pocus, his ultimate conclusion on AI is, more or less verbatim: "it is a mistake to try and reproduce a conscious agent by increasing complexity of circuitry, because consciousness isn't created that way. instead we should use circuitry to open a portal into the realm of conscious agents." I mean. What... what the fuck does that even mean.

I knew what I was getting myself into, but after hitting me over the head with painfully dumb metaphors about computer desktops and VR video games, I was hoping at least for a perspective on AI that wasn't just making alphabet soup from the words we use to think about it. Oh well, that's what I get for delving into popular science, I guess.

At least I have a somewhat better idea of what theoretical physicists mean when they talk about the 3D universe as a "simulation". Even though I still think that's bollocks, it's nice to know there's some math that goes into describing it and it isn't just lazy science fiction. Entirely.
Evil and Unfeeling Arse-Flenser From The City of the Damned.

Doktor Howl

Quote from: tyrannosaurus vex on February 04, 2020, 05:31:22 PM
Well. I finished the book and all I can say is I want both my money and the 10 or so hours I spent on it back.

After pages and pages of pained and overly complicated arguments and appeals to quantum hocus pocus, his ultimate conclusion on AI is, more or less verbatim: "it is a mistake to try and reproduce a conscious agent by increasing complexity of circuitry, because consciousness isn't created that way. instead we should use circuitry to open a portal into the realm of conscious agents." I mean. What... what the fuck does that even mean.

I knew what I was getting myself into, but after hitting me over the head with painfully dumb metaphors about computer desktops and VR video games, I was hoping at least for a perspective on AI that wasn't just making alphabet soup from the words we use to think about it. Oh well, that's what I get for delving into popular science, I guess.

At least I have a somewhat better idea of what theoretical physicists mean when they talk about the 3D universe as a "simulation". Even though I still think that's bollocks, it's nice to know there's some math that goes into describing it and it isn't just lazy science fiction. Entirely.

I went into the wrong line of work, really.
Molon Lube

tyrannosaurus vex

Quote from: Doktor Howl on February 04, 2020, 05:32:03 PM
I went into the wrong line of work, really.

The title of the Appendix is literally "the right to be wrong".
Evil and Unfeeling Arse-Flenser From The City of the Damned.

Doktor Howl

Quote from: tyrannosaurus vex on February 04, 2020, 05:34:52 PM
Quote from: Doktor Howl on February 04, 2020, 05:32:03 PM
I went into the wrong line of work, really.

The title of the Appendix is literally "the right to be wrong".

Mine would be "the right to be stupid."
Molon Lube

chaotic neutral observer

Escape From Colditz (P.R. Reid)

During World War II, the castle at Colditz served as a German high-security prisoner-of-war camp.  It was reserved for officers who had escaped from other camps, been recaptured, and were considered at high risk for escaping again.  Later in the war it was also used for high-profile prisoners (relatives of politicians, and such) whom the Germans thought would make good hostages, if the war were to go badly.

The Germans considered the castle to be inescapable; it had tall, sheer walls, was located on a steep hill, and the guards outnumbered the prisoners.  But putting so many escape-prone officers in one place had predictable effects.  Many attempted escape, a few were shot, several managed to get away, and some even managed to make it all the way home.

It's a true account, but it reads like something out of a Hollywood movie.

Everyone here knows how to pick locks, how to make disguises.  Hacksaws are made from old razor blades, radios are constructed from smuggled parts.  If a civilian contractor enters the prison, and removes his coat for any reason, he'll leave without it.  If he leaves his truck unattended on the grounds, he might find it missing a wheel bolt, or two.

The different contingents--English, French, Polish--need to appoint Escape Officers, who coordinate activities so that different groups don't interfere with each other in their attempts.

There are diversions, tunnels, feigned illnesses, hidden compartments, secret passageways.  The Germans, to educate themselves, create an "Escape Museum", containing artifacts of past attempts.

Guards' movements are carefully monitored and recorded; one escape is coordinated with split-second timing.

One prisoner resembles a high-ranking German closely, and is disguised to look like him, so he can deflect suspicion during his escape attempt.  He even seems about to pull it off, until the original appears.

A section of an attic is walled off to make a workshop.  A glider is constructed, with the object of a rooftop escape; but, the war ends before they get to test it.

This book is more concerned with the art of escape, rather than the war itself, but there are a few reminders.  A number of French Jews who were interred in the camp for a time have no motivation to escape; if they were to be recaptured, it would mean death.  The S.S. executes four hundred slaves in the nearby town.  The involvement of the Gestapo means torture.  The Wehrmacht, who operate the prison, seem comparatively civilized, but they'll still shoot at you if you run.

This is an entertaining read, and an easy recommendation.  "Escape from Colditz" is actually comprised of two earlier publications, "The Colditz Story" and "Men of Colditz"; the first is an account of the author's personal experiences, up until his escape; the second is a compilation of later events in the prison, which he learned of after the war.
Desine fata deum flecti sperare precando.