Memorandum by Professor Harold Thimbleby,
Middlesex University, London
What needs to be done to create confidence and
to stimulate e-commerce?
1. Regulatory uncertainty and personal worries
over privacy are factors undermining confidence. Personal worries
over privacy focus much debate underlying regulatory concerns.
Trust needs to be encouraged between e-commerce parties (on short-term
issues) and between investors and regulators (on long-term issues).
2. Abuse of personal privacy, say, by using
cryptographic techniques to conceal unlawful activities, is of
concern. Unfortunately the techniques that can be used by criminals
are techniques that can be used fore-commerce. Inevitably, companies
that sell CDs securely will be able to sell seedy products securely.
3. Keep regulation in perspective. Although
roads can be used by both good and bad people, good drivers are
only very exceptionally hindered by road blocks. Unlike roads,
e-commerce is based on new technologies that have not been fully
developed. Imposing "road-blocks" is likely to be premature
4. Privacy is not an issue so much as building
up trustas it were, most effort should be put into encouraging
good driving rather than concentrating on bad drivers. When parties
can trust each other, intrusions of privacy are secondary. Trust
can be built up by degrees, and depends on (i) the certain identification
of the parties concerned (ii) the ability to locate the parties
in the future (iii) the possibility of sanctions that can be imposed
outside of the relationship. In the physical world, these factors
are taken for granted. In the virtual world, we have become accustomed
to anonymity. Thus hackers can easily cause havoc that cannot
be traced back to them: they have no identitiesand this
is what vandals conventionally relish. Trust is a new concept
on the Internet.
5. Because of widespread anonymity (eg where
do cookies come from?) people tend to over-react and emphasise
privacy as a protection. Partially; but identity is required.
If we knew where the cookies came from, and could in principle
get back to their originators, impose sanctions on them, and so
forth, trust would be built up.
6. Identity can be ensured within any nation
state by a physically-based registry (eg employing key escrow
connected to IDs). The Internet is far bigger, and solutions that
depend on single nation, centralised, databases will be of limited
value. Moreover political culture from nation to nation will undermine
each nation's "rational"but culturally specificsafeguards.
7. Strong cryptographic techniques, such
as zero knowledge proof (I can prove I know something without
revealing the knowledge) and key exchange (I can exchange keys
with no third party knowing them) require the use of methods that
are subject to regulation. It should be said at once that these
technologies are at their early stages of development.
8. Whilst debate focuses on privacy, the
technologies that are required for trust will not be developed
(nor will experimental solutions be tested).
9. The Internet (as currently based on TCP/IP)
is inadequate; regulation based on its absence-of-trust mechanisms
will be temporary. Technologiesnot just legislationrequiring
CPU IDs in packets, tamper-proof hardware, and so forth will be
required. Without tamper-proof hardware it is easy to confuse
trusted parties for trusted computers. So faroutside of
very special situationswe have no trusted computers, and
hence no trusted parties.
7 March 2000