Dear Damien Bruneau,
We offer a few comments, below, to the consultation process regarding the consultation Making a Will (Consultation Paper 231, 2017).
This response is not confidential.
- It is nearly 70 years since Maurice Wilkes designed and built the first stored-program-controlled computer in Cambridge, and thereupon discovered the surprising difficulty of writing error-free computer programs. [For which see Maurice V. Wilkes, Memories of a Computer Pioneer (Cambridge, Mass: MIT Press, 1985), 145 (cited in Electronic Evidence, 6.61)] The manifestations of this discovery have been with us ever since, never more so than today, when a torrent of software patches remains unequal to the task of keeping our software correct (and sometimes introduces new errors). As a result of the exploitation of such errors we have seen the technical secrets of major intelligence agencies strewn over the internet and the funds of major banks pilfered in large amounts.
- The computer industry has been well aware of its difficulty, and has indeed deployed a formidable armoury of legal techniques for reducing its exposure to liability for imperfections in its products. It justifies its approach on the ground that production of error-free software (and, be it whispered, error-free hardware) is impracticable.
- Perhaps surprisingly, these facts have left undisturbed a general intuition that when machines fail, they do so in an obvious way and often catastrophically. This intuition probably derives from the experience of the nineteenth century and the first half of the twentieth, when steam boilers exploded, bridges shook themselves apart in gales and ships struck icebergs and sank. As an intuition it served well enough in its time, but it is profoundly deceptive when it is applied to the subtle, hidden, exploitable vulnerabilities to be found in computer hardware and software connected by a global internet.
- We think that the Paper has not always been well served by the intuitions of its authors; and this background is also relevant to the problems of authenticating electronic documents.
- In the light of these preliminary observations we turn to consider electronic wills.
- The Paper suggests (at 6.12) a small number of advantages in enabling wills to be made in electronic form: that they would be easier to amend and thus keep up to date; that they would be easier to store; and that executors would find them more easily after the testator’s death.
- Virtually all professionally drawn wills are prepared in electronic form, and even if the will is executed on paper there is no obstacle to preserving the electronic form from which it was prepared if that is thought to assist preparation of a future will in amended form. Those who make their own wills will often do so in electronic form, perhaps after buying an electronic precedent, when the same consideration applies. Otherwise they will use printed will forms, and miss out on the convenience of having an electronic equivalent. Only those who today would buy a printed form, but who in a hypothetical future would choose to make an electronic will, will gain this benefit.
- Is it really a benefit? In our modest experience of making and remaking our wills, never once has a professional will-maker evinced the slightest interest in any earlier will, let alone a desire to follow its form after comparing it with the professional’s own up to date precedents to check what amendments of form are required. Would a client wish to pay the extra charge inevitably attracted by such an exercise, as compared with making a clean start with a new will in the latest form? It seems to us that of the few people likely to obtain the contemplated advantage, almost none if any would get the least benefit from it.
- Are electronic wills easier to store? Some effort must be put into looking after a will, whatever form it takes. Solicitors often store originals for their clients, providing them with one or more copies to keep as they please (which will usually identify the firm storing the original). Sometimes clients keep the original, together with any extra copies they want, leaving the solicitor with a copy. Solicitors and other will-makers may well keep the electronic version from which a paper version was prepared - storing electronic documents is not expensive. Solicitors’ accounting and other records, almost invariably in electronic form, make it easy to investigate whether a particular person made a will and where it may be found. In this context it seems not to make any discernible difference to ease of storage whether a will is on paper or in electronic form. If a client is the sole repository of a will in either form, there is a risk of loss from accidental destruction. Successfully moving materials from an old to a new personal computer is not free of risk. Ransomware can effectively deprive computer users of access to their documents. And users of the increasingly popular methods of full disk encryption can deprive themselves or others of access to their own documents if they lose their key or fail to pass it on to their families. People’s intuitions about how to safeguard paper documents are of much greater antiquity and a good deal more reliable than their often naive intuitions about how to manage their computers, and we do not find the suggested benefit of easier storage even faintly convincing.
- Executors looking for a will ex hypothesi know they are executors and that there is or was a will. They will certainly find it easier to search a computer than to search a house, but only if they can get access to the computer, which may be impossible. It is hard to see any real benefit in this supposed advantage.
- The Paper goes on (6.13) to imagine a system in which an electronic will is “stored ready to be submitted for probate in electronic form automatically and efficiently on the testator’s death.” It is not clear how this will be achieved (if testators will wear sensors to detect their deaths and initiate the submission of their wills, this is not mentioned). The Paper contemplates that this will produce great gains for testators and the whole probate system.
- The administration of an estate is not an especially quick process. Assets must be traced and valued, and some perhaps sold. Creditors must be identified and their claims checked. There may be negotiations about death duties. Compared with these processes, the submission of the will to the probate registry is a task of elementary simplicity, whatever form it takes. It is hard to see that much if anything can be gained by streamlining this minor if important task.
- For automatic submission to be workable, there must be some means for detecting whether an electronic will has been revoked, perhaps by a later will made on paper. Either there would have to be a system for registering wills and their revocation, or the making of an electronic will would have to preclude the later making of a valid paper will. Such requirements can hardly be reconciled with the maintenance of testamentary freedom, to which the Paper attaches commendable importance.
- Electronic documents (at least in the form of text documents rather than audio or video recordings of recited text) do offer one distinct type of advantage over paper documents, although wills are generally among the documents least able to benefit from it. We have in mind the use of hyperlinks and their derivative, tooltips. Where a document makes use of defined terms, for example, each use of a defined term could constitute a hyperlink to the definition, so that the reader could by the click of a mouse jump from the term to the definition, and then by another click jump back whence he came. Tooltips can display the linked text when the cursor is hovered over the linking term (sometimes inconvenient if the linked text is extensive and overlays the linking term while being viewed). Complex documents can be made much easier to understand by such techniques, and by others available only in electronic documents, such as maintaining several copies open on the screen at the same time (or on multiple screens), displaying different pages which it may be useful to have open at the same time for cross-reference.
- These techniques are at their most valuable in complex documents like legislative material or major project contractual documentation; and we suspect that the very great majority of wills do not call for them. They would hardly justify legislative change, especially because in the rare case where they were useful, they could be applied to an electronic copy of a will made on paper with no loss of advantage.
- (We cannot however forbear from pointing out how much irritation they would save readers of Law Commission Consultation Papers in electronic form who are at present forced to scroll down the page to read the footnotes ands scroll back to find where they left off.)
- Just to gain a fresh perspective, we asked a family member, an active user of social media aged under 40, for her thoughts on the future of wills in tomorrow’s world. She proposed the iWill, an artificial intelligence app which monitors the user’s social media activities and family and other communications, combining them with web-derived information about the user. The app uses this information automatically to create and update a will continuously, its dispositions reflecting the state of the user’s preferences and inclinations from time to time without user intervention. It takes effect on death automatically. Unattractive as we find such a notion, the intuitions on which it is based seem no less plausible than those which have motivated the Paper.
- At 6.34 the Paper says
“Our starting point has been to recognise that, while the status of electronic wills is currently uncertain, it is highly likely that their use will become commonplace in the future.”
- At 6.80 the Paper says:
“Furthermore, stakeholders have indicated that there is little demand for electronic wills.”
- We think that by “recognise” the Paper means “assume,” and that the assumption rests on its intuitions (since no other basis is suggested).
- We have argued that the advantages of electronic wills are at best trivial. The advantages of the paperless office are very great, but despite some 40 years of energetic promotion by the computer industry, it remains mostly unachieved. The only motive we can imagine which might give rise to a demand for electronic wills would be that they somehow became fashionable. It is hard to see why this should happen, and equally hard to see why it justifies legislative anticipation.
- Clients seeking legal services fear delay, worry and expense. They are delighted if applications of technology reduce their incidence. They expect their lawyers to have telephones and use email and text messaging (and maybe still to have their old fax machines). It is in our experience a matter of profound indifference to them whether their land transfer forms are submitted to the Land Registry on paper or in electronic form, or their tax returns likewise to HMRC, or their litigation documents likewise to the Court Service.
