On 10/2/07, Jef Allbright <[EMAIL PROTECTED]> wrote: > On 10/2/07, Vladimir Nesov <[EMAIL PROTECTED]> wrote: > > On 10/2/07, Jef Allbright <[EMAIL PROTECTED]> wrote: > > > Argh! "Goal system" and "Friendliness" are roughly the same sort of > > > confusion. They are each modelable only within a ***specified***, > > > encompassing context. > > > > > > In more coherent, modelable terms, we express our evolving nature, > > > rather than strive for "goals." > > > > > > > Terminology. Note that I did talk about subproblems of 'goal system': > > 'goal content' (textual description, such as Eliezer's CV) and > > property of system itself to behave according to this 'goal content'. > > Word 'goal' is a functional description, it doesn't limit design > > choices. > > What do you mean by context here? Certainly goal content needs > > semantic grounding in system's knowledge. > > Fundamental systems theory. Any system can be effectively specified > only within a more encompassing context. Shades of Godel's theorem > considering the epistemological implications. So it's perfectly valid > to speak of goals within an effectively specified context, but it's > incoherent to speak of a supergoal of friendliness as if that > expression has a modelable referent. > > Goals, like free-will, are a property of the observer, not the observed. > > When I speak of context, I'm generally not talking semantics but > pragmatics; not meaning, but "what works"; not linguistics, but > systems.
In this case there is no distinction: semantics is a label for perception machinery; that machinery implements integration of behavior, experience and 'goal content'. Goalness of goals is in existing in form of separate subsystem prior to integration (learning). -- Vladimir Nesov mailto:[EMAIL PROTECTED] ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244&id_secret=49015315-f3791f
