[Cs5500] Fwd: improved game design

[Karl Lieberherr]

I wrote
http://www.ccs.neu.edu/home/lieber/courses/se-courses/cs5500/f11/maintenance/game-repair/welfareClaimsSubjectToRefutation
after my discussion with Ashish.

I would like to share Ashish's view with you. In class today we will
have to resolve those issues. (See project 5)

-- Karl


---------- Forwarded message ----------
From: Ashish Joshi <joshi.as at husky.neu.edu>
Date: Fri, Oct 14, 2011 at 4:06 PM
Subject: Re: improved game design
To: Karl Lieberherr <lieber at ccs.neu.edu>


improvements in Design
1.) Implementation of the Welfare set: This is where all the optimal
claims would go
    There are 2 ways to do it
a.) flag the claim as optimal
b.) flag the avatar as a teacher avatar
In the latter case the teacher avatar can still make a suboptimal claim before
reaching the optimal one and hence the welfare set could be diluted with
some suboptimal claims
The first case is the easier one to do but requires a change in the structure of
a claim which would lead us to change the avatars in hsr and csp along with mmg
We'd need a boolean (1/0) feild to identify a claim as optimal
PS: The Welfate Set is maintained on a per tournament basis which means it's
similar in nature to the forbidden list
All the learning that happens in the avatars between tournaments is by means
of the avatar creator studying the log files (smart/raw) and determinig where
their avatar is loosing/gaining reputation

2.) Challenging claims in the welfare set: This is where all avatars
babay/clever scan the
    the welfare set to determine if all the claims present in it are
optimal. If an avatar
    feels that a claim is suboptimal it can challenge the claim by
strengthening or refuting
    it. If it sucessfully strengthens the claim then it wins a large
amount of reputation
    and the person who made that claim will loose a large amount of
reputation (20% of their
    current reputations...just an arbitrary number).On the other hand
if the challenging
    avatar is unable to strengthen or refute the claim in question
then it looses huge reputation
    which goes to the avatar making that claim. This will lead to a
cleaner Welfare set.
3.) Duplicate claims: If a claim is flagged optimal and added to the
welfare set then the avatar
    should be allowed to make that claim again. One simple way of
doing it is not adding a claim
    flagged as optimal to the forbudden list. This would keep the
existing functionality of the
    baby avatar intact
These things are needed so that the teacher avatar does not need a
large number of distinct claims
just to prevent itself from being kicked out.
4.) Survivor/AtLeastAsGood Protocols: As per these protocols there is
no need for agreement or
    strengthening. Each avatar just says I'm the best or at least as
good. How does the avatar
    works in this case? How does the playground work? We still need to
work on these topics
Let me know if I've missed out something
Cheers
Ashish Joshi
On Fri, Oct 14, 2011 at 3:08 PM, Ashish Joshi <joshi.as at husky.neu.edu> wrote:
>
> hey professor
> will send it it a little bit
>
> Cheers
> Ashish Joshi
> On Fri, Oct 14, 2011 at 2:35 PM, Karl Lieberherr <lieber at ccs.neu.edu> wrote:
>>
>> Hi Ashish:
>>
>> please send me the improved game design that we developed yesterday asap
>> so that we can share it with the class.
>>
>> -- Karl
>