Sientifik theorys and methods model'd on natural selektion.

To index of simpler spelt pages.
To home page.

( Kapital-i, in 'I, myself', now spels Il as in isle or aisle.
Leter y spels ryd for reed or read and partys for parties.
Leter w spels swn for soon. )

Links to sektions:

Manfred Eigen's theory of inorganik natural selektion.

To top.

The most basik diferens betwyn lIf forms on erth is not betwyn plants and animals but betwyn sels with or without a nukleus. Ther is a fundamental diferens of emfasis in the ways wich the tw kinds evolv. Sels without a nukleus, such as bakteria, evolv partly by random mutations of jyns, som of wich ar naturaly selekted over others. This is the traditional konseption of evolution sins Darwin. It is an important kaus for bakteria, bekaus they kan reprodus and multiply so rapidly, that smal mutations to ther heredity swn tak efekt over ther houl jyn pwl.

The other important faktor in the evolution of bakteria is kal'd DNA rekombination or global trading of jyns. This is rather as if humans had a natural kapacity for jenetik enjinyring, being abl to splIs bits of ych others' jenetik inheritans.

How-ever, kreaturs with a sel nukleus wer orijinaly separat kreaturs that hav permanently kombin'd. Lynn Margulis sujested that this inkorporated symbiosis, or 'symbio-jenesis', is the main way hIer organisms hav evolv'd. The klaim is that lIf achyv'd its global empIr not by kombat but by networking.

It has long byn question'd that lIf aros simply out of the natural selektion of chans jenetik variations. LIkwIs, the natural selektion from a menu of randomly kombin'd kemikal swps has sym'd wildly improbabl. This is despIt Miller's famus experiment that produs'd, in a flask, som basik konstituents of lIf, by elektrolysis, simulating lItning on a fw kemikals present on erth in its primeval stait.

Manfred Eigen sujested that lIf-lIk, but not living, stably reprodusing systems of kemikals evolv'd thru natural selektion, befor evolving into lIf itself. The metabolism of living sels depends on katalysts, mainly enzyms. Katalytik re-aktions form komplex net-works, inkluding klos'd lwps or katalytik sIkls. Enzyms produs'd in uon sIkl akt as katalysts in a further sIkl, a so-kal'd hyper-sIkl.

Eigen propos'd this mekanism for self-organising kemikal systems, with inorganik hyper-sIkls in kompetition for natural selektion. Thys systems ar far from equilibrium, that is 'fed' by an enerjy flow. This may fors instabilitys in the system. But lIk the Chinese word for 'krisis' it kan also myn 'oportunity' for nw development, wich may be amplify'd thru positiv fyd-bak lwps.

Environmental selektion in memory, konektionist komputation and nural net-works.

To top.

Western thot has depended mostly on analysis but ther ar revivals in the holistik aproch. Ther was Kurt Lewin's gestalt sykolojy as an alternativ interpretation to that ofer'd by behaviorism. Behaviorism, itself, has byn liken'd to a Darwinism of the environment. ( Talcott Parsons' The Structure of Social Action ).

The brain has byn kompar'd to a komputer. Komputers or thinking macyns hav byn bais'd on linear thinking, starting with a fw basik prinsipls and then folowing thys to ther lojikal konklusions.

In the 1980's, a diferent kind of komputer was bekoming fashonabl as a model for the working of the brain, syn as a nural net-work. Konektionist komputation is a desendant of the nIntynth sentury asociationist sykolojy, wich koms from British filosofy of empirisism, rather than the rationalist filosofy wich syks to deriv nolej from first prinsipls.

I kan best deskrib the diferens betwyn the tw aproches, from my personal experiens as a child. Wen lerning to ryd, I didnt much bother to konsult a diktionary. It wil defin the myning of a word yu dont know. But that is a bit of an interuption of an adventur story.
For that mater, wen adults yus an obskur word, it wud be a very skolarly child, hw went off to konsult a diktionary. Uon wud lws the rest of the konversation and be wors off, than staying put.

Similarly, this child stay'd put at his bwk. Funily enuf, I lern'd the myning of words I didnt no, tho I had never found out ther myning from a diktionary. I belyv this was bekaus I gradualy pik'd up the myning of a stranj word, from koming akros it several tIms in bwks and bilding-up a sens of wat it ment, from the slItly diferent kontexts in wich it was yus'd, without ever having byn aktualy told its myning.

