[MUD-Dev] Social Networks

[Jeff Cole]

From: Matthew Dobervich
> Jeff Cole wrote:

>> While evolution in content is indeed important, I think it begs
>> the question of whether the existing content is efficiently or
>> effectively implemented.

>> That a social network is scale free, implies that the fitness
>> curve is going to largely depend on the nodes and not the
>> content; that is, that additional or different content is not
>> going to drastically change the networks link distribution.

> I would have to agree with Dave.  Jeff, if I understand you
> correctly the only factor that determines differences in fitness
> curves, number and concentration of hubs, and network distribution
> of one persistent world's population as compared to another's
> would be the "type of person" a given world would attract.  What
> else would "depends on the nodes" mean?

You do not understand me.

Suffice it to say, that my position is that developers have much
more effective and efficient methods than new-and-improved content
to affect the topology of the social network.  Transactional costs
associated with establishing links necessarily reduce any effect of
content on network topology.

For a given set of game "goals," the game is going to draw some
audience.  Those are your nodes.  Period.  With respect to network
topology, such goals are likely to have its greatest effect upon the
degree exponent of the power curve.

The nodes have certain social proclivities (read: fitness).  Whether
you want to conceive it as an absolute "Fitness" consisting of many,
varied and smaller "fitnesses," of which only a number of which
affect a node's behavior in a particular network (i.e. the "Fitness
DNA" composed of "fitness genes"), or you want to speak of "fitness"
and as that quality of a node that affects its behavior in a given
network is immaterial.

Take 100 people and put them in a social space.  A certain network
topology will evolve.  Take the same 100 people and divide them into
10 groups of 10 and impose some cost upon cross-group interaction
(any cost: lingual, physical, let your imagination run wild etc.) 
and a completely different social network will evolve.  Same nodes,
different topology.

Beyond that, it is naive to think that a fitness, let alone Fitness,
is truly quantifiable, or that all nodes with equal numbers of links
are equally fit.  So, to redistribute the 100 people among the 10
groups, maintaining the same cross-group transactional costs, you
will likely see yet another topology.

The manners in which a Gameplay Mechanic segregates nodes and impose
costs upon cross-group linking will affect network topology far more
than additional content.

> I'm not saying more content (defined as more mobs, weapons, skill,
> spells, quests, etc.) will change the topology of a given
> peristant world's social network, but I DO believe that content
> (defined as game systems) can clearly do this.  Examples of this
> type of content (and I think it's what Dave was getting at)
> include communication systems, guild systems, economic systems
> that require player cooperation, and PvE/PvP goals that require
> player cooperation.  I would also argue this is the ONLY way to
> change the topology of a given world's social network.

To define such as "content" reduces to tautology the issue.

> To me the only power a developer has to increase the number of
> hubs (and hence the clustering coefficient of a given world) is to
> increasing the number of activities (facilitated by additional
> game systems) that create player connections, hence activities
> that REQUIRE player cooperation.

It's not necessarily about increasing the number of hubs.  If
Barabasi et al. are correct, then the social network will develop
the number of hubs that it can support.

It's not about a network having more hubs so that the network
resists the effect of losing a hub-- it is about making sure that
when (yes, it is when) the network loses a hub, how quickly (if at
all) can the network promote a node to take its place.

When a network loses a hub, all of the nodes that were linked to
that hub are likely to seek a new hub.  Rather than simply
redistributing orphaned links among the remaining hubs, the
scale-free model suggests that an existing node will become a hub.
The lower the cost to a node of establishing a link, the more
quickly the network can redistribute the links and recover.  The
greater the cost, the greater the number of orphaned nodes that will
simply leave the network (and the game).

> The only way I see to decrease transaction cost is to increase the
> number of communication tools at a player's disposal, although
> this is required to facilitate an increase in the clustering
> coefficient, I don't see it increasing the coefficient by itself.

There are plenty of costs beyond communication.  Travel time.  Class
imbalances.  Experience penalties.  Also, I am less concerned with
the clustering coefficients of the nodes and more concerned about
the degree exponent of the network.  That said, I don't understand
why you would think that the ability of nodes to communicate could
*not* dramatically affect a node's clustering coefficient.  Greater
communication will likely *decrease* a node's clustering coefficient
because it will likely increase a node's weak links.  As Raph has
pointed out, the work of Granovetter et al.  suggests that weak
links are integral to the network topology.

>> Also, decreasing such costs would increase the likelihood that a
>> network could recover from the loss of a hub insofar as remaining
>> nodes could more easily and quickly establish new links.

> Again, only if decreasing transaction costs by itself created new
> links.

Again, I don't understand how you can argue that decreasing such
transactional costs could result in anything but the creation of
more links-- especially weak links.

> I think what's required to see the profound effect on distribution
> you speak of is a combination of decreased transaction costs and
> game systems that REQUIRE player cooperation.

Well, to a large extent, of course.  But that's what we're talking
about.  Given a Gameplay Mechanic, a network will develop; it is the
Gameplay Mechanic which dictates which and to what extent "fitness"
characteristics impact the network topology.

I would add, though, that your use of caps implies too rigid an
application.  I am always surprised at those that dismiss the value
of a good hang with friends (online and otherwise).  That they
cannot recognize it suggests they missing much that MUDs and MMO*'s
offer.

> Remember, nodes don't WANT to link by nature, players work
> together because they have to.

Bah, cynics.  You're semantically splitting hairs for no good reason
on the definition of "want."

> I have a hard time picturing events that would cause this
> selective node loss, but would be curious to see some
> brainstorming of plausible scenarios.

MUD and MMO* history is replete with examples.  Boredom.  Burnout.
Exploiting.  Release of the next-gen.

> I see increasing the clustering coefficient of a world's social
> network as having two effects.  Guilds will be less likely to
> leave a world (as will individuals), but if they do leave, they'll
> leave together.

How does the clustering coefficient affect whether a guild leaves or
doesn't?  Seems to me that it depends upon whether the hub
(singular, as I think guilds are exhibit much more star topology
rather than scale-free) migrates.  I agree that increasing the
clustering coefficient would increase the likelihood that they leave
en masse.

Yrs. Affcty,
Jeff Cole


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev