In re comment on Financial Privacy blog

This post is in response to a comment on my blog post about Financial Privacy. See


I use terms like unlinkability and anonymity in the academic vernacular, not in respect to any legal definition. After all, the law can define a word to mean anything it wants. The technique used to anonymize the transaction is similar to Anonymous Lightweight Credentials (see for more information on ASL). Breaking the anonymity would require solving the discrete log problem. Solving that problem would put in jeopardy much of the cryptography upon which the world relies today, so I’m reasonably confident of its security for the moment.  Spending a token under the Microdesic system based on the technique allows the user to prove they have the right to spend a token without identifying themselves as a particular person who owns a particular token.

Now, as far as de-anonymization under fraud, if a user double spends the same token, they reveal themselves. If I were to offer a somewhat real world analogy, it would go like this: I walk into a store. If I’m minding my own business, the store can’t distinguish me from any other customer in the store. I can purchase what I want and remain anonymous (subject to the store taking other measures outside this scenario, like performing facial recognition). However, if I commit a crime (in this case fraud), the store forces me to leave my passport behind. (It is sometimes hard to create real world analogies of the strange world of cryptography, but this should suffice).

In other words, prior to committing that fraudulent act, I’m anonymous. In the act of committing that fraud (in order for the store to accept my digital token/money), I’m standing up and announcing my identity and revealing my past purchases.

Returning, now to the law and specifically Recital 26 of the GPDR, it states “To ascertain whether means are reasonably likely to be used to identify the natural person, account should be taken of all objective factors, such as the costs of and the amount of time required for identification, taking into consideration the available technology at the time of the processing and technological developments.”  There is clearly a temporal element. In other words, we need not account for a super computer in the distant future, or someone solving the discrete log problem. I also doubt the GDPR contemplates forcing the user to reidentify him or herself as a reasonable means of reidentification. Surely, they aren’t saying that if you rubber-hose the user and tell them to identify when and where they made a purchase, that’s reidentification. The data subject always knows that information, the question is whether anyone else can ascertain it without the user’s assistance. Under the Microdesic system, at the time of a non-fraudulent transaction, there is no reasonable means of reidentification (i.e. you must solve the discrete log problem).

The Middle Man

The subject of my previous post was financial privacy vis-à-vis decisional interference. The comment to which this post replies posed the question of whether Microdesic becomes the middle-man with the ability to interfere in the decision-making capabilities (i.e. spending decisions) of the user. Let me first explain by counter-example. When a payment authorization request comes in to PayPal, it knows the account of the spender, the account of the recipient, who those parties are, how much is being transferred and some extra data collected (such as in a memo, etc.). At that point, PayPal could, based on that information, prevent the transaction from occurring. Maybe they think the amount is too high. Maybe the memo indicates the person is purchasing something against PayPal’s AUP. The point is they can stop the transaction at the point of transaction. The way Microdesic works is different. A user in the Microdesic system is issued fungible tokens. From the system perspective, those tokens are indistinguishable from user to user. In fact, the system uses ring signatures which mixes a user’s tokens with other user’s tokens, to reduce correlation through forensic tracing. The tokens are then spent “offline” without the support of the Microdesic server. All the merchant knows is that they are receiving a valid token. Microdesic has no ability to prevent the transaction at the time of transaction.

Now for a bit of a caveat. Because the tokens are one time spends, the Merchant must subsequently redeem the tokens, either for other tokens or for some other form of money held in escrow against the value of the tokens. Microdesic could at this point require the Merchant to identify themselves and prevent redemption. Merchants that weren’t approved by Microdesic might therefore be excised from the system by virtue of being unable to redeem their tokens. However, the original point remains. Unlike a PayPal or credit card system, which authorizes each and every transaction, Microdesic has no ability to approve or disapprove of a particular transaction at the point of the transaction.

Financial Privacy and CryptoCurrencies

Financial privacy is most often conceptualized in terms of the confidentiality of one’s finances or financial transactions. That’s the “secrecy paradigm,” whereby hiding money, accounts, income, expenses prevents exposure of one’s activities, subjecting one to scrutiny or revealing tangential information one’s wants to keep private. Such secrecy can be paramount to security as well. Knowing where money is held, where it comes from or where it goes give thieves and robbers the ability to steal that money or resource. Even knowing who is rich and who is poor helps thieves select targets.

