[ExI] Evolution

Ben ben at zaiboc.net
Sat May 23 09:19:08 UTC 2020

On 23/05/2020 00:56, bill w wrote:
> Completely innocent and ignorant question:  how far are we from 
> examining a gene and predicting what it will do?  If we could do that, 
> we could design genes, splice them in and do our own evolution, eh

Bill, I'd say we are very, very far from this, in general, when faced 
with an unfamiliar gene.

When faced with a familiar one, where we already know what it will do 
under specific conditions, we can probably tell that changes to it will 
make it dysfunctional.

One of the problems is that genes are not as simple as they are usually 
made out to be. A 'gene' is just a sequence of DNA that codes for a 
protein, but that's just the beginning of things, and is not even 
strictly true in many cases. A gene can code for a fragment of a 
protein. Or a fragment of a number of different proteins. Or the 
precursor to one or several proteins, or slightly different versions of 
the same protein, depending on how its transcripted RNA is cut up into 
smaller bits, and recombined with some of these bits, or bits from other 

Then, the end-product of translating this RNA into a protein often goes 
on to do different jobs in different places under different circumstances.

Finally (or probably not finally, there may well be other mechanisms at 
work that I'm not aware of), the gene is subject to control over if and 
when it's expressed at all, by other stretches of DNA (usually 
'non-coding' DNA), and chemical groups that sit on the DNA strand (this 
is what's known as epigenetics), which are themselves gene products, 
which are subject to control... And so-on. There is even evidence that 
the same stretch of DNA can code for more than one thing, with what you 
get depending on where you start (frame-shifting), a little bit like 
this string of letters representing different sentences, depending on 
where you start, if you make the sentences by selecting every third 
letter: AATRBWEIODGECDGAOGRGS. Genetics is full of tricks like this.

It's basically a massive spaghetti program, written in a language we 
only partially understand. Predicting what one gene will do is like 
predicting what will happen if you change one variable in a monolithic, 
multi-million-line program written in BASIC, with heavy reliance on GoTo 
statements. In fact, I'd say it's worse than this.

It's probably no more possible to tell what a single unknown gene will 
do that it is to tell what a particular unknown pattern of cells will do 
in Conway's Game of Life. The only way to find out is to run the program 
and see what happens. So we'd have to have a massive simulation of all 
the biochemistry of an entire organism before we could tell what a 
single gene does, and we are a very long way from that.

This is why no-one has made more than trivial progress with genetic 
engineering to produce novel features. People talk excitedly about using 
CRISPR to give people things like an extra thumb, or extend our vision 
into the ultraviolet. We have no  clue how to do things like this, and 
we would be fumbling in the dark in a room full of mantraps to even try it.

I very much doubt that genetic engineering is the way to enhance 
ourselves in significant ways. I hope I'm wrong (because it would mean 
that we are, collectively, MUCH cleverer than I suspect we are), but I 
have severe doubts.

Ben Zaiboc

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20200523/0112d912/attachment.htm>

More information about the extropy-chat mailing list