On Mon, Dec 10, 2012 at 7:25 PM, Russell Wallace
<[email protected]> wrote:
> (first draft of this message was sent partway through by accident)
>
> On Sun, Dec 9, 2012 at 2:32 PM, John G. Rose <[email protected]> wrote:
>> I don't know if anyone knows anywhere near what the minimal K-complexity is
>> for running general intelligence.
>
> I know what it is. It's the length of the program 'start with the Big
> Bang and simulate a universe under standard physics until intelligent
> life evolves'.  Estimated K-complexity no more than a few thousand
> bits.
>
> You will observe that this is not remotely useful, which is the point:
> K- complexity is about the shortest program given infinite computing
> power, which is a very different thing from the most efficiently
> constructible program given limited computing power.

Yes, you are right. Thank you for bringing that to my attention.

Simulating the universe requires a program of only a few hundred bits
on a computer with 10^120 bits of memory in 10^120 steps. (See
http://arxiv.org/abs/quant-ph/0110141 ). We could do even better by
Levin search for intelligent life: simulating a multiverse with all
possible laws of physics using a program size of just a few bits and
10^240 steps.

The real problem, of course, is my estimate for the complexity of the
human genome that is based on each generation adding about 1 bit. To
execute that program you need a planet sized molecular quantum
computer using 165 petawatts of solar power for 3 billion years.
(Quantum because modeling chemistry by solving Schrodinger's equation
is otherwise exponential).

So I should be using the actual size of the genome (6 x 10^9 bits)
with some reasonable compression algorithm. DNA doesn't compress very
well. A quick test with 7zip -mx on chr22.fa from the hg19 reference
genome compresses to 68% of the uncompressed 2 bits per base pair.
There are other compressors that would probably do a little better. So
I think 4 x 10^9 bits is a fair estimate.

That's about 100 million lines of code at a cost of $10 billion. Not
that it matters in the least. If it were $10 trillion it would still
be overwhelmed by the cost of hardware and knowledge collection.



-- Matt Mahoney, [email protected]


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to