By Iain G. Mitchell QC
Nothing new under the sun
What has been is what will be, and what has been done is what will be done; there is nothing new under the sun. [Ecclesiastes 1:9]
Those words from Ecclesiastes might seem an odd way to start an article on the latest thing in digital connectivity: the new way of building trust that has powered the rise (and fall, and rise again) of digital currencies, promising to build trust and utterly transform markets of all sorts, including the art market; but there is the eternal truth that human nature does not change, however much technology may transform the way in which we do business.
Currency as a medium of exchange supplanted barter and provided a store of value which was liquid. Double entry book-keeping gave an ability to record and understand transactions. In the seventeenth century, the founding of the Bank of England and the creation of the national debt – lending to the government which the investors did not expect to be repaid, provided that the government continued to service it – unleashed a huge tide of investment. One can think, no doubt, of other examples, but what lies at the heart of all of them is the notion of trust. You will not do business with another person unless you have trust. It might be trust in the other person – I should not want to lend to the government unless I trust that the government is be able to service the debt - or it might be trust in the medium of transaction – I shall sell you my chickens in the market place not so much because I trust you, as because I trust the coin which you will give me as I hand you my chickens – or it might be trust at a more abstract level – I can risk making an investment in your business (or you might risk investing in my business) because I know, through double-entry bookkeeping just what my net worth is.
If trust is impaired, business is, at best, inhibited and, at worst impossible. It is no coincidence that the national economies which prosper are those which have in place legal structures which tend to engender confidence and in which the judges are not corrupt. In the virtual world, where you cannot see the person with whom you are transacting, nor visit him in his shop, the issue of trust assumes particular importance. In its Consumer Conditions Scoreboard, The European Commission sees the building of trust as essential for unlocking the potential of e-commerce in the EU.
It is arguable that the art market has not yet achieved its potential for building a trusted space in which to do business. The authority of catalogues raisonnées and the personal reputation of dealers help build trust, but there lurk in the background, to a greater or lesser extent, areas of doubt over authenticity and provenance amongst other concerns.
What blockchain technology can potentially bring to the party is the building of trust to create conditions in which markets can flourish. However, if we lack a clear idea of how blockchain technology actually works, a discussion as to its potential benefits and risks can result in little more than a utopian wish list – a triumph of hype over experience.
The Technical stuff
One of the great frustrations of reading about blockchain is that many of those who set themselves the task of explaining it tell you what they believe it does, rather than explaining what it is, and often what they think it does is received wisdom, leading their expositions to founder on the Scylla of over-simplification. Others, who do understand what it is, often presume on the part of a general readership a level of familiarity with what might appear to be arcane technical concepts which such a readership does not possess: anyone for Byzantine Fault Tolerance? Their expositions thus founder on the Charybdis of incomprehensibility to all but fellow experts. Neither approach really facilitates a consideration of the benefits nor an appreciation of the risks involved in the use of blockchain technology.
First, then, what is blockchain?
It is a distributed open ledger which exists in digital form. But what does that mean?
Essentially, it is a database which contains a record of transactions, for example, that A transferred an asset (cash, a container load of tuna, or whatever) on a given date and time and at a particular price. That is, essentially, the sort of information which people have been recording in ledgers almost since the first cuneiform characters were inscribed on the first clay tablets in ancient Nineveh. Of course, we don't use clay tablets any more. Some of us still use paper ledgers, but, mainly, we use spreadsheets which are maintained in digital form.
Now, a merchant keeping a paper ledger, or a spreadsheet, will keep a closed ledger, which is to say that only he can write in it, and he may or may not want to let a potential customer look at it. So, the issue of trust arises. Can you trust what the merchant has written in his ledger?
This is where the idea of an open ledger comes in. It is a ledger which is potentially open to everyone both to maintain and to consult. However, just because everyone can see it, does not make the information recorded in it any more trustworthy. This is where the idea of a distributed open ledger becomes important.
The ledger does not exist on a single computer: it is distributed amongst all of the computers, or nodes in the network. This means that there is no single “authentic” version. All of the nodes in the network store the ledger and all iterations of the ledger are the same, and, therefore, equally authentic. For this to be achieved, all of the iterations of the ledger require to be synchronised with each other, so that if a change is made in one iteration, all will show that change. Therefore, for the distributed open ledger to be maintained, constant interconnectivity is required.
