Psychodynamics cont...

From: primeradiant (
Date: Thu Oct 28 2004 - 12:09:29 MDT

Domini Fratres Magnificentissimi!

>I'm sorry, did you say `a prisoner of a rave ward'?

No, I wrote 'gravity well'. Perhaps you should have gone to Specsavers.

>associate of the Warsaw Orthodox
>Theological Seminary,

Of course. Apart from psychology I've a musicology degree. I'm an expert in the transcription of Byzantine & Russian Church Chant (neums). Takes all sorts, you know.

>SL4ers, meanwhile,
>are likely to view "psychodynamics" as more of a cult
>than a science, and will react with the characteristic
>hostility of people on the receiving end of an
>unrequested psychoanalytic diagnosis.

First, psychodynamics is not a 'cult'. Current advances in neurobiology, neuroscience, cognitive science, evolutionary and experimental psychology are now beginning to vindicate and render more precise, the intuitive concepts of early psychoanalysis (for starters, try On the Incompatibility of the Psychoanalytic Model with Human Neuroanatomy by Joseph Slap - URL: and proceed form there to Schore's Affect Regulation and the Origin of the Self (1994)). We've come a long, long way since Freud. Group psychodynamics has a rigorous quantitative dimension these days. Hasn't anyone heard of Bales' Interaction Process Analysis, Stryker's Symbolic Interaction Model, the structural-functional models of Parsons, the cybernetic models of Deutsch, Wasdell's Matrix Methodologies or any of the other methods developed at the Tavistock and Rice Institutes, or similar models employed by consultant psychologists working for NASA or other large organisations? Read Anzieu, Bion, Czander, Stein, Wasdell, Young or others on task-sentient boundaries and other subliminal structural/dynamic phenomena that inhibit the goals of task-oriented groups. There' a multiplicity of models, true, and we're no nearer synthesis than you are from Seed AI, but we left Oedipus in the cradle long ago.

Second, the research material of SIAI and the SL4 dialogues suggest that a collusive, cultic mentality does pervade the organisation (like most others). The self-designation 'SL4ers' certainly evokes more than a whiff of Jim Morrison's 'smell of the tribe'. Look how you all project your collective anxieties into Eliezer, forcing the role of prophet onto him. What happens if, through no fault of his, the 'thing' doesn't work? Read Bion on this one. And 'good heavens', what's all this about 'evangelization'? It's quite clear from the responses I've received so far (and from the website) that SI/SL4 has acute anxieties regarding its collective boundaries, yet at the same time proposes to create an entity that is supposed to 'speak for us all' ('collective volition' - whatever that is). Oops! Now that does give me the heebie-jeebies.

Third, I have not offered an 'unrequested psychoanalytic diagnosis', only a few general points regarding the psychodynamics of task-oriented groups which any group so utterly brilliant as to be working seriously on such a critical project as FAI is surely well aware of as a matter of course, right? Any complex, recursively self-enhancing entity created by human programmers is likely to develop some functional layers analogous to the human DU. If we are not aware of the morphogenetic origin of our own motivations, a group of 'imperfectly deceptive social organisms' may well end up inadvertently creating a perfectly deceptive antisocial machine. There are no safeguards. Without greater personal and interpersonal insight on the part of the task-oriented group, not Solomonoff, not Bayes, not Kolmogorov nor any combination of causal validity, external reference or shaper/anchor semantics can guarantee this will not happen. How can we create a generalized intelligence when we don't yet understand the EEA-constrained human instantiation of intelligence? Most of what passes for psychological knowledge on this site belongs in Psych101. Hasn't anyone read Husserl, Varela, Maturana, Haken, Pribram, Frölich, Searle, Edelman etc. etc. (none of whom are psychoanalysts by the way)? I write all this only because I want FAI to succeed. Non enim aliquid contentione quaerimus, sed res profundissimas modestissime nosse desideramus (Cassiodorus of Vivarium). But while I'm not Don Quixote, I've more than my fair share of the Cassandra Complex, as you can see. It's the occupational disease of psychohistorians.

By the way, social scientists been using Bayesian statistics for years (as you probably know). It's considered vieux chapeau now by APA and BPS statisticians but for my part, I rejoice mightily to see it revived in the world of AI.

Please tell me what you think of Petitot's comment - which IMHO has great relevance for current SI/SL4 thinking. Or must I translate it for you?

>I'd have thought it's closer to the dismissive scorn >appropriate to an
>unrequested Scientological come-hither while crossing the >campus. But maybe
>it's my years of listening in increasingly angry disbelief to >Lacanian
>drivel talking.

Really! Lacan is not psychoanalysis, nor psychoanalysis Lacan. He's no topologist either. And what has scientology to do with it? I suggest you broaden your psychoanalytic reading somewhat. I study AI material - avidly. Get to grips with mine. Oh, and mind the semantics - drivel doesn't talk, people talk drivel. If we can't write English, how will we ever write error-tolerant FAI source code?

Ok, so I seem to be coming from another planet. But I'm not. I'm from the same one the SI lives on. And I'm deeply interested in AI research, which promises much for psychological research, not to mention the general benefit and evolutionary future of this species. If I must qualify for entrance into the sanctum sanctorum of SI, the SI must qualify itself to me as an institution if I am ever going to support it and write favorably about it in a paper on AI psychology I'm working on at the moment (hopefully a more rigorous update of Mazlish's Fourth Discontinuity - which you all know about, right?). It's one of two. The other's on psychological factors in deep-space missions. It's all part of the generalized psychohistorical matrix.

Brilliant hackers come these days in packs of 50 - just add water. One sufficiently stable, wise, metacomplex and capable of programming Singularity-level FAI? I haven't met one yet.

Oh yes, and physics decidedly precedes maths. Somatically-encoded, hyletic experience necessarily precedes symbolification, as Husserl and others have shown. Human symbolic systems can be Gödelized , but not the invariant properties of eidetic data flow.

And my final Delenda est Carthago: the human mind is not Turing-computable.

Live Long and Prosper!

Paul Ziolo

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:49 MDT