[MLB-WIRELESS] Re: Node x is over this way -was- Applications

[Ben Anderson]

> >>example - problem: bandwidth congestion
> >>how does mojo attempt to solve this? by creating a class system where
> >>those with more links get more mojo, while those on the outskirts, or
> >>those who cant afford to put 5 cards and 5 antenna on their roofs are
> >>penalized by the system. so how could those people gain more mojo? by
> >>sending more traffic. thus actually increasing the amount of traffic on
> >>the network, as these people try to earn "credits" so they can download.
> >>mojo is like a ratio FTP site, but part of the network, instead of just
> >>some horrible idea on a ftpd.
> >>
> >
> >I agree, horrible idea on an ftpd.
> >
> and on a freenet

_Why_ is it so bad?

> > And implemented 'nastily' it'd make the
> >network suck.  I'm proposing using the mojo only to give people the
ability
> >to get low-latency access to the network.
> >
> at the expense of everyone else

Of course.  I think this is better than allowing leaf-nodes to fully
saturate the bandwidth of the network trading mp3's, movies, wares or porn
(not that this content has anything inherantly wrong with it as data), just
that it DoS's the low-latency people.  Some tradeoff has to be made.  Why is
your tradeoff in the direction of 'good for leechs' (which is inherantly bad
for network scalability, as it encourages people away from taking
responsibilty for addint to the network infrastructure, and secondly,
encourages altrustic people who do provide infrastructure to provide less
infrastructure as people complain about shit access) better than my tradeoff
that allows low latency traffic in exact proportion with the networks
capability to carry it?


> > Packets without mojo would
> >traverse the network as if the mojo didn't exist
> >
> except for the fact that they have low priority?

Lower priority.  An overloaded link is obviously going to be overloaded,
nothing will change that.  Nodes not contributing to the bandwidth of the
network only get to use the networks unused capacity.  Ie it's all you can
eat "unguaranteed" bandwidth.  With data with a QoS metric (mojo) being the
priority traffic.


> >-- overloaded nodes would
> >still be overloaded, except now people could actually still get
low-latency
> >access to it, in what I think is a fair manner (and in a manner that
causes
> >the network to grow to encourage increasing this 'resource in demand')
> >
> again at the expense of everyone else

There's a tradeoff.  Read my comments on the first paragraph.

> >There's a lot of economics theories that cover this concept quite well.
> >People wouldn't earn credits sending information onto the network.
> >
> so how *do* they earn mojo?

By providing services that are in demand -- I haven't worked out the exact
metrics for what makes sense to earn mojo, and how much -- i'm reccomending
simulation for that.  Though I'd suggest that complete leaf-nodes, who
provide nothing should never earn mojo.  A leaf node that caches data for
another node, reducing the traffic from another part of the network should
receive mojo proportionally.  Leaf nodes could also set up nodes in
congested areas, effectivly 'buying' mojo, shifting it to their leaf system.
Of course, users who are altruistic can give their mojo to leaf nodes.  If
the network would scale without mojo, then there should be enough people to
give mojo to the leaf nodes such that everyone always has enough mojo for
their needs.


> >Your solution requires someone altruistic enough to actually set the link
> >up, and make it available.
> >
> can you stop saying that? i've shown several times why someone would set
> the link up

I've also shown why the system will evolve away from people being altrustic
with their time, and given examples of it.  Do you require more examples of
people getting sick of being altruistic?
And the 'mojo' solution doesn't stop people giving the mojo away for free,
either.  It just provides a QoS metric that can be used to prevent DoS
attacks, and also provides a congestion control technique.
Show me a workable alternative.

> >  What's more likely to happen in that scenario is
> >that the people with congested links stop giving it away because hundreds
of
> >users are flaming them for providing a crap service.
> >
> why would people flame them for helping the network become less
> congested? your assumptions here are quite odd. and besides, if the net
> is still bad, one of the others can provide another link, or add another
> card combined into the same antenna to double their bandwidth. READ:
> they already have a motive to do this, without mojo.

No, people would flame people for having a congested link...  example, I'm
in area a, you're in area b.  We have a directional link.  We each have a
hundred users in our areas that want to talk to each other, send movies,
etc.  The network slows down to 3kbits/sec/user, and suddenly you have 200
emails telling you you're crap, should upgrade, make the system work, loser,
"we don't need a committee" and generally people don't deal with this lack
of appreciation, decide it's not worth it, and just turn it off.  READ:  the
'reason' will get evolved away by large numbers of leaf nodes.


> >Prime example, look at what happened with the mailing list.  Few flames
> >happened, the people being altruistic with their time WRT committee stuff
> >decided it wasn't worth their time, and packed up and left.
> >
> not exactly, i think they've reconcidered. the problem is different with
> a mailing list. half the people flaming have no intention of doing
> anything for the network, they just like to bitch for the hell of it. on
> this network, its going to require a lot of personal interaction, going
> to the other person's house to test LOS, being on the cell phone with
> them while aiming antenna, and finally a beer when a link is finally
> had, you're going to end up knowing the people you're linking to, its
> going to be a community building project as much as it is a network
> building project. if that same guy you just spent a weekend with setting
> up a link starts flaming you, he's got a mental condition. ;)

Half the people connecting to the wireless network will have no intention of
providing other people with better access.  They'll bitch at people who do
provide "for the hell of it"
In the short term, it'll be very 'beer and pringles on roof' community
building exercise.  This is good.  But as it becomes larger, and scales
towards being a general comoddity item, the people accessing the
infrastructure will just plug in, and turn on.  No community building, no
face-face contact.  They'll just expect it to work.  And when it doesn't
they'll be pissed that their mp3 got cut off halfway through...

> >  Very
> >dissapointing in my opinion.  I think I can come up with a lot more
> >real-world examples where relying on peoples altruism to scale stuff
breaks.
> >
> aren't you just a ray of sunshine in a dark world? have some faith in
> people, you may be surprised.

I put my faith in people all the time.  I get dissapointed with people all
the time.  Every now and again, a gem pops up.  All too often, that gem gets
squashed by people taking advantage of them.  I don't see how adding a
'guarantee of scalability, and DoS prevention layer' precludes people giving
stuff away for free anyway.  It just implicitly protects against people
being assholes.
Would you argue the GPL is a bad thing?

Ben.


--
To unsubscribe, send mail to minordomo at wireless.org.au with a subject of 'unsubscribe melbwireless'  
Archive at: http://www.wireless.org.au/cgi-bin/minorweb.pl?A=LIST&L=melbwireless
IRC at: au.austnet.org #melb-wireless