[extropy-chat] I'm back.

[Dan Clemmensen]

Robin Hanson wrote:

> On 11/12/2003, Dan Clemmensen wrote:
>
>>    In 1996l, on this list, I asserted that the singularity would 
>> occur prior to 2006. I had no viable reason for
>> that asertion. I still have no viable reason for that assertion.
>
>
> To be clear, are you still making that assertion?

Yes, but I'm feeling pretty shakey about it. I also asserted to Damien 
(who quoted me in the original Auatralian version of "The Spike") That 
any such assertion is invalid more than about a year in advance, because 
the real evidence is tenuous. If you recall our discussins back in 1996, 
my premise is that the singularity will be a phase change in a 
supercritical technological substrate, analogous to the instantaneous 
freezing of supercooled water when it is shocked.

This hypothesis here is that an SI will include a very large 
co-operating computer system running newly-developed software that can 
improve itswelf. With sufficient available computer resources, such a 
system should very rapidly improve its ability to improve itself.

More available computing power translates to a higher liklihood that the 
"net" is supercritical.

Since 1996, we've seen the an increase in available computing power of 
perhaps 10,000 to 100,000, depending on how you count. We've also seen 
the emergence of Beowulf clusters, SETI at home, grid computing, and 
Google. All of these techniques add to the toolbox and open new avenues 
that might let soem clever programmer make the breakthrough.

The difference between my hypothesis and Elizier's "friendly AI" is that 
my hypothesis does not specify the nature of the self-improving system. 
His friendly AI is an instance of a class, but any member of the class 
results in the SI.

My hypothesis is that the mroe "supercritical" the system is, the 
simpler the seed can be. Eventually, the power on the "net" will be so 
large that a trivial assembly of programs and people will suffice.

Will this happen by 2006? Well, "I still have no viable justification 
for this assertion."

Sorry for the weasel wordsa. As you see, I'm mostly holding this 
position for old time's sake.