- If there were a way of executing an electronic will which was discernibly simpler than executing a paper one, it might have some attraction. If one thing is clear from the Paper’s discussion of the choices of how to do so, it is that it will not be simple. Some means will have to be found for associating the electronic signatures of the testator and the witnesses with their makers. In the case of biodynamic signatures, which have the attraction of providing good biometric evidence, but have not caught on in practical use in Europe (although used in the United States of America by banks), a body of samples would have to be provided by each signatory and securely stored by some trusted person so as to be available for future evidential purposes. In the case of digital signatures, the signatories would have to obtain key pairs and certificates for them, and the verification keys and certificates would have to be stored for the long term with the identity evidence linking them with the relevant signatory. Apart from the inconvenience attendant on these steps, there would be the expense of engaging the services of the third parties involved, which we think would not be trivial. (In the case of digital signatures we think it is one of the reasons why, in the absence of a requirement to use them by governments, they are not in general use and are never likely to be.)
- Paragraphs 6.79 and 6.80 discuss the point that in the case of a digitally-signed will, as compared with a contract, there is no counterparty to check the validity of the signature made by the testator (or the witnesses). It is not clear why the Paper sees this as a problem to be solved, saying that “In a will-making context, there is no natural counterparty and one would have to be engineered.” We suspect that this may reflect a misunderstanding either of the functions performed by Certificate Authorities (CA), or of how digital signatures are made and may be verified.
- When a CA is asked to provide a certificate for a public key, it checks that the person requesting the certificate has control of the corresponding private key by requiring them to sign some arbitrary data so that the CA can verify it using the public key. The CA also checks that they have the same name as the certificate will link to the public key. (It should be noted that the CA cannot certify that the person in question has sole control of the private key, as it cannot know that.)
- When a digital signature is made by the use of the signatory’s private key, it is complete when it is made. It can be verified by anyone in possession of the relevant public key, either then or later, and whether they are a counterparty or not. There appears to be no need to “engineer” a counterparty to a will. Will-makers would probably wish to verify a digital signature on a will to ensure that no errors had been made, and eventually so would the probate registry.
- CA certificates are issued for a defined period, after which they expire. The public key contained in an expired certificate will continue to verify the signatures made with the corresponding private key, but verification software may be expected to issue a warning, more or less delphic, about certificate expiry. Verifiers, such as the probate registry, may be expected to reject as invalid wills whose signatures cannot be verified without provoking such a warning. The result would be that testators and their witnesses would have to continue to renew the certificates for their keys as they expired, and bear the resulting expense. The death of a witness before the testator might lead to difficulties in continuing renewal, as might the expiry of the testator’s key between the time of his death and the verification of the signatures by the probate registry. Errors in renewal might also lead to problems. The proposed dispensing power might be invoked to cure the perceived defects, though at some expense.
- Another problem which would impede and might frustrate subsequent recognition of a will as duly executed would be prior cessation of business by the CA which issued any of the relevant certificates, whether through insolvency or other causes. The primary revenue source of a CA depends on its certificates being recognisable by verifiers’ computers; and this depends on the willingness of computer makers and browser suppliers to include in the computers and browsers they distribute the necessary means for recognising the certificates of that CA. If a CA makes sufficiently serious errors in the issuing of certificates, computer makers and browser suppliers may treat it as untrustworthy and exclude its certificates from recognition. The result would be immediate insolvency, as was suffered by the Dutch CA DigiNotar in 2011 (for which see Electronic Signatures in Law, 15.15; the Wikipedia entry under ‘DigiNotar’, https://en.wikipedia.org/wiki/DigiNotar and ‘History of Risks & Threat Events to CAs and PKI’, illustrating the risks to CAs and SSL certificates, which are little understood but heavily relied upon, at http://wiki.cacert.org/Risk/History).
- Customers whose certificates’ validity was thus rendered unverifiable might not be notified, or might fail to understand the significance of any notification they received, and so fail to obtain new certificates for their public keys from another CA.
