A    As users, concerned with confidentiality and legal professional privilege.

B    As advisers to the private citizen, concerned with

• Privacy, as part of the liberty of the subject
• The powers of the State, as expressed in the criminal law.

A    As users: confidentiality, legal professional privilege

It protects:
 
• Communications to and from the lawyer seeking or giving legal advice
• Documents prepared for litigation
Its purpose:  promotes fair administration of justice

Its special status is reflected in the law on powers of search and interrogation, where special judicial scrutiny may be required before access can be granted if privilege may apply, and where privilege may protect from interrogation.

Quote from City firm's Electronic Mail Policy
We wish to encourage wide-scale use of the email for communicating with clients and others. However, because of fears over email confidentiality, we should not use email to communicate with a client regarding the client's affairs unless the client has agreed to communicate by email. We can take the client to have agreed this if the client initiates email communication or requests it. In other cases, we should check with the client first. As when sending confidential faxes, before sending an email with confidential material we should check that the client is happy to receive it in this way. In some companies, email may pass through an intermediary before reaching the intended recipient (rather like a fax may go through general office). It is inadvisable to send highly confidential (for example, price sensitive) information by email without using some form of encryption.

Given the low risk from loss of access to communications (as compared with stored data), since communications can be repeated, there is little benefit from key recovery or key escrow services.

Given that there is little benefit from key recovery or key escrow services, there is no possible justification for lawyers exposing private confidentiality keys to risk of third party access.

In any event, access to private confidentiality keys (whether through escrow or under a warrant) is plainly excessive, because it gives access to all communications to the owner of the key, not just to messages where this is justified by law enforcement requirements.

The proposed legislation requires scrutiny in due course to ensure that:

• the special status of privileged communications ensures that judicial authority is obtained before access under warrant can be demanded
• the person to whom the warrant is addressed is given enough time to respond to it to enable an application to be made to the court to rescind the warrant if appropriate (section 55 of the Drug Trafficking Act 1994 offers a suitable model)
• the provision of plaintext or a session key, and not a key giving access to all communications, is sufficient to satisfy the warrant

B    As advisers to the private citizen

In the context of crime prevention and detection, this offers two significant advantages:
 

• It reduces opportunities for fraud
• It enables the police to have secure communications with informants

What about secure communications between criminals and honest citizens (banks, doctors, travel agents, car hire firms)?
 

A warrant is appropriate to justify a bank or doctor releasing confidential information.  For a travel agent or a car hire firm, where the information is hardly even confidential, a warrant is almost more than enough.

None of these cases justify a key escrow requirement, because access to keys provides secret access to all messages to a particular recipient, and not just those messages where it is justified.  This is clearly unacceptable in the case of doctors and banks, and an undesirable invasion of privacy even in the case of travel agents and car hire firms.

It is not inconceivable that a case could be made for secret online law enforcement access in very special circumstances.  No such case has been made.  No serious attempt to make that case has even been made.  Secret online law enforcement access should be opposed until there is hard, clear, public, convincing evidence to justify it.  The anecdotes of urban myth are not enough.

 As previously mentioned, legal professional privilege must justify objection, there must be time to object, and provision of plaintext or a session key must be sufficient.

The legal consequences of failure to respond to a warrant require scrutiny:  what if a key is lost or corrupted (or claimed to be)?  What if there is a flat refusal.  Is there to be a criminal offence?  Are adverse inferences to be drawn (and if so, of what)?  Note that a warrant makes lawful what is otherwise unlawful, whether the breaking of a door or a confidence:  what is sought here is not a warrant but a power of interrogation.  Such powers exist (for example the powers of the Serious Fraud Office under the Criminal Justice Act 1987, or those of Inspectors appointed by the DTI under the Companies Acts):  but there are strict limits on the scope of these powers and what use can be made of material obtained through them.

C    As advisers to commerce and industry

Contracts can be made by word of mouth, by gesture (as in auctions), by telegram, by telex, by fax:  it is simply absurd to doubt that they can be made by electronic mail.

Some things have to be done in writing; for example:  guarantees; wills and codicils; tax returns; and transfers of property such as debts, copyright, patents and land.

