On Monday 10 November 2003 04:19 pm, Ken Corson wrote:
Tom- If I follow you correctly, you are saying that a node may choose to slant it's bandwidth usage in favor of sending queries on behalf of the local node, and disfavor the service of data transfer to others. If a node sets rtMaxNodes=200 , he can hit more nodes with requests, but only if he tones down his data transfer responsibilities. And it will cost him more to develop useful/decent estimators per node. So how can we encourage other nodes to stand up to their responsibility of providing data transfer to the network ?
Ken
note: the network currently favors local requests by default. There is no since in stopping this.
Tom - I'm not suggesting we actively attempt to defeat this behavior. Only
to take notice of it. The reason I asked the above question was to make
people consider what I described as "routing symmetry," in a message
dated 11/05. I only received feedback from Matthew, and hoped to see
more discussion of it... he also believed it is worth discussing. Basically
all node-pairs would communicate in both directions, as requestors and providers. No one said this was a horrible idea, or a good one either.
If the remote node is good at providing data people want, I don't care
if that node is the original publisher of that data or not. Conversely,
I don't care what percentage of his requests are actually originating on
his node. As a publisher, he _would_ be providing a beneficial service
(of content provider) that can be measured by my local node. And I would
reward him by giving him a higher priority within my set of peers.
If you want to make sure they rout others data too? You can't. Especially if we are tolerating transients.
Agreed, for now. Routing symmetry -could- be a step in the right direction for correcting this. It deserves to be talked about more, because I haven't seen any other suggestions for improving this condition, other than building a more complicated web of trust between nodes. Which I believe we should also support, long term. I specifically disregarded transients in my "routing symmetry" suggestion... supporting transients is another issue that needs discussion eventually.
If you want a long term and more ideal solution to insure fair routing and attack resistance go back and read my proposal for a "decentralized trust based" routing scheme a little while back.
I want to read this, please mail me a (date posted) pointer. Thank you. I just scanned many of your past 2 week's worth of postings. Was this the idea that somewhat approximates IIP's tiered structure? Or did it involve your discussion of multiple hashings (like hash cash)?
Incidentally, I must apologize for my initial reaction (two weeks back) upon reading your message, which began :
"WAIT. I've got it! Add another level of hashing. So the content is encrypted with it's hash, and it is stored in the hash of the hash of the hash, and attached to the request is the hash of the hash. "
I fell on the floor laughing, because I /thought/ you were being sarcastic, suggesting yet another method to make the whole network fall hard on its face. When I saw serious responses to it, I had to go back and reread! The nuances lost in this form of communication are troubling. I was reading too fast, and "hash of the hash of the hash" did not seem a serious concept at first. My apologies :)
-Ken
_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
