Approaches to Quantum Gravity

String theory, holography and Quantum Gravity 197
of chemistry and biology we know and love, but it would surely be more satisfying
to derive it from a less restrictive assumption.
We do not even have a precise mathematical definition of our much less restrictive
observerphilic principle. However, if we make some assumptions, we can see
some of its consequences. Assume that the observerphilic part of the universe will
eventually evolve to be an FRW space-time. The p = ρ universe is infinite. Obviously,
in entropic terms, it is preferable for the low entropy, observerphilic part of
the universe to involve as few degrees of freedom of the full system as possible.
In particular, a finite number is infinitely more probable than an infinite number.
This principle thus predicts that the cosmology of the observerphilic part of the
universe should have causal diamonds with bounded area. There are two ways to
achieve this: the observerphilic part of the universe could end in a Big Crunch, or
asymptote to dS space. The life-time of observers in an asymptotically dS space of
given horizon size, is exponentially longer than in a Big Crunch space-time with
the same size maximal causal diamond. 11 Thus, if one looks for an observer in a
p = ρ universe sprinkled with observerphilic regions of various sizes and types,
one is more likely to find it in an asymptotically dS region. Note that, unlike the
anthropic principle, the observerphilic principle lends itself to simple calculations
of probabilities, and makes no assumptions about particular biological structures,
or the nature of low energy particle physics, except that it is well described by
quantum field theory. It predicts that an observerphile will want to search for the
objects of his or her affection in an asymptotic de Sitter universe, with the maximal
value of the cosmological constant consistent with whatever version of observers
he/she wants to insist on.
It is amusing that the necessity that a locally FRW region should be asymptotically
dS can also be derived directly from General Relativity [19; 20; 21], by
assuming that there is, for all times, a consistent interface between a particle horizon
sized bubble of normal universe, and the p = ρ background. In all cases
where the particle horizon expands indefinitely, the parallel components of the
Israel junction condition show that we can only make an interface between the
two FRW systems if the coordinate volume of the less stiff fluid shrinks with time.
By contrast, we can match the future cosmological horizon of a single observer
in an asymptotically dS space, to a marginally trapped surface embedded in the
p = ρ background. Black holes in the p = ρ background cannot decay, because
the background space-time already saturates the maximum entropy bound. I view
this as evidence that a complete quantum theory of a p = ρ background sprinkled
11 In an asymptotically dS space, observers are destroyed by thermal nucleation of bursts of radiation, or black
holes, at their position. The probability of such processes vanishes exponentially with the dS radius.