4st wrote:
On Wed, Jun 14, 2023 at 5:41 AM juan via agora-discussion <
agora-discussion@agoranomic.org> wrote:
I would really love some feedback.
----------------------------------------------------------------------
EQUALITY (Power 1.0)
For the purposes of this rule, a Player Property is defined by a set
of values (its range) and a way of naturally and unambiguously
assigning at each point in time a value in that set to each player at
that time.
For the purposes of this rule, a Criteria on a set of values is
defined by a subset of that set of values specified naturally and
unambiguously.
For the purposes of this rule, a Class is a set of players defined as
those whose value under a clearly specified Player Property is inside
the set of a clearly specified Criteria.
Equality is a natural player switch tracked by the Nomos.
The Protected Classes is a singleton switch tracked by the Nomos with
values on lists of Player Properties, without repetition, defaulting
to the empty list. To protect a class means to set The Protected
Classes to its former value with the specified Player Property
appended. A Player Property is protected if it is listed in The
Protected Classes, and unprotected if it isn't.
A Policy is a document unambiguously specifying a Class of players.
A Policy is Discriminatory if the specified Player Property of its
Class is protected.
A player CAN, with agoran consent, protect a class, by specifying the
Player Property to be protected.
A player CAN, once a week, by announcement, enact a Policy, provided
that Policy is not Discriminatory. When e does so, the players in that
Class have eir Equality increased by one.
----------------------------------------------------------------------
Just to be clear: my idea is to create an exploratory experiment on the
concept of protecting classes of people. I want to see the limits of
how can formalistic reasoning protect against discrimination based on
certain characteristics.
As far as I understand, in law (at least U.S. law) the idea is that a
rule is valid, despite desproportionally affecting individuals in certain
protected classes, if it can be shown that it is the least burdensome
way to accomplish certain valid policy goals.
That is, a rule is discriminatory if it affects protected classes
disproportionally in any way. I.e., if the criteria it establishes is
not independent (in a statistical, or event-based sense) to the protected
criteria. Since this would rule-out basically any meaningful legislation,
there are provisions for allowing this indirect discrimination when it
is justified.
So, the rule I'm proposing is exploring how far we can get in protecting
classes (through politically motivated action) when all that is forbidden
is *direct* discrimination.
There are some restrictions, though: the “natural” criteria is to
avoid certain logical sheananigans that would sidestep the
protections. I know, it isn't formal, but I can't think of a better
alternative.
--
juan
Ok, so, I'm just imagining the implementation of this right now.
The only way player properties exist is by a clear and unambiguous
definition, which the nomos would have to track each definition if it has a
reference. Are definitions meant to be like "player birthdays" or
something? otherwise I'm not sure what other protected properties exist
"how many assets owned by that player" "player names" "total blots in the
past year" "number of CFJs judged" "total months in an office" "number of
reports published" "total rules proposed".
As I understand it, player properties are indeed any such things that
already exist based on otherwise-existing gamestate, however the Nomos
only needs to track the ones currently in the Protected Classes list.
I also imagine that just because
it is clear and unambiguous doesn't necessarily mean "easily resolvable",
making this a very complex office if e is responsible for determining
outcomes.
Yeah, I would suggest shifting the burden to the player adding a player
property to the Protected Classes, and require em to accompany eir
announcement with a good-faith effort of specifying which players have
that property at that time (e.g. "Alice, Bob, and possibly Charlie
depending on CFJ 9001"). Then the Nomos's report of Equality (including
things like "Charlie: 2 or 3, depending on CFJ 9001") would self-ratify
as usual.
But that's just what happens when they're added. What happens after
they're added is also complex and unclear:
4st> Secondly, Policies don't do anything, which might be intentional,
4st> so that's probably fine I guess.
juan> Yes they do. Or rather, things are done with them. When players
juan> enact policies, certain players gain points in the form of the
juan> Equality switch.
But, based on this proto alone, the Equality switch doesn't do anything
either (e.g. grant radiance). And there's a strong implication that
Policies /should/ do something more, but no context for what that
something might be. At least with the "publish the hash of a message
ahead of time" proposal, we were able to come up with some reasonable
examples on our own, but this is so much of a blank slate as to just be
rather baffling.