This means that if, say, a painting is sold in Lyon & Turnbull at a particular sale on a particular price, that sale is entered in the distributed ledger and appears in all synchronised iterations. If there is a later sale of the same painting, that also shows in the ledger. The utility of this in terms of transparency and the recording of provenance is obvious. So far, so good, but there are obvious problems if that were all there were to it, such as the risk of false entries being made or retrospective changes being made to genuine entries.
With a conventional centralised database, you have to trust the person, for example the auction house, which maintains the database, but you also have to trust the robustness of its defences against hacking and other unauthorised alterations in the record. With large and reputable actors, trust in the person may not be so much of an issue, but there have been so many high profile and successful hacking attacks that there is a lingering impairment of confidence in the system – the very issue highlighted, in the context of e-commerce, by the EU Commission in the paper referred to above.
It might be thought that the same issues would arise to a more pronounced degree with a distributed open ledger, not least because of its open nature – Surely anyone could write in, or delete data from the ledger? In the event, this is neatly circumvented in the case of blockchain technology by a combination of the use of encryption and the very architecture of the system itself.
Blockchain employs a distributed ledger in which the data is grouped in blocks, which, upon reaching a certain predetermined size are then chained (“hashed”) to previous blocks. The ledger is distributed across all the computers, or nodes, in the network. With a “permissionless” ledger, anyone may join a node simply by downloading and running the relevant publicly available software. This is the model on which bitcoin operates: a permissionless distributed open ledger which uses blockchain technology. Not all ledgers are permissionless. It is perfectly possible (and, indeed, not uncommon) to have a virtual private network which runs on blockchain technology.
The ledger relies on a two stage verification process with asymmetric encryption. Every user has a public key, which is shared with others to allow transactions to take place, and a private key which is not shared. The private key has a mathematical relationship with the public key, as a result of which the private key is able to decrypt data which has been encrypted through the use of the public key. The data which is added to the ledger does not necessarily have to be encrypted, but unencrypted data can, of course, be read by anyone with access to the system, and because of the necessarily limited size of blocks, the storing of unencrypted data is not generally practicable. Accordingly, the usual model for a blockchain is that data is stored in the form of a plain text header, containing a timestamp and the identity of the data's source with the rest of block content (sometimes referred to as the “payload”) being encrypted.
In a blockchain network each node (or, in some networks, only certain designated nodes) act as “miners” - a reference to the first use of the technology by bitcoin. It is the function of the miner to aggregate data so as to create a new block which is then hashed to the chain. As a result, a chain can end up being comprised of a long series of blocks. The hash is cryptographic. Such a cryptographic hash is a one-way function which cannot be reverse engineered. In other words, once a block has been hashed to a chain, it cannot be altered or deleted. By this means, the problem of retrospective alteration, referred to above, is circumvented.
However, what of the problem of false data being appended in the first place?
This question is essentially the same as a well-known thought experiment in the field of cryptography known as The Byzantine Generals problem. Suppose that a Byzantine army is surrounding an enemy city. The army is split into four divisions, each of which is under the command of a general. The divisions are separated from each other and the generals in charge of each division require to communicate with each other through messengers, who may require to pass through enemy held territory. If all four divisions attack at the same time, the city will be taken, but if only three or fewer divisions attack, they will not be strong enough, will fail to take the city, and will suffer heavy losses. In order to ensure that any attack is co-ordinated, all the generals have to agree to co-ordinate their attacks: if one of the generals has not agreed, the others might attack not realising this. Therefore, they agree a protocol whereby a message, say “attack at dawn tomorrow” is transmitted by each of the generals to each of the others and repeated back. Unless all have repeated the message, there is no assurance it has been received by all. If all four generals are trustworthy, that is no problem, as they will each respond as they are supposed to, and all will know that all four divisions will attack simultaneously.
However, suppose that it is known to the generals that one or more than one of them is, or may be, a traitor, but it is not possible to say which. In that event, each of the loyal generals will relay the message that the other generals have ordered an attack at dawn, but the traitor generals will relay the message that the agreement is (say) to retreat. In that event, each of the loyal generals will have received contradictory commands, but they do not know who is loyal and who is a traitor. What are they to do?
They could do nothing unless there is unanimity, but the nature of the problem is that there is no unanimity. Accordingly, they require to develop a protocol for reaching consensus.
This is essentially an analogy for the situation in an open distributed ledger. The generals are nodes on the ledger network. In the case of the generals, as it happens, consensus on a reasonable plan to attack at dawn can be achieved if more than two thirds of the generals are loyal. Thus, in the example above, if two or more of the generals are loyal, consensus can be achieved, but if only one of them is loyal, there is a problem.
In the case of blockchain, rather than generals, there are several ways to achieve consensus. One is to design an algorithm which has a defined “Byzantine fault tolerance” to achieve a consensus amongst the nodes as to whether data inputted is accurate. This is the version which is the closest analogue to the thought experiment. However the protocol used in bitcoin mining is the protocol known as proof of work in which the process of hashing a block to the chain is effected through the solution by a miner node of a mathematical puzzle derived from the data involved. Such a puzzle can be solved only through the use of (in computing terms) brute force. The puzzle is offered simultaneously to all miner nodes and devised, so far as possible, to give them all an equal chance of solving it – the main variable is the level of computer resource that is put into solving it. In order to provide an incentive (and recompense for the cost of the resources deployed), the first miner to solve the problem is given a small financial reward. Though immensely difficult to solve, the solution is designed to be easy to verify, in this instance by the other nodes on the network. If verified according to the consensus protocol then the node's solution to the puzzle will constitute a “proof of work”, which proof is required to permit the block to be hashed to the chain.
There also exist other consensus protocols, for example, those using proof of stake, which is based on the use of digital signatures, but these are not further considered here.
In short, the consensus protocol which is used ought to eliminate the risk of false data being inputted to the ledger.
Characteristics of Blockchain
From the above discussion, it will be seen that a distributed open ledger using blockchain technology has the following characteristics:
Having those characteristics, such a ledger can be used to be used to trace assets. For example, the WWF (World Wide Fund for Nature) is presently running a test project to use a blockchain to trace tuna from where it is caught, down the supply chain to final sale to consumers.
Utopia next stop?
The benefit which clearly jumps out is benefit 3, improving provenance and reducing art forgery. Clearly, blockchain could be used for creating a central register on which a painting whose authenticity has been established is entered and each time it is sold, the details of the sale are added. This will unequivocally establish provenance from the first time it is placed on the register, going forward; though, of course, its provenance back to the original artist is only as good as its attribution in the catalogue raisonée. In the case of new art, however, a complete provenance can be established: the art work is placed on the register by the original artist and each sale is recorded. This also has the effect of increasing transparency as to price and, it is suggested, giving to the market the level of confidence required to grow the market and drive up prices.
Benefit 1, it is suggested, will allow artists to embed their digital works in the blockchain and thereby control the sale of copies up to whatever edition limit they might have set.
Benefit 4 flows, it is suggested, from benefit 1, facilitating a further payment to the artist on second and subsequent sales. This resale right, is, of course, a statutory right in the UK, but is not in other jurisdictions.
Benefit 2 might be the most intriguing. Works of art are already value stores. Art works can be held as investments, and sometimes are kept securely in secure stores, rather than exhibited on the walls of a home, office or museum. If the art work is recorded on the register, it is but a short step to regarding the entry as a form of (or at any rate, certificate of) ownership. It becomes, as it were, an analogue of a bitcoin, and can be split into shared ownership. Very few people in this world can find $450,000,000 for Leonardo's Salvator Mundi, but it is entirely possible for 450,000 people to find $10,000 each.
Are we, then, heading for an exciting new future for the art market?
Well, although the enthusiasts have some basis for their predictions, there's a few possible flies in the ointment. There is, first, the obvious point that no system involving human agency is ever fool-proof, however good the technology may be. For example, the proposals for tracking artworks usually involve the affixing of some form of tag to the painting, and tags can always be swapped, perhaps taken from the genuine article and attached to a copy; and I am sure that the would-be art criminals among you can think of other points of vulnerability. However, there are other, more profound systemic problems, some of which are discussed in a considered way in the Turing Institute's paper The Art Market 2.0 – Blockchain and Financialisation in Visual Arts.
Legal and Regulatory issues
Many enthusiasts tend to assume that all distributed ledgers running blockchain will be open, permissionless ones, facilitating a true democracy founded upon egalitarian values. A similar naïvety attended the early days of the internet. Kehoe in Zen and the art of the internet (1992) suggests, in relation to usenet groups, that signatures containing advertisements have become “the graffitti of computers” and are liable to cause those who do this to be flamed. One recalls the following quotation:
"We are assured that the world is becoming more and more united, is being formed into brotherly communion, by the shortening of distances, by the transmitting of thoughts through the air. Alas, do not believe in such a union of people. Taking freedom to mean the increase and prompt satisfaction of needs, they distort their own nature, for they generate many meaningless and foolish desires, habits, and the most absurd fancies in themselves.”
You might think the language a little archaic to describe the new technologies of blockchain, or, before that, the internet, and that is understandable as it was, in fact, Dostoevsky in The Brothers Karamazov describing the invention of the telephone.
What of course, Dostoevsky recognised, even if he did not approve of it, is that the gratification of meaningless and foolish desires is a perfectly sound business model. The meeting of needs and desires (whether foolish or not) is what underlies e-commerce, and builds a flourishing market. It is a characteristic of unregulated markets that their competition can become distorted with a tendency for enterprises to achieve dominance or even monopoly status. In the digital space, this is complicated by the way that the internet transcends national boundaries, enabling the rise of players like Facebook and Google and bringing the many issues with which we are familiar from our daily news. Distributed open ledgers are not immune from these pressures and tendencies.
As we have seen, distributed networks do not require to be permissionless. It is perfectly possible to run blockchain on a virtual private network, or to develop a register which is open, in the sense that it may be consulted by the public, but closed, in that to become a node on the network requires permission which might be restricted to, say, artist and art sale house “members”.
The creation and running of an art register running on blockchain technology is a perfectly feasible model. The core members would fund it through fees or subscriptions, and it could be made accessible to consult by anyone either on a fee-paying basis or entirely freely, but with advertising being delivered to those who visit, and with their data being gathered and monetised in the way that is now familiar with Facebook and Google. It is entirely possible that such a register could be set up and launched. The Turing Institute paper points out the risk of such an early entry player achieving dominance in the market for providing an art register. On the other hand, if that is avoided, does it really help the realisation of the benefits which are mentioned above to have a multiplicity of art registers operated by different companies and entities, possibly each incompatible with the other? Could the solution lie in the principal players in the art market co-operating to build a joint not-for-profit ledger?
The problem is not, however, purely economic. There arise substantial regulatory concerns arising from the trans-national nature of most distributed ledgers. There are, of course, the market regulation issues, hinted at above, but, more pressingly, it is questionable whether such a register is even possible against the presently-existing regulatory background.
The EU General Data Protection Regulation (GDPR), broadly forbids the processing of personal data unless certain conditions are met. The issue of the relationship between blockchains and the GDPR is fully discussed in the paper by Michèle Fink Blockchains and Data Protection in the European Union (Max Planck Institute for Innovation and Competition Research Paper No 18-01, 2017) but it is worth highlighting some of the issues here.
A distributed register running on blockchain has personal data everywhere. As described above, in a typical blockchain, the headings are unencrypted, and typically contain personal data, at any rate where the person whose details are shown is an individual rather than a corporation. Further, where data in the blockchain are encrypted that results in anonymisation of the data, but the threshold for anonymisation under the GDPR is that the data is processed so as irreversibly to prevent identification. Since encrypted data may be accessed by use of the private key, it still falls in the category of personal data for the purposes of the GDPR. Even with hashing of the data, which is not susceptible of reverse engineering, the view of the EU Article 29 Working Party on Anonymisation Techniques ( Opinion 04/2014) is that it is still possible to link the dataset with a data subject, with the consequence that, although hashing may achieve pseudonomisation, it does not effect anonymisation. Thus the data remains personal data.
The effect of the GDPR extends, however, beyond transactional data to the very architecture of blockchain. As discussed above, encryption is a necessary component of blockchain, and it is not possible for public keys to meet the requirement of anonymity as laid down in the GDPR.
This does not, of course, mean, of itself, that personal data cannot be processed so as to allow the distributed ledger to work, but it does mean that the personal data requires to be processed in conformity with the requirements of the GDPR.
The first issue which arises is the identification of the Data Controller, on whom obligations fall under the GDPR. The nature of a distributed ledger is that there is no single data controller, data residing upon and potentially being processed by any or all of the nodes, nor are the nodes joint controllers under Article 26(1) as the nodes do not “jointly determine the purposes and means of processing”. As Finck comments: “while a blockchain is fuelled by the interplay of various nodes, they don't determine the modalities of data processing of other nodes.” In these circumstances, it is unavoidable that every node is a data processor, which makes it very difficult to make processing on a distributed ledger compliant with the GDPR, since the GDPR assumes a centralised control agent.
