On 1/11/12 05:36 AM, danimoth wrote:
Hi list,
I was thinking about a slightly modified Kademlia network, with
an economic incentive to reduce free-riders.
I thought this incentive as bitcoin's fractions exchanges that may take
place as payment for services provided by nodes.
For example, there are A and B. B send a FIND_VALUE to A; if A has the
value, before sending the resource it requests a payment from B. When B
pays, the resource is sent.
There shouldn't be big modifies to be done to already deployed Kademlia
networks, but I think it must be joined by a reputation system to avoid
routing and payments problems, which I identified in [1].
So, my questions are:
1. Is this idea any good or it's only a waste of time?
The essential solution to all trade imbalances relies on money. So if
your problem is some form of asymmetric trading, you need a payment
system, of some form, and you need an exchange of some form.
Beyond this simple statement, however, is a sea of ideas, in which one
can easily drown. E.g., you've identified a simple exchange process,
discovered a weakness, and then proposed a reputation system to cover
the weakness. Adding a reputation system to solve issues is like a deux
ex machina in systems; Rep systems are little understood and generally
or frequently crap, so chances are you'll end up building something that
won't work, and wasting a lot of time in doing it.
Better to avoid that and come up with a payment system that doesn't need
reputation - or at least one that doesn't lean so heavily on it.
As a field you can research it, but you have to be extremely skeptical
because much of what is written is unreliable at some level or other.
For one example, everything written about gold is tainted by Central
Bank marketing (for their own currency). This makes it very confusing
if one just reads and assumes what is written is fact...
Alternatively, one can build it and try it. But the cycle times are
long, it takes a year or so to write a decent money system and get it up
and rolling.
Alternatively, you cut the gordian knot and make everything free. The
system has to work under this constraint. That works for somethings
(open source software, songs sharing, etc) but not for all things.
2. There have been previous experiments similar to what I'm proposing?
Mojo Nation tried to be an economically informed p2p system, but seemed
to run out of grunt as a project. It failed because it tried to solve
every problem, and drowned.
http://financialcryptography.com/mt/archives/000571.html
http://financialcryptography.com/mt/archives/000572.html
In contrast, the projects that spun out of it - BitTorrent? Tahoe? -
reduced their problem set dramatically. Either way, you might find
Mojo's design to be well worth studying, people say the design wasn't wrong.
[1] Enforcing Collaboration in Peer-to-Peer Routing Services
(by Tim Moreton and Andrew Twigg)
That's an unfortunate turn of phrase there, which rather strikes at the
heart of the problem you are trying to solve :)
iang
_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers
