[ExI] list test

Anders Sandberg anders at aleph.se
Sun Jul 28 06:33:59 UTC 2013


On 2013-07-28 00:34, BillK wrote:
> On Thu, Jul 25, 2013 at 10:34 PM, Andrew Mckee wrote:
>> Hi there, looks like the list just went dead a few days back, so anyone else
>> out there still?
>>
>
> That's what happens when Spike goes on holiday.........

And no doubt when many of the rest of us are away too.

Email lists need a critical mass: if you see an email you might respond 
with some probability p, but without email you are unlikely to remember 
the list (post probability q << p). So the number of posts next day from 
N people will be
posts = N[p(1-(1-p)^N) + q(1-p)^N]

If this is larger than one, then the day after that there will be more 
posts, and so on, and the list shifts into high activity mode (I ignore 
the ceiling effect, where too high post rates makes people post less 
since they cannot keep up; feel free to extend the model). This 
transition happens when N becomes large enough (about N>1/p): you need 
enough posters for a given p to maintain reliable activity. If N and p 
decreases (holidays) the list can shift to a low state. q is what 
ensures there will be occasional revivals when things have fallen 
silent: Andrew's post is a fine example of a q-type post.

Another social list I am on has a far lower q, interestingly enough. It 
was completely silent for several weeks, until finally somebody asked 
Andrew's question, at which it jumped to its normal high activity mode. 
We seem to have a large N but even us actives have a somewhat low p, 
while the other list has a smaller N but higher p.

-- 
Anders Sandberg
Future of Humanity Institute
Oxford University



More information about the extropy-chat mailing list