The situation is further complicated by the likelihood that, with a distributed ledger, especially one dealing with an international market like the art market, the nodes could be anywhere in the world. Some, no doubt, will be in the European Economic Area, but others could be in the United States, or Russia, or Burkina Faso, or wherever. Are all of the data controllers equally susceptible to EU jurisdiction?
In fact, there is presently a case before the European Court of Justice concerning the “right to be forgotten”, Google v CNIL (C-507/17) which concerns whether a search engine operator is required to implement a de-referencing request in such a way as to make it impossible for the link to be accessed from anywhere in the world, and not just from within the EU. At the time of writing, the court has yet to advise, but the Opinion of the Advocate General has been published. It his view that the EU Treaties clearly apply amongst the member states and within their territory. Such exceptions to that principle as presently exist he treats as being truly exceptional and not a basis for extra-territoriality in the case before the court. He gives considerable weight to considerations of comity: extending the de-referencing right would, in effect, export EU data protection rights throughout the world and could give rise to a situation where observance of EU data protection law might compel a search engine operator to infringe laws of other jurisdictions, such as laws protecting free speech; further, this gives rise to the prospect of whether EU censorship of the global internet might encourage censorship by less benign polities who do not permit a free press and who might seek to impose similar global censorship in respect of, say, journalism. In effect, this presents the risk that such a move by the EU might lead to a “race to the bottom” for the global internet.
Of course, the context of a distributed ledger is different from that of a search engine, but it is difficult to see a way forward in light of the Advocate General's comments. Because the non-EEA data controllers (i.e. the overseas nodes) would be outside the EEA, the operation of the blockchain would inevitably lead to the exporting from the EEA of personal data by the data controllers who are subject to EU jurisdiction, contrary to article 44, unless those data transfers could be brought within the provisions of articles 45 to 49. Article 45 might come to the rescue where the overseas nodes are resident in third countries where the EU Commission has certified that those countries have standards of data protection comparable to those of the EU, for example, the United States under the Privacy Shield Agreement, but as for other countries, it is difficult to see what provisions of the GDPR could be made to fit in the case of a distributed ledger, other than consent of the data subject under article 49(1)(a). Such consent may be straightforward to engineer with a private distributed ledger, but may be more problematic with an open ledger. However, assuming that a system based on consent could be employed, that would have to admit of the possibility that a data subject would withhold consent, which might therefore impair the ability of the ledger to do its job.
This then moves us on to the question of, assuming a distributed ledger can be engineered which avoids the jurisdictional problems, just how the substantive obligations are compatible with both the fundamental architecture of blockchain and the existence of multiple data controllers. The Max Planck Research paper has a lengthy section on this, but suffice it to say that creation of a compliant internationally distributed ledger capable of permitting compliance by all data controllers with the substantive GDPR obligations is highly problematic.
This is an issue which is appreciated by the European Data Protection Supervisor which, in its 2017 Annual Report stated:
“It is essential that data protection experts begin to examine the concepts behind blockchain technology and how it is implemented in order to better understand how data protection principles can be applied to it. An integral part of this process should be the development of a privacy-friendly blockchain technology, based on the principles of privacy by design.”
It may, however, be that these comments wish the result without proving the means.
Blockchain technology is not as scaleable as it should be and does not come without environmental cost.
Blockchain typically can process about seven transactions per second whereas a conventional system has a much higher capacity. For example, Visa can process 56,000 transactions per second. Furthermore blockchain is highly energy intensive. The Proof of Work model of verification depends on a large number of nodes working over an extended period at maximum computing power. This puts a huge demand on energy supplies. Bitcoin alone accounts for 0.22% of the world's electricity consumption.
A distributed register recording art market transactions, of itself, is not likely to be seriously impacted by, or impact upon those statistics. It may well be that a processing rate of seven transactions per second is substantially more than is required, and its energy consumption demands are likely to be substantially less than those of bitcoin, though if blockchain is taken up in other markets and areas, and on a global scale, the environmental impact cannot be ignored. One possible mitigation measure would be to move to a Byzantine Fault Tolerance model, which requires substantially less computing power than a proof of work model.
The need for standards
If the way in which distributed ledgers develop in the art market is towards having a number of different registers, either competing or complementary or both, it might, sooner or later, be thought desirable to allow them to talk to each other. If they develop entirely independently, there may be incompatibilities which would limit or exclude interoperability.