- Probably only a government could carry on a certificate-issuing operation with a sufficiently convincing guarantee of indefinite availability to meet the needs of testators, and such a non-commercial venture might be expensive for users unless subsidised from general taxation, which would seem difficult to justify. Governments usually undertake commitments of indefinite duration and cost only when essential; for example, to protect public safety in the context of the decommissioning of nuclear power stations. It is hard to see that the preservation of the validity of electronic wills falls into the same category.
- The need for certificate renewal would also prevent the adoption of a potentially valuable security precaution by testators and witnesses, namely the destruction of the private keys immediately after their use to sign the will. Such destruction would be by far the best safeguard against the risk of the private keys coming into the wrong hands through accident (disposal of an old computer without secure erasure of its contents) or malice (improper access to the computer by family members or carers, or malware infection). Certificate renewal might depend on using the private key to prove to the CA that it was still under the owner’s control, however. Accidental erasure of the private key could of course have similarly awkward consequences.
- As an aside, note also the problem illustrated by Stefanie Fischer-Dieskau and Daniel Wilke, ‘Electronically signed documents: legal requirements and measures for their long-term conservation’, 3 Digital Evidence and Electronic Signature Law Review (2006) 40 – 44.
- (Another misunderstanding about signatures is found in 6.74 and later paragraphs, in which signing a document is described as being accomplished by encrypting the document with the signatory’s private key, thus enabling those in possession of the corresponding public key to decrypt it and thus verify that it was encrypted by the signatory. Although the logic is correct, that is not what is done, because the consequence would be inconvenient: the document would have to be decrypted to be understood. What is done in practice is to append to the document (or include in a separate file) a hash or message digest of the document encrypted by the private key. Verification consists of decrypting the hash, making a new hash of the document, and checking that the hashes match. These facts would be obvious to the Paper’s authors if they had ever inspected a digitally signed document and its signature.)
- Wills in the form of audio or video recordings would have the potential advantage of providing evidence as to testamentary capacity, and also knowledge and approval, as well as intrinsic evidence that the testator was their maker. They also provide opportunities for forensic examination to detect tampering. In these respects they have advantages over text documents; but some of those advantages could equally be obtained by a video-recorded discussion between will-maker and client followed by the execution itself. The disadvantages of the recorded form is that it requires the testator and witnesses to carry out a performance, and presumably to repeat it as many times as necessary to do so accurately. The discipline of doing so, especially for a long will, would probably make the whole idea unattractively irksome to most clients. (It would be unsatisfactory for the recording to incorporate corrections, as the result might be uncertain; and the stage-directions of a master of ceremonies could not presumably be admitted to probate but could not be excised without giving rise to suspicions of tampering.) It is not clear whether the Paper contemplates that the oral statements of the participants would constitute their signatures; if not, the problems of executing the recording would apply as they would to an electronic text.
- For a will of any length and complexity, an audio or video recording would be distinctly inconvenient as a working document for administrative purposes, for which a transcript would probably be needed. But if the recording were based on the reading of a previously prepared text, then it would remain available with only the inconvenience of checking that the testator followed the text correctly. (Wills made extempore without a prior draft would surely be a rarity.)
- When a testator signs a will on paper (and the witnesses do likewise), they know what they are doing and how it works and can see the whole result of their acts. When the electronic equivalents take place, they have to trust their own or some other computer to be running the software that it is supposed to be running, they have to trust that it works as they have been told (in ways they probably do not understand) to produce the results that it should. The computer systems they are using have large numbers of unknown flaws which may be vulnerable to exploitation to their disadvantage in ways they cannot imagine. To take one example, they may have seen and checked the display of their will on the screen; but unknown to them, this may not be the document that is in fact signed with their signing key – the will-maker may have colluded fraudulently with their relatives to obtain their execution of quite a different document. To take another example, the will-maker’s computer may without his knowledge have been infected by malware which copies the signing keys used by the testator and witnesses and sends them to criminals who can use them to make a fraudulent will or otherwise misuse them.
