Re: Economic assumptions
Eric Hughes:
Let's take remailers as an example. One current suggestion is to add some sort of money system to the remailers as a condition of use. This is exactly the wrong priority at the current time. The remailers are already hard enough to use, and adding a payment system on top of that will make them used even less. Making a system harder to use increases the transaction cost.
The current priorities should be to lower these costs. When the remailer system begins to be overloaded, then adding some restriction on use, perhaps by means of payment or a payment analogue, will be warranted, because it will lower overall transaction costs, trading off ease of use for throughput and reliability.
What are some of these costs that should be lowered?
-- Finding out that remailers exist and what they do. build a remailer "who" server into each remailer -- Finding a remailer to use. ditto -- Deciding what remailer to use. ditto (remailer server should list remailer properties like keylength, private?, delay length, chaining?, mixing?, padding?, encryption required? etc) -- Figuring out how to use a particular remailer. standardize remailer help system, standard remailer command format (but not neccessaily the commands themselves) Sorta like an SGML for remailers -- Formatting a message for a remailer. see above -- Receiving mail through a remailer. Get/Creating a nice client. At the moment, 100% of the mail in my mailbox is encrypted. I wrote a script called "deliver" which encrypts incoming mail, then pipes it through procmail/slocal. I modified morepgp and made it a lot more user friendly (and recursive).
Reducing complexity cost: All of this could be lowered by creating an easy-to-use remailer client which is compiled (or perl/tcl interpreted) and installed with every unix out there so it becomes ubiquitous. If someone's machine doesn't have a client, they can telnet to a machine where one is set up (just like gopher, archie, www) by some generous cryptoaltruist. The current remailer solution of putting all of the remailer system complexity on the server side can't make remailers too easy to use. My Extropians list software attempted to make it easy to use by allowing commands to be contained in-band with messages to be posted. It's still too complex for the user who wants hot-key style operation. (which is why I will eventually write a client for it) Once you write a generalized client that can communicate with standardized remailers, you can easily include digicash/postage in the system.
There much more need for improving the ease of use of remailers than for paying for them.
Are you objecting to paying for remailers on a philosophical grounds (anti-property/money)? No one has proposed paying real money for remailer use (although that is a future possibility). There needs to be some way to authenticate remailer users and limit use in a "free" sense (instead of top-down rationing) The best way to do this is to use some form of monetary system.
The less expensive privacy is, the more privacy there will be. Privacy has non-linear benefit; the more that people are private, the better any individual's privacy actually is.
Every standard is enhanced by more people using it. However, this alone can't be a justification for making services into public goods which are free to everyone. If the Detweilers of the world take advantage of totally free remailers, they could end up limiting the privacy for all. The same "free" philosophy has killed many a porno/music/book site (or created absolutely long user queues reminiscent of food lines in the xUSSR) Spamming/Spoofing attacks on remailers must be dealt with. The situation is not helped by either-or logic. We need both ease-of-use and some notion of postage. -Ray -- Ray Cromwell | Engineering is the implementation of science; -- -- rjc@gnu.ai.mit.edu | politics is the implementation of faith. --
me:
One current suggestion is to add some sort of money system to the remailers as a condition of use. This is exactly the wrong priority at the current time. [...] [re: other transaction costs] The current priorities should be to lower these costs.
[later]
There much more need for improving the ease of use of remailers than for paying for them.
rjc:
Are you objecting to paying for remailers on a philosophical grounds (anti-property/money)?
Four words: Libertarian Political Correctness Witchhunt. If it's not really clear that I was making a statement about priorities, I don't think that repeating it a fourth time will help. If, of course, I'm not all in favor of monetarizing remailers immediately, could it be that I'm not in favor of ... money? Please.
The situation is not helped by either-or logic. We need both ease-of-use and some notion of postage.
Are you talking about me? It appears that you are, but I thought I was only comparing priorities. Enough of this. I'd rather discuss lowering transaction costs. rjc comments on my list:
-- Finding out that remailers exist and what they do. build a remailer "who" server into each remailer
I point out this doesn't help if you don't know where the first remailer is. What I was specifically referring to was public education. Were remailers ubiquitous, there would be a chapter on them in each of the latest rage of 'how to use the internet' books. They could be a well-used service, like archie. In fact, they are not. There are numerous reasons for this, some of which are self-referential (as in, there aren't a lot of remailers yet) and some of which are not. For example, there's no FAQ for comp.mail.remailer, because there's no such group. Why shouldn't there be?
-- Finding a remailer to use. ditto
I specifically made this a separate item because it has a different solution. Let's assume the potential user has some beginner's document about remailers. How do they go about finding out what remailers exist? Well, the document could have a list of them, but that doesn't exactly work well in the face of rapid changes. Some centrality in the initial query seems called for. That could be a stable machine, or some stable name, even. What the query actually looks like is less important. We need DNS or something like DNS for this purpose. We need something where changes can propagate outward rapidly, which pushes data out, and unlike BIND (the standard implementation of DNS), which pulls it in after it times out. The standard DNS query format could be kept, but the current back end may not quite work. And what about users on Compuserve, AOL, Genie, Delphi, and Prodigy?
-- Deciding what remailer to use. ditto (remailer server should list remailer properties like keylength, private?, delay length, chaining?, mixing?, padding?, encryption required? etc)
Certainly a standard way of listing the properties of a remailer would help. This seems to be mostly a matter of syntax. There is, also, the question of trustworthiness. That mythical beast the reputation system might be applicable, but I know of none to judge for suitability. More generally, there are questions of policy. What, for example, is the policy of the remailer in case of administrative request for mappings? Are there liquidated damages available to someone whose privacy is breached? These legal issues are not so easily made into syntax.
-- Figuring out how to use a particular remailer. standardize remailer help system, standard remailer command format (but not neccessaily the commands themselves) Sorta like an SGML for remailers
I think the commands ought to be standardized, just like RFC-822 standardized on the To: field. I realize this is going to create a little havoc for the half-dozen or so remailer developers who have all chosen not to talk to each other during their developments. If you don't have standard commands, then you need a way of specifying semantics for all these various commands. Not good.
-- Formatting a message for a remailer. see above
Personally, I don't think we need multiple algorithms for this. Is there any compelling reason, other than to avoid wasting existing but not yet deployed code?
-- Receiving mail through a remailer. Get/Creating a nice client.
There's a transaction cost to switching clients which is huge. It's completely unrealistic to expect everyone to use a particular client for remailers. It just won't happen. Far better is to rework existing clients to support remailers and to get those changes into the main distributions.
Reducing complexity cost: All of this could be lowered by creating an easy-to-use remailer client which is compiled (or perl/tcl interpreted) and installed with every unix out there so it becomes ubiquitous.
The dream of universal software. When I can unpack some software and type 'make', and do nothing else except read the man pages that 'make' caused to be formatted, I'll call that universal software. And not before. I'm glad lowering these transaction costs garnered a response. But what I really want to see is, what did I forget about transaction costs to use remailers? Eric
participants (2)
-
hughes＠ah.com -
rjc＠gnu.ai.mit.edu