To som extent, I was selekting the bwks I red. But the chanses, of wich bwks I hapen'd upon, wer naturaly selekting for me the mynings of words in the partikular environment of bwks, wer I found them. Thys chanses -- mor or les -- re-infors sertain kontexts of word-yusaj and diskriminat against inaplikabl yusajes.

Pasing from uon bwk to an other, the ryder inherits an akumulation of slitly varying mynings to any given word. Thys mynings wil vary about a norm, with som tolerans for deviation from that most usual myning of the word in question. This tolerans of variation alows words to adapt ther mynings, apropriatly to a chanjing sosiety.

In a living languaj, no hard and fast definitions ar laid down by som text with a monopoly of authority. Insted, al the authors, uon ryds, ar kompyting authoritys, by wich a konsensus of myning is rych'd. Even diktionarys kompyt as gId-lIns.
Just as sher'd spych is a demokratik yusaj, so is the sher'd authority of many authors.

Words deriv a spesys-spesifik myning from ther kontexts in a semantik ekolojy of words. As languaj evolvs with a chanjing sosiety, the mynings of words may be adapted to nw situations. It has byn klaim'd that the evolution of languaj and of twls ar the tw traits of human sosiety most akin to biotik evolution.

As a child, lerning a nw word was a bit lIk peging down a tent. Ych tIm uon kaim akros it, was lIk a peg in uon's memory, until uon had its ful myning sekurly ankor'd in uon's mind. Uon may mov the pegs about a bit, in relation to ych other, to straiten the tent's konstruktion. LIkwIs, uon's idea of a nw word may shift about a bit, as uon bilds up a nolej of its varying yusaj, into a setl'd patern.

Margaret Boden kompars the mekaniks of konektionist komputers to a klas of children. ( 'Komputers' yus'd to be the naim for pepl, hw kud do extra-ordinary kalkulations in ther hed, without even noing how. ) Som of the children no bits of information but not the houl 'story': they ar lIk input units. The output units ar the children hw anouns 'the story so far'. Other children ar the 'hiden units' that neither hav any input or output to ofer but mediat betwyn thos that hav.

This so-kal'd 'paralel distributed prosesing' is lIk a gesing gaim, in wich som children hav klus, wich they pas on to others. Geses may be re-infors'd by repetition of the saim klus or thru asociations. Other geses ar dismis'd as they depart from the patern, bilding-up. A konsensus or equilibrium is aproch'd on the basis of probability of wat the objekt, to be ges'd, is.

An objekt is not defin'd by explisit ruls, lyding yu to konklud exampls. Rather, glimpses or bits of nolej, exemplifying an objekt, ar amas'd, til regularitys ar implisitly lern'd, in a bild-up of sensory asociations. Thus, the natur of uon's environment akts as a selektor on the ruls uon taks in.

It is lIk the diferens betwen formal akademik edukation and 'lerning by doing'. And, indyd, the traditional klas-rwm has a poverty of environment to experiens, bekaus uon is, or was, not ment to lern from uon's suroundings but from a sylabus. In this respekt, the klas-rwm restrikts lerning rather than promots it. A klasik kritisism of formal edukation is that it is irelevant or not adaptiv to lerning the ruls of survival in the ryl world.

Mathematicians tend to sy fylds of study in terms of 'fylds', in the sens of abstrakt spases. Thys may be visualis'd in thry dimensions, say, as a memory 'land-skaip', tho the aktual mathematiks, as yus'd by John Hopfield, jeneralises the idea to multi-dimensions of memory asociations, perhaps simulating the nural net-works of the brain.

TrIing to remember a naim is lIkened to seting a marbl roling on a hily terain. In uon of the valys of memory is the korekt naim but ther ar klosly asociated, over-riding memory basins that the marbl insists on roling into, insted. For exampl, the Skotish inventor of television was not kal'd Yogi Bear but it was som-thing lIk that. ( Logie Baird, in fakt. )

LIk a skratch on an old rekord, a dyper memory kuts akros a klosly related memory's impresion, and kauses the nydl of rekolektion to slip in its traks and go round and round the rong way, til uon finds a way of puting it bak on kors to wer uon wants to be.

The memory land-skaip, that experiens bilds from lIf's chanses, may be a fairly reliabl map of the past. But a konfusion of traks on the mind, left by the senses' past trafik, kan play us fals. The point is that the natur of our experiens bilds a memory land-skaip that selekts our memorys for us. Memory bekoms subjekt to our experiens's environmental selektion

The world, as a houl, dos not stamp a formaly korekt impresion, lIk a rekord, on our minds. Rather, it is a huj mix-up that we hav to mor or les mak sens of, as far as posibl, and with doutful sukses, jujing by the kurent stait of the planet.

Evolutionary reform of English speling.

To top.

A ded languaj is uon in wich mynings ar fix'd, bekaus aproximating to a form of sosiety that no longer exists. The sukses of English depends on the extent of its demokratik evolution. English is a hybrid of languajes that welkoms nw worlds of experiens by fryly adopting the words of other languajes. Kritiks of English as a posibl world languaj protest it isnt a languaj at al, bekaus it dosnt konform to ther standard of a klasikal languaj, that is a ded languaj, with no nw input.

Indyd, the idea of a world languaj is the miskonseption that wud impos som rijid form on the houl world. WI English most lwks lIk a potential world languaj is presisly bekaus it has abandon'd the idea of som fix'd linguistik form and is evolving out of itself, al the tIm.

Wer-as, the failur of English is presisly in its failur to evolv a mor rational speling that wud enfranchis the twenty per sent or mor of its spykers, hw ar funktionaly iliterat. This failur is du to the falasy of 'korekt' speling yusaj, the dogmatik insistens on som monopoly of authority, such as Dr Johnson's diktionary. In fakt, diferent English spyking kuntrys hav typikal speling variations from ych other. The Oxford English Diktionary has adopted both British and Amerikan spelings.

The problem of English speling reform is tw-fold. Partly it depends on tolerating frydom of speling, so that any-uon not speling the saim way as Dr Johnson or Teddy Rooseveldt is not lwk'd-down on. Partly, it depends on the nyd for an edukation in how to spel English as it sounds, with a tolerans of aksents. We dont insist that Australians shud drop ther spesys of Cockney twang. We shudnt insist that al English spyking pepl shud konform to uon speling 'aksent'. But to be abl to spyk in any aksent depends on an edukation in basik English fonetiks. BesIds the orthografik pedants, the other problem, to saving English literasy for al English spykers, wud be the fonetiks pedants.

Rational English speling, for jeneral literasy, nyds not only to ensur a klos aproximation to uon leter uon sound. It also nyds to ration the number of leters in the reform'd alfabet to about the saim number as the existing English-Roman alfabet of 26 leters. ( This is delt with, on an other of my web pajes. )

Supos edukators achyv'd a short rational English alfabet tolerating aksented variations in speling. This wud not lyd to anarky any mor than Dickens' novels, with ther fonetik transkriptions of aksents ar anarkik. Mor-over, pepl hw wanted to be understud, as wIdly as posibl wud spel in the most universal English aksent, 'mid-atlantik' or wat-ever.
A sort of natural selektion wud operat, wer rItings, in the most standard aksents, wud be the most wIdly understud and therfor the most wIdly red.

English kud evolv in its speling, as in its gramar, wich aros from a 'natural selektion' of the simplest syntax, betwyn ordinary pepl of divers bak-grounds, with limited understanding in komon. This kontrasts with the hIly inflekted languajes elaborated by skolars, or with monitoring by a languaj akademy, justifying its existens by imposing a supos'd klasikal exelens, purjing popular introduktions of forein words.

The sektion below, on evolv'd komputor programs, has a leson for languaj reform. 'Korektnes' of programs kan only be achyv'd by total kontrol over its rIting by the programer, hw nos every lojikal step of the way. But for komplex programing problems, this 'korekt' aproch syses to be praktikal. The ryson is analjus to wI the 'komand ekonomy' has to giv way to the market ekonomy. Ekonomik relationships ar tu komplex for a totaly top-down manajment. Diktatorial kontrol has to be relinquish'd for individual frydom of initiativ.

Lee Smolin's theory of kosmolojikal natural selektion.

To top.

Klasikal fysiks wud sujest that blak hols draw mater into infinitly dens points or 'singularitys'. But Smolin firstly asums that the modern ruls of quantum fluktuations wud over-rId this notion. Blak hols gravitationaly atrakt mater, beyond a bound of no return. But the mathematiks of this proses wer study'd in revers, as akin to the explosiv orijin of the univers, in a 'big bang'.

