[freehaven-dev] Discussed Today

[dmolnar@belegost.mit.edu]

* Figured out a preliminary notation for updates which will go into 
	/doc/model.tex 

 	- We all pointed out that we should try to find some nice 
	mathy problems which we could then pose to our friends involved
	in that discipline. A minor sticking point : MIT math doesn't
	require a thesis or AUP, instead requiring courses.

	- Hopefully Guruswami's paper will help us formulate these 
	questions and guide our modelling. Heaven knows we have 
	a lot of stuff left to model...
 
* I mentioned that I had heard about the Dempster-Shafer "belief model."
  No one seemed to know about it, so I'll go read about it this week and
  hopefully we can discuss w/Brian next week. It seems to be a
  math framework for dealing with partial information. 

* Political discussion. Roger points out all the ways that the world 
  fails to acheive any basic notion of fairness or equity. 

	As I understand it, the argument goes like this 

	1. The world is in the death grip of large corporations and
	states. This has led to horrible consequences. 

	2. Therefore, the current world order is not worth preserving.
	More than that, it is worth shaking up and pulling down. 

	3. Free Haven will be a tool which may help in pulling down the
	current order. It won't do it by itself. It may not even be 
	that great a tool, but gosh darn it, it's a hell of a lot better
	than doing *nothing*. 
	
	4. Therefore Free Haven is a good thing, over and above any 
	effect it may have on "nice kinds of speech." 

  We then had a brief argument about capitalism and peoples' motivations,
  manufactured consent, connotations of the word "greed", and how to 
  link this general principle "shaking up the world is good" to specific
  examples. 

* ACLU of Georgia v. Miller was brought up. 

http://www.hcs.harvard.edu/~dmolnar/miller.htm

is the preliminary injunction, which is more or less the document you want
to read (the others are just cut and dried "so and so ordered.")

press releases at
http://www.aclu.org/news/n062097b.html 
http://www.aclu.org/news/n092496a.html

		Short summary 
		- Miller is the general counsel for the state of GA
		- the law in question outlawed the following : 

	" knowingly to transmit any data through a computer network . . .
for the purpose of setting up, maintaining, operating, or exchanging data
with an electronic mailbox, home page, or any other electronic information
storage bank or point of access to electronic information if such data
uses any individual name . . . to falsely identify the person . 

and also

any person . . . knowingly to transmit any data through a computer network
. . . if such data uses any . . . trade name, registered trademark, logo,
legal or official seal, or copyrighted symbol . . . which would falsely
state or imply that such person . . . has permission or is legally
authorized to use [it] for such purpose when such permission or
authorization has not been obtained."

The motivation seems to have been fraud and "online impersonation." 
		
		- one of the first, maybe the first state law to 
		impose content restrictions on speech. 

		- found unconstitutional because
			- too vauge ("falsely identify" ??)
			- no intermediate measures attempted by the State


		- What does this mean for us?
		IMHO it's nice for U.S. Free Haven servers, but 
		the decision relies too much on the U.S. constitution and 
		U.S. law to be a firm ground. We should look at 
		the ACLU brief instead.


The Ohio case is the one on disseminating anonymous campaign materials.
More on that later. 

Lexis also reveals some interesting law review articles on the subject of
online speech & law. Snarfed and posted on my page :

"State Regulation of Anonymous Internet Use After ACLU of Georgia v. Miller" 

Covers the Miller case and then reviews current attempts by states and
federal government to regulate anonymous speech. Outlines a strong case
for the proposition that anonymity is Constitutionally protected. 
This is really good...but only if you're working in the U.S. legal system.
It's not yet clear to me how much can be extracted from the morass of
precedent and used elsewhere.

http://www.hcs.harvard.edu/~dmolnar/anon.htm

"Fraud: American Civil Liberties Union of Georgia v. Miller"

Reviews the case and then suggests what future laws would look like in
order to avoid the same problems which killed the Georgia law. Could
be a blueprint for successful future laws of this type. Never, never 
trust the Supreme Court...

http://www.hcs.harvard.edu/~dmolnar/berkeley.htm

"NOTE: THE GERTZ DOCTRINE AND INTERNET DEFAMATION"

Discusses defamation in the face of the Internet. The "Gertz Doctrine"
refers to a particular case's decision that defamation about a public
figure had to be proved to be "actual malice." 

Discusses the idea of a "marketplace of ideas" and whether the ability to
reply to "defamation" constitutes a kind of "self-help" remedy. 

http://www.hcs.harvard.edu/~dmolnar/gertz.htm

A panel on "should the Internet be a family-friendly zone? and how?"

Discusses kiddie porn and anonymous speech. 
Goes into issues of liability for public libraries offering internet
access, and from there touches on provider liability in the U.S. 

http://www.hcs.harvard.edu/~dmolnar/family.htm


* TIMING ATTACKS

	We discussed attacks in which two servers conspire to frame
	a third server. The attack depends on the two servers 
	giving misleading timestamps to the third server, but 
	timestamps which are "off" by a small enough time to 
	be explained by "clock drift." 

	The exact details of the attack should be a separate message
	(this is already becoming a behemoth)

	We speculated that a low-latency mix-net would reduce or eliminate
	timing issues of this sort. We also need to look into logics
	or simulation methods which can tell us about this kind of
	attack. (Paul Syverson's papers?)

* BUDDY ATTACKS and ALTERNATIVE MODELS

	We went over buddy attacks again. Again, separate message. 
	I don't think I'll write that tonight..

	Alternative models proposed :
		
		- Everyone keeps a record of every trade ever made.
		Then randomly decide to check on trades now and then. 
			- Note : creates audit trail, but 
			people may decide to break the trail with
			some probability.
		
		- trading "responsibility for share X" the same
		way we trade shares. Not sure how distinct this
		is from buddy trading. 

* CITIZENSHIP IN THE SERVNET

	At one point we realised that squawking is not always in 
	a node's best interest. To be precise, if a node of little
	status squawks about a node with high status, 

		a) no one believes that the high status node is bad
		b) everyone looks at the low status node funny

	Why, then, would a low status node squawk??
	Because it is a good "citizen of the servnet." 
	This concept was sort of poked at, then tabled. 
	It seems like something we should make VERY EXPLICIT.
 
	Especially since my default mode for evaluating protocols is
	to assume that if there's no good self-interested reason to do it,
	a party won't. Changing that assumption needs to explicit and
	very very well motivated. 

		
* TRADING witht he assumption of a RELIABLE (acked) CHANNEL

* That's what I can remember. Post if I forgot something. 

Thanks, 
-David