[ExI] Fail to the chief?

[Anders Sandberg]

Testing out systems of policymaking and governance in the small is very 
helpful, since if there is a bug you need to test it for a time roughly 
proportional to the mean time between failure for that kind of bug. 
Ideally you do it independently in parallel to get data faster.

However, social technologies have scaling properties that matter. The 
behavior among 10+ team members is very different from 100+ groups or a 
100,000+ population. Social dynamics matter: small groups often get 
effects from the individual relationships, while larger groups have 
anonymity effects. Also, the number of minds trying to find loopholes 
and ways to crack the system increases with the group size.

If there is a chance p per participant of finding a problematic 
loophole, the chance that it will be found is 1-(1-p)^N, which becomes 
large for N=-ln(2)/ln(1-p) ~= ln(2)/p. So if p is 0.01, then you need 70 
people to have about 50% chance of finding the bug. Once N is on the 
order of millions, you can no longer run tests - your system is part of 
society (or is society), so stuff with p less than one in 1.4 million 
cannot be tested away, you have to deal with it as it happens for real.





On 2015-12-20 02:30, Flexman, Connor wrote:
> On Fri, Dec 18, 2015 at 12:14 AM, Adrian Tymes <atymes at gmail.com 
> <mailto:atymes at gmail.com>> wrote:
>
>     On Thu, Dec 17, 2015 at 6:49 AM, Jason Resch <jasonresch at gmail.com
>     <mailto:jasonresch at gmail.com>> wrote:
>
>         Check out Liquid Democracy, now used at Google:
>         http://www.tdcommons.org/cgi/viewcontent.cgi?article=1092&context=dpubs_series
>
>
>     Good luck refining it into something most people will be able to
>     understand enough to trust.  (Not a sarcastic comment: that really
>     is a large problem getting this system deployed on more than niche
>     electorates.)
>
>
> Perhaps if more companies begin using it like Google does, or we roll 
> it out in very sub-national realms, people will slowly become 
> comfortable with it? It looks promising enough that it might be worth 
> pushing for on a smaller scale, to see how far people can take it 
> before insurmountable problems show up.
> Connor
>
>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat

-- 
Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20151221/5883eefc/attachment.html>