Re: Possible Idea For a Sem Web Based Game?

On 15 December 2010 09:39, Pierre-Antoine Champin
<swlists-040405@champin.net> wrote:
> Melvin,
>
> (sorry to the others, I used the wrong address to post to the mailing lists,
> so my previous message didn't get through)
>
> you wrote:
>> You dont need smart servers, just socially aware cloud storage.  Flat
>> files are fine, you can let Agents do all the middleware.
>
> ok, I should'nt have used the term 'server'; I was not considering
> cloud-storage (yet)...
>
> It does not really change my point, though: if you only trust a single agent
> ("dungeon master") to manage game-data and enforce game logic, you end up
> with a rather centralized system.

You can trust multiple agents.

>
> On the other hand, distributing the game logic is harder:

Harder but more fun!

> - how do different agents maintain consistency of the game?

Rules and game logic.  Theres a number of ways, one is simply to
encapsulate game login in the agent code.

> - how do you trust a newly discovered agent?

Web of Trust.  This should be a sem web scale service.

> - how do you know that several agents are not colluding to cheat?

You dont know that but distributed systems have a good track record of
fault tolerance.

>
> But obviously, I merely scratched the surface, while you seem to have
> clearer ideas on the subject... :) -- thanks for the links by the way.
>
> I'll keep an eye on that.

I've summarized some of the links in this thread under the concept
Linked Open Gaming

http://linkedgaming.org/

If you have a game world or client I'll add to the list.  Note that
the ontology now has items.

>
>  pa
>
> On 12/15/2010 12:39 AM, Melvin Carvalho wrote:
>>
>> On 14 December 2010 22:21, Pierre-Antoine Champin
>> <pierre-antoine.champin@liris.cnrs.fr>  wrote:
>>>
>>> Hi,
>>>
>>> this is fun, but we have to ask ourselves: what is the added value of
>>> RDF/sem-web/linked-data here?
>>> What does http://drogon.me/ have that wouldn't be possible with HTML+PHP?
>>
>> To me the Web, particularly the Sem Web is a universal space whose key
>> advantage is interoperability.
>>
>> So, each world can interop with similar worlds.
>>
>> Also worlds can operate with other parts of the Semantic Web Space.  I
>> use the acronym SEMANTIC to describe key areas:
>>
>> Social
>> Entertainment
>> Markets
>> Access
>> Nearby services
>> Trust
>> Information management
>> Currencies
>>
>> So a game can be social, have trading with virtual currencies and
>> markets, you can interact with a personal or public web of trust, with
>> existing information or things in the real world in your locality (eg
>> augmented reality), using web standards.
>>
>> Granted each area on the list is still in an embryonic phase.  But
>> this is a level of interop simply not available in other systems.
>>
>> We've seen linking of basic social and trust in PHP+HTML (facebook)
>> and social and entertainment (zynga) get some traction.  But when we
>> have interop across all areas we'll have a that much more powerful
>> system.
>>
>>> Don't get me wrong, I think those ideas is great, and kudos to you guys
>>> for
>>> turning them into code so quickly!
>>>
>>> My two cents on this question:
>>>
>>> 1/ linking to real world data is definitely an interesting track, because
>>> this leverages existing linked data for the purpose of the game
>>
>> Yes, agree, leverage interop.
>>
>>>
>>> 2/ another way to use linked data principles is that the game can be
>>> distributed, even more so than an HTML-based game.
>>
>> Exactly.
>>
>>>
>>> I imagine that every character, place, item... could have its own RDF
>>> description, linking to each other. A triple between two objects (X is
>>> located at Y, X owns Z...) is considered true only if both the subject
>>> and
>>> the object claim it.
>>>
>>> This implies that the RDF files are hosted by "smart" servers that will
>>> allow updates by anybody, but under certain conditions.
>>
>> You dont need smart servers, just socially aware cloud storage.  Flat
>> files are fine, you can let Agents do all the middleware.
>>
>> http://www.w3.org/DesignIssues/CloudStorage.html
>>
>>>
>>> For example, a place will acknowledge that it contains a person only if
>>> the
>>> person claims to be in that place, and only there.
>>
>> This is game logic.  It need not reside on a server.
>>
>>>
>>> The protocol might be tricky to design for more complex things like
>>> transactions. I imagine that an item would change its owner only after
>>> checking that both the old and the new owner explictly agree on the
>>> transaction
>>>
>>>  <#me>  game:agreesOn [
>>>    a game:Transaction ;
>>>    game:give some:sword ;
>>>    game:receive some:money ;
>>>  ]
>>
>> Im working on an economic aspect.  This is an interesting proposal on
>> transactions and contracts:
>>
>> http://iang.org/papers/ricardian_contract.html
>>
>> I have reasonable confidence we can introduce a sophisticated economy
>> that can be leveraged by all sem web projects, probably before end of
>> next year.
>>
>>>
>>> Plus, the buyer would have to trust the sword not cheat on them and
>>> return
>>> to its previous owner without notice...
>>>
>>> Fights will probably be even trickier... But I think the idea is worth
>>> exploring...
>>
>> Many ways to model this, again agents can handle this.
>>
>> Traditional architecture is
>>
>> client<-->  middleware<-->  data store
>>
>> Web oriented architecture is more flexible and can have, in addition:
>>
>> client<-->  data store
>> client<-->  agent<-->  data store
>> client<-->  data store<-->  agent
>>
>> With trust and PKI regulating actions.  Of course we see why WebID is
>> important here too.
>>
>>>
>>>  pa
>>>
>>>
>>> On 12/02/2010 01:20 AM, Melvin Carvalho wrote:
>>>>
>>>> On 2 December 2010 01:13, Toby Inkster<tai@g5n.co.uk>    wrote:
>>>>>
>>>>> On Wed, 1 Dec 2010 23:06:42 +0100
>>>>> Melvin Carvalho<melvincarvalho@gmail.com>    wrote:
>>>>>
>>>>>> I think the next thing I need to model is 'items'.
>>>>>>
>>>>>> At present need to work out a way to say a location has an item.
>>>>>
>>>>> Perhaps model it the other direction?
>>>>>
>>>>>        <item22>    game:initial_position<node394>    .
>>>>
>>>> I was thinking more along the lines of:
>>>>
>>>> Location x has
>>>>   item 1
>>>>   item 2
>>>>   player 1
>>>>   player 2
>>>>
>>>> With a trusted Agent(dungeon master) adding them to a copy of the game
>>>> world.
>>>>
>>>> The DM is allowed to sparql update the locations via insert and
>>>> delete, contains the game logic, and interacts with players.
>>>>
>>>> In this way you can have 1 or more DM's given access to administer the
>>>> worlds, the best DMs would become 'resident' in the game world.
>>>>
>>>> Agree, it's not the only way to model it, but I like the idea of a
>>>> file based solution mediated by agents.
>>>>
>>>> Make sense?
>>>>
>>>>>
>>>>> --
>>>>> Toby A Inkster
>>>>> <mailto:mail@tobyinkster.co.uk>
>>>>> <http://tobyinkster.co.uk>
>>>>>
>>>>>
>>>>
>>>
>>>
>>
>
>

Received on Thursday, 16 December 2010 21:50:27 UTC