It is uon step further to Smolin's sekond asumption that quantum unsertaintys mIt bouns bak mater from a blak hol to explod a nw univers. Stephen Hawking's 'Black Holes and Baby Universes' is the tItl esay of a popular bwk, he rowt after 'A Brief History of Time.'

Hawking koments on the wel nown paradox of Shrodinger's Kat. He says the paradox depends on asuming the viw-point of klasikal fysiks, wer the prinsipls of quantum fysiks aplI. The klasikal notion is that an objekt has a singl definit history, such as is the kat in the box alIv or ded? But the houl point of quantum mekaniks is a diferent viw of reality in wich an objekt has al posibl historys. Probabilitys of very slItly difering historys, in most kaises, kansel out. But, in sertain kaises, the probabilitys of neiboring historys re-infors ych other; uon such being observ'd as the history of the objekt.
Schrodinger's kat has tw historys re-infors'd: uon as alIv, the other as ded.

Our univers may be konsider'd as only uon of many universes, or a 'multivers'.
To kontinu the analojy, of baby universes, the greit majority of universes ar lIkly to be stil-born. This is asum'd bekaus the only ryson this univers kan exist depends on an inkredibly delikat ajustment of fysikal konditions. ( Uon of Paul Davies' popular fysiks bwks, 'The Runaway Universe', is about this. )
Therfor, most universes born out of blak hols ar lIkly to get litl further than the sub-atomik skail of initial quantum fluktuations. And most of the remainder ar lIkly to be larjly unstruktur'd and so inkapabl of myting the konditions that mak lIf posibl in this univers.

Smolin's third asumption is that the baby universes inherit the saim laws of fysiks from ther parent univers but for smal random chanjes in the fysikal konstants or parameters, such as thos that ar found to obtain in this univers.

A kind of multiversal memory land-scaip, or 'mind of God', as Stephen Hawking wud put it, is implisit in this asumption that a baby univers remembers, with hI probability of akurasy, the fysikal laws of its parent.
Smolin yuses the saim sort of mathematikal land-scaip for kosmolojikal natural selektion, as is yus'd for memory's selektion from environmental experiens ( mention'd in the abov sektion on memory ).

The kurent best theory of our univers, on wich most fysisists ar agry'd, is kal'd the standard model. This kan not explain every thing from first prinsipls. Ther ar about twenty parameters or arbitrary konstants, wich kan not be dedus'd, but ar mesur'd by experiment, and plug'd into the theory, to fil in the gaps in its rysoning.

The kosmolojikal theory of natural selektion wud only konflikt with a totaly unify'd theory of fysiks, that had no parameters. Einstein atempted a unify'd theory and he asum'd quantum theory was inkomplyt bekaus of its statistikal natur. Natural selektion is a statistikal theory, wich Smolin has adapted to kosmolojy, on the asumption that the multivers is not absolutly determin'd, but kontains uon or mor fry parameters.

After the fashon of Darwinian theory, wich taks many jenerations to much alter spesys, many jenerations of univers wud be nyded to mak big chanjes to ther initial konditions. HIly restrikted is the ranj of parameters, or konditions, for wich atomi nuklei, and therfor stars exist, and the blak hols that form from som of them. But such universes with many blak hols ar lIkly to hav many projeny. Given that this is a blak hol-rich univers, its existens is explain'd as having the konditions of a typikal univers.

Out of the twenty or so fysikal parameters, Smolin givs a simplify'd model of his theory with just tw parameters, say, proton mas and elektron mas. He tryts thys as lenth and bredth ko-ordinats, lIk the tw sIds of a hily fyld. The hIts of the hils ar proportional to the number of blak hols that any given kombination of valus for thys tw parameters is lIkly to produs. In turn, the number of projeny, from ych univers, is proportional to the hIt of the parametrik 'land-scaip', wer it is situated.

Lee Smolin's theory maks an analojy of fysiks parameters to biolojikal jyns. The spas of parameters ( wich wud be of a very hI dimension ) is kompar'd to the kolektion of al posibl sequenses of DNA.

