IBRIColloquium, 22 Jan 1991
Dr.Robert C. Newman
BiblicalTheological Seminary
COMPUTERSIMULATIONS OF EVOLUTION
Introduction
Nota literature search
Notcovering origin of life question
tho2 programs on diskette are self-reproducing automata
REPRO- Langton's automaton in my JASA Spr 88 paper
BYL- Byl's in his JASA Spr 89 paper
Notdealing with competition & spread of varieties
gooddeal has been done on ecology/population genetics
Rather,a description & investigation of three programs that relate to the mechanism
--one devised by self
these3 programs also on diskette available from IBRI for $5
Program BIOMORPH
Describe:
program,slightly simplified from Dawkins, for building "organisms" fromgenetic information, selecting among mutants
geneis sequence of eight small integers
generates"tree" controlling branch length, angles,
#of levels of branching, with mirror symmetry
givenoriginal gene, program constructs all "one-step" mutations, displayson screen
operatorselects which mutant to succeed parent
Lessonsfrom BIOMORPH:
mutationoperates on DNA
selectionoperates on developed form, not on DNA
seethat:
identicalforms can conceal diff genetics
leavingroom for neutral mutation
Program SHAKES
Describe:
Dawkinsseeking to circumvent "monkeys typing Shakespeare" problem ofenormous times involved
choosetarget sentence/phrase
startwith gibberish of same length
Dawkinsgets convergence in typically 40-70 generations
Dawkins'version:
Notdescribed in detail, so can't tell how he generated mutants, how many mutationsper generation
Myversion:
Onemutant each generation, compared w/ parent
Betterof mutant/parent survives
Iget much slower convergence, taking over 1000 generations
Lessonsfrom SHAKES:
showsthat a "rachet mechanism" does work
importantreason why many convinced evolution must
becorrect
butthis is "guided evolution,"
whichis considerably more efficient than even artificial selection,
tosay nothing of natural selection!
doesnot solve time question
whichversion is more realistic?
mutationrate in eukaryotes is 10-8 per replication
bothignore time involved for mutant to take over
population!
myversion suggests a problem
formutating into complex or optimal structures:
lastpieces of puzzle are highly constrained
Program MUNSEL
Describe:
simulatemutation & natural selection by analogy with human language
letterstring is both gene and organism
mutationis random change in content and/or length
selectionis "naturalized" by requiring that each
groupingin string be an English word
currentversion has operator do selecting,
butcomparing with a spell-checker would be more objective
generateswords of 1-4 letters rather easily
relativefrequency of space character (and nature of selection) tends to keep wordsshort
littlesuccess in getting intelligibility in 100s of steps
Lessonsfrom MUNSEL:
complexorganisms involve hierarchies of structure
somewhatlike intelligible writing
letters> words > phrases/sentences > paragraphs
mutationonly works at lowest level
nucleotides<=> letters
non-selectedmutation => gibberish
neutralmutations spread only by random walk
functionalisolation seen here (as in terrain analogy)
manycombinations cannot be reached by single mutations from acceptable smallergroups
foreach level of hierarchy?
canyou really get there from here?
complexorgans/organisms