Closely paralleling “financial privacy as confidentiality,” is identity theft, using someone else’s financial reputation for one’s own benefit. In the US, Graham-Leech Bliley’s Safeguards Rule provides some prospective protection against identity theft, while the Fair Credit Reporting Act (FCRA), Fair and Accurate Credit Transactions Act (FACTA) and the FTC’s Red Flag Rules provide additional remedial relief.

However, as I’m found of saying in my Privacy by Design Training Workshops, “that’s not the privacy issue I want to talk about now.” All of the preceding examples are issues of information privacy. Privacy, though, is a much broader concept than that of information. In his Taxonomy of Privacy, Professor Daniel Solove categorized privacy issues into four groups: information collection, information processing, information dissemination and invasions. It’s that last category to which I turn the reader’s attention. Two specific issues fall under the category of “invasions,” namely intrusion and decisional interference. Intrusion is something commonly experienced by all when a telemarketer calls, a pop-up ad shows up in your browser window, you receive spam in your inbox or a Pokemon Go player shows up at your house; it is the disturbance of one’s tranquility or solitude. Decisional interference, may be a more obscure notion for most readers, except for those familiar with US Constitutional Law. In a series of cases, starting with Griswold v. Connecticut and more recently and famously in Lawrence v. Texas, the Supreme Court rejected government “intrusion” in the decisions individual’s make of their private affairs, with Griswold concerning contraceptives and family planning and Lawrence concerning homosexual relationships. In my workshop, I often discuss China’s one child law as a exemplary of such intrusion.

The concept of decisional interference has historical roots in US privacy law. Alan Westin’s definition of information privacy (“the claim of individuals, groups, or institutions to determine for themselves when, how, and to what extent information about them is communicated to others”) includes a decisional component. Warner and Brandeis’ right “to be let alone” also embodies this notion of leaving one undisturbed, free to make decisions as an autonomous individual, without undue influence by government actors. With this broader view, decisional interference need not be restricted to family planning, but could be viewed as any interference with personal choices, say government restricting your ability to consume oversized soft drinks. I guess that’s why my professional privacy career and political persuasion of libertarianism are so closely tied.

But is decisional interference solely the purview of government actors, then? Up until recently, I struggled with coming up with a commercial example of decisional interference, owing to my fixation on private family matters. A recent spark changed that. History is replete with financial intermediaries using their position to prevent activities they dislike and since many modern individual decisions involve spending money, the ability of an intermediary to disrupt a financial transaction is a form of decisional interference. A quick look at Paypal’s Acceptable Use Policy provides numerous examples of prohibited transactions that are not necessarily illegal (which is covered by the first line of their AUP). Credit card companies have played moral police, sometimes but not always at the behest of government, but more often being overly cautious, going beyond legal requirements. Even a financial intermediaries’ prohibition on illegal activities is potential problematic, as commercial entities are not criminal law experts and will choose risk-averse prohibition more often than not, leading to chilling of completely legal, but financially dependent, activity.

This brings me to the subject of crypto-currencies. Much of the allure of a decentralized money system like Bitcoin is not in financial privacy vis-a-vis confidentiality (though Zerocoin provides that) but in the privacy of being able to conduct transactions without inference by government or, importantly, a commercial financial intermediary. What I’m saying is not a epiphany. There is a reason that Bitcoin beget the rise of online dark markets for drugs and other prohibited items. Not because of the confidentiality of such transactions (in fact the lack of confidentiality played into the take-down of the Silk Road) but because no entity could interfere with the autonomous decision making of the individual to engage in those transactions.

Regardless of your position on dark markets, you should realize that in a cashless world, the ability to prevent, deter or even discourage financial transactions is the ability to control a societ. The infamous pizza privacy video is disturbing not just because of the information privacy invasions but because it is attacking the individual’s autonomy in deciding what food they will consume by charging them different prices based on external intermediaries social control (here a national health care provider). This is why a cashless society is so scary and why cryptocurrencies are so promising to so many. It returns financial privacy of electronic transaction vis-a-vis decisional autonomy to the individual.

[Disclosure: I have an interest in a fintech startup developing anonymous auditable accountable tokens which provides the types of financial privacy identified in this post.]