The averaj number of universes produs'd by a univers with a partikular set of parameters is kompar'd to the averaj number of off-spring from kryturs with a partikular set of jyns. This 'fitnes' of kryturs depends on ther situation on a 'fitnes land-scaip' in an abstrakt spas of jyns. The rait of reproduktion is mesur'd for how strongly it varys with variations in possible jyn kombinations.
Smolin qualifys his analojy as only aproximating to the simplest or krudest of biolojikal senarios, a singl spesys evolving in a fix'd environment.

Evolving robots and programs.

To top.

A robotiks resercher's work was shown on tv. This apyr'd to konsist of a flok of litl metal toys moving around the flor. Som-uon thot this was a wayst of tIm and the funds shud be kut. But I had red Kevin Kelly's Out Of Control, so I nw wat this aparently childish resercher was about.

Robbie the Robot, in Forbidden Planet kud do humdrum things lIk produs a pIl of bwtleg vintaj, besIds al the out-of-this-world efekts. The old movys hav mayd us imajin that robots ar larj, if not human-sIz macyns that pionyr enjinyrs kan at lyst begin to mak do yusful jobs, lIk Hoover the hous. Ther ar robots, koming into komercial yus, that kan do rutyn tasks lIk klyn a hospital flor.

A mor resent aproch to robot development is to kopy how lIf has evolv'd, from the simplest to mor komplex organisms. As robots ar esentialy mobil komputers, this also involvs a radikly diferent way to kreat programs. Traditionaly, programs hav byn riten for a spesifik purpos, achyv'd by folowing a set of rijidly kontrol'd instruktions, that kan only work in a definit environment.

The alternativ is not to defin the environment for uon komplex macyn's operations, but let the environment defin lots of simpl macyns' operations. This is akin to natural selektion. Gradualy, mor komplex jobs kan be don, wen the simpl jobs hav byn lern'd. And, lIk insekts, many smal robots kan do big jobs.

The evolutionary aproch to robots and to programs kompars to the toy konstruktion gaim, 'Leggo'. As in objekt-oriented programing, many smal existing programs, lIk leggo briks, kan be yus'd to put together a program for any nw purpos, insted of having to rIt a houl nw program from skratch, to fulfil a given gol.

Also, problems may be tu komplex to solv ysily, or at al, by rational desIn. But they may yild to masiv random trials. For instans, a molekular form must be found for a drug nyded to nutralis a disys mekanism. By the evolutionary method, bilions of random molekuls ar trI'd til uon fits against its lok, the rest being wash'd away.

Karl Sims' 'equation-genes' ar smal lojikal units of a komputer languaj ( LISP ). Ych modul is an arithmetik komand, lIk ad, multiply, kosIn, etc. Lojikal units ar evolv'd and randomly flip'd, to kreat nw funktions, that the komputer drw into often stuning paterns. Sims selekted and bred variations of thys, by swoping branches of ther lojik trys, analjus to the sexual exchanj of jyns.

Dr Robert Smith, at the University of the West of England, yuses such jenetik algorithms, or asortments of koud, set the problem of test-flIing playns. They ar selekted and bred lIk 'animals' until they bekom spesys-spesifik to ther problematik nw 'environment'.

Evolv'd solutions involv a greit dyl of trial and eror. The kriterion for such komplex solutions is not that they ar eror-fry, the property of smal systems, but that they work or that they ar flexibl enuf to surviv the tests they hav byn set.

LIk the jenetik koud, itself, evolv'd komputer kouds kontain much redundant material. Uon ryches the solution, by giving up total kontrol of how uon arivs and without noing how uon got ther. God is belyv'd to hav given fry-wil to his kreations, akording to ther evolving komplexity. So, evolv'd problem-solving givs up komplyt kontrol of uon's lojikal kreations' development. This helps uon to be kreativ beyond uon's pre-konseptions.


Margaret Boden: The Creative Mind. 1992.
George Johnson: In The Palaces Of Memory. 1992.
Fritjof Capra: The Web Of Life. 1996.
Lee Smolin: The Life Of The Cosmos. 1997.
Stephen Hawking: Black Holes And Baby Universes and other essays. 1993.
Kevin Kelly: Out Of Control. The new biology of machines. 1994.
Sanjida O'Connell: 'Those magnificent genes.' ( The Guardian, 21 June 2001. )

Richard Lung.

To top.

To index of simpler spelt pages.
To home page.