I'm not sure anyone can retrieve the original wording at this point; the wiki article has pretty much everything known about the conjecture.
"originally formulated before 1992" means that it's not in LifeCA email archives (that group was started in 1992).
The "locally stable over a rectangle R" formulation in the wiki article came from a 1992 rewording by Alan Wechsler, for a list of fifteen Life-universe open problems analogous to David Hilbert's twenty-three:
(
EDIT: sixteen open problems and questions, actually -- I didn't notice the list starts with zero!)
From: Allan Wechsler
Date: Fri, 10 Apr 1992 11:01-0400
Subject: Missions
I think this is the first time that so many influential Life hackers
have had such fast access to a common forum. Wouldn't it be a good
occasion to restate some of the most outstanding problems? Here are a
few that spring to mind:
0. Provide a coherent, intuitive, rigorizable account of what
constitutes a Life "object".
1. Life is known to be universal. Write a compiler whose target is
Life.
2. Find a stable pattern that is its own only predecessor.
3. Prove the existence of oscillators of all positive integral periods.
4. Prove the existence of gliders of all velocities permitted by
2(dx+dy)<=dt.
5. Characterize the functions F(t) such that Life patterns exist whose
populations grow as O(F(t)). Provide explicit examples where possible.
6. Find the smallest area containing a pattern with no predecessor.
7. Find the minimal population for a pattern with no predecessor.
8. Provide bounds for the maximum period of an oscillator that fits at all
phases into an XxY rectangle.
9. Provide bounds for the maximum period of an oscillator that fits at
some phase into an XxY rectangle. In particular, consider X=1.
10. Demonstrate the existence of a pattern which possesses parents but
no grandparents.
11. Provide a statistical but rigorous account of the end state of
randomly initialized universes. Characterize "fires" and "quiet areas"
and provide probabilistic analyses of the prevalence of these two phases
with time.
12. Prove Schroeppel's "Cool Out" conjecture: if a configuration C is
locally stable over a rectangle R, then there exists a configuration C*
such that (a) C* is locally equal to C over R; and (b) C* is globally
stable.
13. For what oscillation regimes other than stability is an analog of
Schroeppel's Conjecture true?
14. Settle the question of how many distinct three-glider collisions
there are.
15. Does there exist a finite stable pattern which can defend itself
against any single glider?
Rich Schroeppel never objected to the limitation to rectangular areas, so it's at least a reasonable guess that the original conjecture was also intended to deal with rectangular areas, with a boundary region around them to be used for stabilization. And notice that in 1992 people seemed fairly confident that the conjecture would eventually be proven rather than disproven:
From: Allan Wechsler
Date: Mon, 14 Sep 1992 18:17-0400
Subject: Re: Infinite p2 with density 1/2
John Conway wrote: Date: Thu, 10 Sep 1992 09:45 EDT
From: John Conway
Thanks - it seems likely that any simple infinite still-life or flip-flop
can probably be bordered so as to become finite?
For the period-1 case this is Schroeppel's "Cool Out" conjecture. My
phrasing was: if you have any configuration C with a rectangle R that is
unchanged in the successor (that is, C is locally stable over R), then
you can build a configuration C' with finite support, that is equal to C
over R and is globally stable.
John Conway wrote: I hope at any rate that
you can finish this one off.
It certainly seems as if only a finite repertoire of fringe textures
would be necessary to prove the conjecture.
However, the claim that Conway's period-1 case is the same as Schroeppel's Coolout Conjecture doesn't entirely pan out. You can't make an infinite still life out of the counterexample that Schroeppel eventually found. So the Coolout Conjecture is apparently a (mistaken) claim about a larger set of patterns than just rectangular subsets of infinite still lifes.
