[ZS] Quantified Prestige

[Michael Hrenka]

On Monday, 10 September 2012 20:14:18 UTC+2, Mark Nuzzolilo wrote:
>
> On Sun, Sep 9, 2012 at 12:34 PM, Michael Hrenka <meta... at gmail.com<javascript:>> 
> wrote: 
> > I just finished the updated version of the documentation of my 
> > reputation/money system with the updated name Quantified Prestige 
> (formerly: 
> > Prestige Fluido). 
>
> This is very good stuff, but it must be handled with care.  Similar to 
> what I was working on last year.  I'm obligated to ask a few tough 
> questions. 
>
> 1) This prestige score appears to be linear.  Any thoughts on having a 
> two-dimensional score, score as a function of context? 
>

No, how should that work? Context isn't even 1-dimensional, but much more 
complicated. If you really want different contexts, then use a different 
Quantified Prestige Network for each context. Finally, I want to create a 
service that makes it very easy to set one up. Like what Wordpress did with 
blogs. You can get one on wordpress.com, or you can download the software 
from wordpress.org and host one on your own.
 

>
> 2) How do points get taken away?  In my system, a negative vote from a 
> person would equal the inverse of about six positive votes from that 
> same person, *but* a reason would always have to be given for the 
> negative vote, and those reasons could themselves be voted on, thereby 
> lowering the vote's weight if it did in itself receive negative votes. 
>  This heavily gives incentive to quality over quantity, and helps 
> ensure that the scores are themselves indicative of the quality of an 
> individual's contributions.
>

User can take Esteem Points they allocated to others away at any time 
without having to give a reason. Think about it like an unrestricted 100% 
money back guarantee. Of course, you can give a reason for giving and 
taking back EPs to give clear information, but that's option. I don't think 
it's a good idea to force people to rationalize their decisions.

I had thought a long time about including negative scores in my system, but 
I decided against it, because it's hardly possible to prevent any serious 
abuse of such a system. I'm a proponent of positive reinforcement.

Your system sounds interesting. Is there any incentive to vote on the votes 
of others, or even to have a look at them? Most people probably won't 
bother with meta-voting, unless they see it as some kind of game or hobby. 
Unless they are so fed up with systems that don't allow meta-voting that 
they see it as obligation to meta-vote where it's possible. But then such a 
system would become like Reddit on steroids.

Anyway, if you have good ideas for modifying Quantified Prestige to a kind 
of "Qualified Prestige", you are free to develop that alternative concept.

>
> 3) Do two relatively equal political opponents have a widely different 
> amount of prestige points based on their popularity?  How do you deal 
> with bias? 
>

It's a bit much to expect tools to fix psychological problems. Every voting 
system is prone to biases. Is there any way that is proven to reduce the 
level of bias in such systems?
 

>
> 4) There are many different pieces to this system, including Esteem 
> Matrix, re-allocation, user circles, etc.  Are each of these an 
> absolute necessary component to the system and have you completely 
> worked out why they are necessary?  I would strongly suggest including 
> the "why" in the paper, and if you can't come up with a reason, 
> *consider* dropping the mechanic from the system. 
>

Reallocation is necessary because otherwise a system with a fixed quota of 
points would stop working once all your points are used up. There is a 
fixed quota in order to prevent Prestige inflation. It might be possible to 
use a monthly quota instead and to weight points with how recently they 
were allocated. I haven't given such a system a lot of thought. Need to 
think about that option.

Many parts of the system are just for convenience or security. They are 
optional in that sense. The core of the system is Esteem Point allocation 
(-> Esteem Matrix) that determines a Prestige Score for each user (-> 
Prestige Vector). Strictly speaking the Spread Factor isn't necessary, but 
it provides a strong incentive to give out points generously to many 
different users - and to get a meaningful representation of relative 
prestige.

I really tried to make my system as simple as possible and as complex as 
necessary to avoid (almost) obvious failure modes of the system - and to 
make the system comfortable for the users, because otherwise it won't 
become popular. And a reputation/money system that's not popular isn't very 
useful.

>
> 5) Will ordinary users with low prestige find it easy or difficult to 
> "get in at the ground floor" and be able to use this system to help 
> build their prestige?  Or will these methods only be more accessible 
> to people who already have large networks of people? 
>

Having large networks of people may be useful, but their usefulness is 
limited if the Esteem Trust system is used, because it makes Esteem from an 
independent individual much more valuable than Esteem from a cluster of 
people that all know each other. The system should work fine if new uses do 
anything cool or useful that catches some attention by many users. It's 
supposed to encourage creating and doing publicly useful stuff. 

>
>
> This is really great.  Just be careful because widespread adoption of 
> systems like this can easily have negative effects in addition to 
> positive effects.  You want to reward people for "doing the right 
> thing" and not necessarily for doing what is popular.
>

In principle, I totally agree with you here. But how do you make a clear 
distinction between what's right and what's popular? Who is supposed to 
judge that? I mean, that's exactly the problem of democracy: It can't 
distinguish between good (and popular) parties and bad, but popular 
parties. However we haven't found a system yet that works better than 
democracy when nobody has the clear authority to judge what's good and 
what's bad.

Anyway, thanks for the great feedback! It's really helpful! :)  

-- 
-- 
Zero State mailing list:
http://groups.google.com/group/DoctrineZero




----- End forwarded message -----
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE