Joined: Sep. 2006
|Quote (Zachriel @ Jan. 12 2012,10:26)|
|We posted this above, but as we can't comment on Uncommon Descent, and as kairosfocus has asked us not to email him, we have decided to answer him here, then send the message to denizens at Uncommon Descent for forwarding. |
|kairosfocus: For instance, of course it is easy to go cat –> rat –> ran –> man –> mat, etc. But, once we add significant complexity — say about seven letters for English words, the gaps between words will be all but impossible to consistently bridge (the islands of function issue emerges . . . ). |
Interesting that he picks the same limit as Sean Pitman did in days of yore.
|Hi kairosfocus, |
Saw your very interesting comment on Uncommon Descent.
kairosfocus: For instance, of course it is easy to go cat –> rat –> ran –> man –> mat, etc. But, once we add significant complexity — say about seven letters for English words, the gaps between words will be all but impossible to consistently bridge (the islands of function issue emerges . . . ). And that has been pointed out over and over, just ignored.
What you are saying is that if we have a population of short words, and they are subjected to mutatation and recombination, and only those offspring that form perfectly spelled words are allowed to enter the next generation, that you will not see words of more than six or seven letters? This, because they are isolated on islands, and there is no way to cross between them?
Member AMF, Angelic Motive Force:
Pushing planets on celestial spheres — one epoch at a time.
Just to clarify, by mutation we mean a single random letter change, insertion or deletion. By recombination, we mean a random part of one sequence, in whole or in part, replacing a random part of another, in whole or in part. In any case, after each such change, if it doesn't form a perfectly spelled word, it is discarded.
So, you're saying that given the scenario of a population of small words that evolve through mutation and recombination, there is no pathway to words longer than seven letters. Let's try. We'll start our population with just a single-letter word:
Consider this lineage:
o, a, an, ant, cant
With another lineage that descends from the same ancestor like this:
o, or, our, out, pout
Our population now includes
ten nine species of word. By recombination the ou from pout replaces the a in cant for count.
Meanwhile, another lineage descends like this:
o, a, la, lea, lee, lees, less
By recombination of the ess from less and count, we have countess.
Finally, by recombination of es from less and countess, we have countesses. That's ten letters. The chance of that exact word occurring in a random sequence of letters is less than one in a hundred trillion.
There clearly is a single-step pathway to at least one longer word, countesses. Do you think this is an exception? Or are there many such pathways to longer words?
Now, you might argue that these paths were intelligently discovered, but your original claim was that such pathways didn't exist. If there are no such pathways, then intelligence or no, nonexistent pathways can't be found. Now, if you want to argue that *random* mutation and recombination would not be able to find this pathway in a reasonable period, then you should make that argument instead of saying there are no such pathways whatsoever. Is that your new claim?
Kairosfocus has refused to defend his position unless we first apologize for our comment above. Our experiment of directly addressing kairosfocus without the glare of public exposure has not been, um, entirely successful.
Per his request, we are not quoting kairosfocus' remarks.
Tard Acquisition and Repository Department