Quake3World

We begin with an agent. Utility maximization is her ideal, goal development and implementation as the furthering of that ideal. She discounts expected future mortality from her current utility level.

As framed, the agent finds it in her interest to reduce existential risk (ER) exposure. Implicitly, her goal is an indefinite lifespan as confirmed by her best estimate of near-term future mortality at any one time (a limited metric). For continuous ER reduction over an infinite time horizon, a geometric series of conditioned probabilities exists which sum to 0< p <1 (boundary exclusive) for an agent at some time t. She cannot forecast over an infinite time horizon, but she is privy to near-term expected ER rate reductions that she manipulates to produce a limiting p-value < 0.5 of death subject to her continual maintenance across time.

In our previous thread, we discussed goals as arbitrarily derived, which may not be a precise specification for this universe. Agents not as the one above, who do not optimize towards indefinite life or do so in some persistently sub-optimal manner, will cease to perpetuate themselves in the long-run. Reproduction as means of agent-class perpetuation without cognitive inheritance will render each generation of new cognitions insufficiently equipped with respect to their ER-reducing counterparts. The reproductive class may too extinguish either of their own right, or through game-theoretic and/or information asymmetric competition with more developed ER-reducing agents. Therefore, it's conjectured here that goals will arise to further the ideal of utility maximization incorporating an existential risk component. Within the (rather large and poorly defined) space of compatible goals to that end, we may see some arbitrariness and randomization among agents as to the goals they individually pursue.

edit: typo

When you begin fixing conditional environments, "arbitrary" and "random" become less and less probable. However, no answer you find will ever mean anything concrete. Shit happens yuh know.

It's conjecture, food for thought.

Physical possibility is not addressed here.*

Cliff Notes: Immortality is a logical possibility given ever-present non-zero agent-affective existential risks (other priors implicit/ignored).

* As a preface to any treatment of physical possibility I will say this, we ought take our cohered science as both flawed and spatio-temporally limited in it's predictive power. Engaging in unbounded physical conjecture is a more highly conditioned far less likely projection of supposed true expected outcomes for the entire universe over an infinite time horizon. (Less ambitious predictions are less foolhardy, but miss the point of post #1, a thought experiment rather than a forecast.)

Seriously, when we try to minimize existential risk, we go against our very nature. We create more problems than we avoid. Rather than minimizing it, we should alter our approach to devising method that give advantage to the risk. I don't mean the simplistic approach of people making preventative laws. I mean teaching people to "roll with the punches". If microcosms are observed with attention, it is obvious this approach is the way of the Universe.

Moron.

I concur

this thread contains nuggets of turdles .. slow green turdles with hard shells

and beetles in boxes

To touch on means in the abstract, let us make more concrete the universe as described in post #1, from here on referred to as (1).

This universe is unbound in all consequential dimensions. The agent from (1), we'll call her Sheila, develops and employs a perceptual framework based on her cohered local science. Her embryonic perceptual framework (pattern recognition) was imbued to her through an unguided naturally selective evolutionary process. She has since self-augmented to suit a broader scope of stimuli and pursuits.

Sheila is more learned of her local environment, and therefore less susceptible to local shocks because less is exogenous (unaccounted) to her perceptional and predictive purview. This lays in contrast to increasingly non-local, increasingly less well-known events which may otherwise impact her. While within her neighborhood, she exists as an adaptive intelligent process with a high degree of operating self-knowledge. Still, perfect self-knowledge eludes her through logical impossibility and therefore she is exposed to some due amount of non-zero intra-systemic existential risk (ER).

Sheila is a cognition with interests, first imbued by evolution, and extended through self-manipulation. Curiosity is a class encompassing interests, it is a means of survival and an end to survival (a component in her utility function). Curiosity is the mechanism by which she ultimately increases the scope and predictive power of her cohered science, and consequently her perceptual awareness. Her newly broadened applied science then allows her to construct novel ER-reducing means and methods, and/or to further other ends relevant to curiosity (feedback).

Sheila must grow to encompass more of her universe, materials are not infinitely divisible. She must employ prudent cognitive redundancies to maintain survival of her cognition and knowledge, supposing the prospect of unexpected system-local cognitive destruction. Her science must broaden to encompass and account for more in an ever larger locality. She must remain cognitively competitive, either with respect to her game-theoretic peers, or in response to new and/or unexpected ER exposures.

This is Sheila.

Thank you Sergei.

http://www.google.ca/search?hl=en&q=Exi ... arch&meta=