"Writing" is generally defined too narrowly to include electronic documents (Interpretation Act 1978, Schedule 1).  Oddly, the definition in section 178 of the Copyright, Designs and Patents Act 1988 is wider:  copyright can be assigned by an electronic document, but a patent cannot.  There is no sense in this distinction.

The problem of the unsatisfactory definition of "writing" is well known, easy to understand and easy to remedy.  It is disappointing to find it not even mentioned in the DTI Statement.

A signature is any mark used to show adoption of a document.  Rubber stamps, mechanical signatures, names typed in telexes and names in faxes have all been accepted by the courts.  There is no real reason to doubt that a digital signature is a signature; but legislative acknowledgement could smooth the path of acceptance.  The draft EU Directive, Article 5, identifies the requirement plainly and clearly.

The central legal problem of electronic commerce lies not in acceptance of digital signatures but in their revocation.  Given insecure operating systems and the ease with which weak passwords can be guessed, there is a real risk of unauthorised access to keys and consequent undetectable forgery.

If a bank honours a forged cheque, it carries the loss however good the forgery and however careful the bank.  But for any form of signature other than handwriting, banks use contract terms to transfer the risk to the customer.  Other businesses will almost certainly adopt the same approach to digital signatures.  So the owner of a key will carry the risk of forgery if there is unauthorised access.  This should be seen as a major obstacle to the consumer confidence on which the development of electronic commerce depends.

It therefore seems essential for the user to have a reliable means of revocation.  This might be by direct communication to anyone to whom the user has given the public key; but what about those who have retrieved the key from a key server?  A highly reliable central service accessible online to all users for automatic checking might be considered, backed by some legal basis enabling the user to be certain that notice of revocation to that service will bind all other users.

Users cannot be expected to carry the risks of key compromise unaided given the insecurity of current systems.  If commerce and industry were to unite round a code of practice limiting a consumer's risk to a small fixed amount (as is the practice for ATM cards), users might have the necessary confidence to use digital signatures, and commerce and industry might have the necessary incentive to support satisfactory means of revocation.

Alternatively, much more secure means are needed for the protection of private keys.  At some cost, smartcards using biometric techniques might offer a solution.  The solution is likely to emerge much sooner if commerce and industry bear the risks of insecurity than if they can leave those risks to be borne by consumers.

These important and difficult issues are likely to be important for the development of successful electronic commerce, but they do not seem to receive adequate recognition in the DTI Statement.

By contrast, signature certification receives altogether excessive attention, despite being fundamentally peripheral in its importance.

It is important to notice how few ordinary signatures are backed by any kind of authentication. Even for documents which are witnessed, or for those sworn before a Commissioner for Oaths or a solicitor, there is no authentication of the identity of the signatory or the deponent.  Numerous other examples can be offered:  the Land Registry does not know the signatures on the land transfers it registers; companies do not know the signatures on share transfers they register; the Registrar of Companies does not know the signatures of directors or company secretaries on the returns they file.

Only notaries commonly take pains to check evidence of the identity of those who appear before them, and the transactions in which they act represent a very small minority of all transactions.  To insist on the cardinal importance of general signature  certification for electronic commerce is equivalent to suggesting that all commercial agreements should be signed before notaries.  It really is remarkable that the regulation of a non-existent industry of strictly limited importance should have generated such widespread and heated debate.

It is impossible to account for this phenomenon unless one observes the advantage for law enforcement access to private keys which flows from a scheme where successful applicants for accreditation are obliged to impose on their customers a requirement for the deposit of private confidentiality keys.  Once again, the law enforcement tail is trying to wag the electronic commerce dog.

Fortunately the draft EU Directive limits the freedom of EU member states to pursue this objective.  The draft requires that digital signatures do not depend for their validity on a certificate, and that certificates do not depend for their validity on having been given by an accredited CA.  Article 5 of the draft deserves strong support for this insistence, and in general for the limits it places on the freedom of member states to pursue law enforcement objectives in the context of electronic commerce initiatives.  It is disappointing that the EU draft fails to recognise the importance of revocation issues, and this is one aspect on which efforts to improve it should be focussed.

Successful electronic commerce will depend on user confidence in its security.  Government efforts to bootstrap law enforcement benefits out of electronic commerce initiatives will undermine the very confidence that those initiatives need to enhance.  The present Government is still clinging to the wreckage of the last Government's scheme:  it should have the courage to strike out on its own.