Further, if the other practical and regulatory issues discussed above are to be addressed, then some form of co-ordination (for example, to provide distributed registers employing blockchain in a GDPR compliant manner) will be required.
These and similar issues underline the need for international standards for blockchain. The International Organization for Standardization is presently developing such standards, but these are not likely to be published before 2021. Additionally, the European Committee for Standardisation and the European Committee for Electrotechnical Standardisation are jointly consulting on their White Paper, Recommendations for Successful Adoption in Europe of Emerging Technical Standards on Distributed Ledger/Blockchain Technologies.
These projected standards relate to blockchain in general, and are not unique to its use in the art market, but plainly will have an immense impact upon it. For this reason, art market players should be heavily involved in the standard-setting consultation processes.
Further, because the use of blockchain has been mainly associated with bitcoin and other cryptocurrencies, and because those currencies have caused concern over their use in criminal activities and in money laundering, there is also a head of steam building up over a perceived need for wide and far-reaching regulation of blockchain. Such calls, though understandable in relation to cryptocurrencies are ill-informed when applied to blockchain in general, since blockchain is merely the technology on which cryptocurrencies run. This is a distinction well understood by the European Parliament, which, while calling for regulation of cryptocurrencies, believes that blockchain should be let be. However, this brings us back to the opening comments in this article that blockchain is poorly understood by the general public. Accordingly, those who wish to see the art market flourish through the use of distributed ledgers should be pro-active not only in explaining the benefits, but also in ensuring that the debate is conducted at an informed level.
Envoi – Where does the Art Market want to go?
So where does this leave us in relation to the effect of blockchain on the Art Market?
It is clear that if the quite substantial issues discussed above can be resolved then, indeed, the hoped for benefits of proof of provenance and transparency can be realised, leading to the boosting of confidence necessary to allow the art market to thrive and prosper. However, there's also contained in the aspiration of democratising fine art investment a tiny cloud, no bigger than a man's hand which may either dissipate, or may come to fill the sky and bring with it a breaking storm.
We have seen how a reliable distributed ledger can enhance the use of artworks as a store of value, since they can be locked securely away and be traded virtually. We have seen also how digital ledgers might facilitate art investment for the masses, with multiple owners each having their fractional shares securely recorded and protected. This could represent the final stage in a process in which investors care no more for the work of art as a thing of beauty than they do for gold bars in a bank vault, or, for that matter, company shares. Indeed, depending upon what arrangements the multiple proprietors might make for the management of their investment, one can figure that it may become impossible, or at any rate, very difficult, to secure agreement for the investment ever to be taken out of the vault and exhibited on the wall of a museum.
The end result is a kind of commoditisation of art, to a greater degree than we now arguably have, but a market which, at its top end, is not limited to a few wealthy individuals and corporations, but is available to everyone, through the mechanism of multiple ownership. Many who take the leap into such investment may know little and care less about art for its own sake.
Consider also, that blockchain is ideally suited to the sale and distribution of digital art, which is locked in the blockchain, and has no need of a vault for its safe storage. Bring to mind, also, that neural networks, artificial intelligence systems, are being developed which can create art works without the need for human creativity. The creation of original an artwork is therefore limited neither by the human talent required for its creation, nor the time and labour which can reasonably be expended by a talented individual in creating it.
Taking these factors together, there is potentially created a market in which masses of people can launch themselves into investments which they poorly understand, and where there is no practical limit on the digital investment objects which can be created by artificial intelligence. There is, in short, created the breeding ground for a bubble and subsequent crash.
We have been there before. They were heady days: the market had originally been the preserve of a few professionals, but as the mania took hold, ordinary, middle class families put themselves heavily into debt betting on a sure thing. It was a new economy where price outstripped intrinsic worth many times over. Everyone based their investment strategy (if greed justifies such a title) on a wildly over-optimistic estimate of future income streams and profits that might never materialise. Homes, entire businesses were mortgaged in order to get on the bandwagon. Then people got to thinking: is this sensible? What lies behind the hype? Those who were smart sold out early, but most people hung on only to see the market collapse. Ruin was everywhere.
For many years after 1637, no-one would look at a tulip bulb again.
Beware, therefore, lest the voyage to Byzantium leads not towards Utopia but, instead, to Dystopia.