- The potential disadvantages of the best known means for signing electronic wills can be seen to outweigh by far such benefits as might even optimistically be foreseen for them. We think the Paper’s assessment of electronic will-making is flawed by an inadequate depth of understanding of the technical issues and their potential consequences. Alexander Pope’s celebrated admonition (from An Essay on Criticism) seems to us in point:
“A little learning is a dangerous thing;
Drink deep, or taste not the Pierian spring:
There shallow draughts intoxicate the brain,
And drinking largely sobers us again.”
- In short, we think that electronic will-making will not catch on (as may be the case in Nevada), and that this would probably be for the best. Legislation, especially conferring delegated legislative powers, is dangerously premature.
The Dispensing Power
- At 5.98 the Paper says:
“Our tentative initial view is that the scope of any dispensing provision in English law should be drawn widely. Were a dispensing power to be introduced, there are strong arguments that it should apply not only to traditional written documents, but also where testators express their testamentary intentions in an electronic format, as well as in an audio or audio-visual recording. Those records are all potentially strong evidence of the testator's intentions and, as we have seen, upholding the testator's intentions is the strongest argument for introducing a dispensing power.”
- We support the introduction of such a power notwithstanding what we say above about the drawbacks of electronic wills, precisely because of the scrutiny that such special cases will attract and the clarity of the evidence which in practice is likely to be required.
- The words “where testators express their testamentary intentions in an electronic format” raise a concern, however. A will is the expression of a testator’s testamentary intentions, of course; but a letter of instructions to a will-maker might be similarly so described. The difference is that the former is meant to be dispositive while the latter is not. We think the dispensing power should apply to things intended to be dispositive, and not to those merely intended to indicate intentions.
- Paragraph 6.27 cites, in footnote 22, the comments by Stephen Mason regarding the unique link to the signatory in the EU Directive. The EU legislator, it seems, agreed with his criticism, and the definition was altered in the EU Regulation. This is covered in the 4th edition of Electronic Signatures in Law, 4.14 – 4.33.
Comments to specific questions
Questions 9 and 10
a. We provisionally propose that the code of practice should apply to those preparing a will, or providing an assessment of capacity, in their professional capacity.
- Should a change in the law take effect, the preparation of a Code of Practice will be useful, and ought, if agreed and produced, to be a document that a judge is required to refer to in determining relevant issues in the event of a dispute. If such a Code is agreed, it should make clear that the lawyer must be sufficiently competent in electronic signatures and electronic evidence. Unfortunately, in our experience the vast majority of lawyers cannot be considered to be competent in either of these topics at present. Given that we now live in a digital age, it is incumbent on professionals to ensure they are sufficiently competent to advise lay clients.
- Note Nicholas Bohm, ‘Watch what you sign’ 3 Digital Evidence and Electronic Signature Law Review (2006) 45 – 49 for an indication of what a solicitor’s understanding ought to cover; the article includes specimens of digital document signatures and includes observations on some practical problems of current digital signature methods.
- We provisionally propose that provision for privileged wills should be retained, but should be confined in its scope to:
(1) those serving in the British armed forces; and
(2) civilians who are subject to service discipline within schedule 15 of the Armed Forces Act 2006.
- We agree that is should remain, especially in the light of the possibility that such wills may be made in an amateur way electronically on the battlefield [although when Stephen Mason was in the army 1973-1982 and before going to Northern Ireland in 1976 on a bomb disposal tour, soldiers were required to write a will]. If a paper will is written, a service person might decide to amend it electronically. There are such cases mentioned with manuscript signatures in letters in Electronic Signatures in Law.
- We provisionally propose that:
(1) an enabling power should be introduced that will allow electronically executed wills or fully electronic wills to be recognised as valid, to be enacted through secondary legislation;
(2) the enabling power should be neutral as to the form that electronically executed or fully electronic wills should take, allowing this to be decided at the time of the enactment of the secondary legislation; and
(3) such an enabling power should be exercised when a form of electronically executed will or fully electronic will, as the case may be, is available which provides sufficient protection for testators against the risks of fraud and undue influence.
