[freehaven-dev] what I'm doing for the paper

[dmolnar@belegost.mit.edu]

Hi,

So we met tonight and discussed what to do with our paper. I am personally
committed to the following things:

	1) Revamp the related works section. Introduce a distinction
	between the "robustness" notion used in related works 
	and our "reliability" notion. (maybe come up with new,
	not so confusing terms?). 
	The distinction is that the previous protocols have
	dealt with 
		- an "all or nothing" notion of robustness;
		the distributed MIXes work if k out of n
		are honest, and fail otherwise.
		(I need to read the papers carefully to 
		make sure this is precise; i.e. no gap
		where it's not clear whether the MIX
		fails or continues)

		- solely with ensuring robustness in
		the protocol and not with reputations.
		

	2) Delinate 3 different "approaches to anonymity" 
		1 - via protocol
		2 -  via reputations
		3 - via having someone pay for it (ZKS))

	3) Move the model from section 6.2 to a new
	section 3 "MIX Model and Assumptions" 
	This section will have two parts
		1 - Anonymity-Breaking Adversaries
		which uses the notion of adversary
		inherited from all previous MIX papers.

		2 - Reliability-Breaking Adversary
		which uses the model from 6.2

	We don't know how to unify these yet, but that
	will be future work. The point is to address
	the reviewers' comments that we don't have a model
	or a sufficient definition of passive adversary. 

	4) Fix section 6. Right now it assumes picking
	with replacement. Except as a reviewer pointed out
	this is not the right way to think about it. 
	Also other gremlins which popped in. 

	
-David