From: Adam Shostack <adam@bwh.harvard.edu>
Date: Wed, 24 Aug 94 19:40:55 PDT
To: shamrock@netcom.com (Lucky Green)
Subject: Re: Anonymous questionnaires
In-Reply-To: <199408242201.PAA05838@netcom7.netcom.com>
Message-ID: <199408250230.WAA14274@bwh.harvard.edu>
MIME-Version: 1.0
Content-Type: text/plain


Lucky:

| You wrote to my question about anonymous questionnarires:
| 
| >        Correlation is easy; assign people consecutive numbers or
| >somesuch.  If both participants are anonymous, no problem.  Could you
| >bring by the questionnaire by hand, in exchange for cash?
| 
| No, they are too far away.
| 
| >If not, how
| >about a money order and a PO box?
|
| They would still need my name.

	I think its Duncan who's been talking about secured credit
cards.  As long as the intent is not to defraud, you can call yourself
whatever you want and its legal.  So create a temporary psuedonym.

| >        The audit part of this is the tough part.  Would the HHS care
| >to agree to a broadcast means of verification?  Would participants 44,
| >71 and 94 please come into the re-testing center to verify their
| >participation?  There could be a zero knowledge proof of some type to
| >demonstrate that you are really patient 94.
|
| Can someone suggest a way to accomplish all this?

	With tools that exist no less.  Each participant gets a penet
account, and agrees to maintain it for (some time period).  Part of
their payment is withheld as assurance that the account will be
maintained.

	Each participant chooses a passphrase, and feeds that to
S/key, providing the 100th md4 hash of their secret passphrase.  They
enclose this number in their encrypted response form.  They also
enclose their participant ID #, and an address.

	When HHS asks for verification that the participants were
real, they select a random set of penet IDs, and mails each of them,
asking that they show up, bringing a lawyer to protect them from HHS
intimidation and the 99th md4 hash of their secret passphrase.  They
have thus demonstrated who they are, and can answer questions as they
feel relate to them not being in colusion with the researchers.

	I think that deciding whether or not this is useful would
require reading the regulations imposed on the researchers by the
agency administering the grant.

Adam