- No currently available hardware and software exists which could support electronic will-making in a way which provides adequate security or offers tolerable convenience at moderate expense, and there is no reason to foresee or anticipate its development. Two articles that are relevant:
Nicholas Bohm and Stephen Mason, ‘Identity and its verification’ Computer Law & Security Review, Volume 26, Number 1, January 2010, 43 – 51
Stephen Mason and Timothy S. Reiniger, ‘“Trust” Between Machines? Establishing Identity Between Humans and Software Code, or whether You Know it is a Dog, and if so, which Dog?’, Computer and Telecommunications Law Review, 2015, Volume 21, Issue 5, 135 – 148
- Parliament has at present no basis for reaching a conclusion about whether a move to electronic will-making would in principle be wise or not in the face of the obvious risks, and should not delegate legislative powers on such a sensitive subject: provision for parliamentary control of delegated legislation provides a wholly inadequate opportunity for proper investigation as compared with the process of primary legislation.
- In addition, there are two interconnected issues:
a. Establishing that the person is who they say they are (this might also be considered to be relevant for the witnesses as well), and providing evidence to that effect.
b. Providing an electronic signature.
- It is probable that people qualified to prepare wills are required to ensure they are dealing with the person who asserts who they are, but this is something that judges have had to consider in the case law relating to wills signed with electronic signatures (for which see Electronic Signatures in Law). For instance, in France, lawyers carry a card with two ‘signatures’ – one serves to authenticate them and the second is a for the purpose of signing.
- Regarding remote signatures and witnesses, this has already been achieved in the United States of America with a notary in respect of real property, for which see:
Timothy S. Reiniger and Philip M. Marston, ‘The deed is done: on-line notarization becomes a reality’, 10 Digital Evidence and Electronic Signature Law Review (2013) 144 - 146.
- We provisionally propose that electronic signatures should not be capable of fulfilling the ordinary formal requirement of signing a will that applies to both testators and witnesses (currently contained in section 9 of the Wills Act 1837).
- Response: We agree.
- We ask consultees to provide us with their comments on, or evidence about:
(1) the extent of the demand for electronic wills; and
(2) the security and infrastructure requirements necessary for using electronic signatures in the will-making context.
- We are not aware of any demand. If there were any, it would be unlikely to survive in the face of the inconvenience and expense likely to be entailed by electronic execution.
- Regarding the security of using electronic signatures in the context of making a will, it might be useful to consider the issue from the perspective of future litigation. Where a will was signed electronically using a software product, part of the evidence will be (or ought to be) evidence from appropriately qualified witnesses regarding the system used.
- Given the possibility that some form of digital signature system might be used – which in itself is not an objection – care must be made to ensure that everyone understands:
a. The technical concept of non-repudiation (dealt with fully in Electronic Signatures in Law, 16.14 – 16.26). The problem is, that many lawyers writing on the topic of electronic signatures, especially digital signatures, follow the technical assertion that if a digital signature is used it means that the person whose signature it is, actually causes the signature to be affixed. This is not correct.
b. The need to safely store the private key, as mentioned in paragraph 6.85. The Russian banking cases illustrate the problems of failing to provide for the proper security of passwords and keys, for which see:
Case A12-3342/05-C11. The Federal Arbitration of the Povolzhsky District. Plaintiff: Sovtransavto – Volzhsky Limited Liability Company. Defendant: KOR Joint-Stock Commercial Bank and Volgoprombank Open Joint Stock Company, 4 Digital Evidence and Electronic Signature Law Review (2007) 83 – 85
Olga I. Kudryavtseva, ‘The use of electronic digital signatures in banking relationships in the Russian Federation’, 5 Digital Evidence and Electronic Signature Law Review (2008) 51 – 57
Olga I. Kudryavtseva, ‘Case Note: Russian Federation’, 5 Digital Evidence and Electronic Signatures Law Review (2008) 149 – 151
- Note also the discussion of the burden of security and integrity in Electronic Signatures in Law and the case of Robb at 16.57 – 16.64.
- For additional case law on electronic wills and further articles, see Electronic Signatures in Law, 10.49 – 10.66.
Stephen Mason and Nicholas Bohm
14 August 2017
Electronic Signatures in Law
Edited by Stephen Mason and Daniel Seng
Available now from OBserving Law in the Humanities Digital Library