e-gold and the Compliance Wall Every Centralised Issuer Eventually Hits

Here is how the company described itself on its own corporate history page, in a capture the Internet Archive took on 7 October 2006:

e-gold Ltd. Is a Nevis, West Indies company created to serve as the General Contractor responsible for performance of the e-gold Account User Agreement. e-gold under its charter is completely dissociated from the business risks relating to exchange.

Read that twice, because it is the whole postmortem in two sentences. A company in one jurisdiction, issuing a unit, with a charter drafted specifically to hold a category of risk at arm’s length. Every word of it is a legal construction. Not one word of it is a technical one.

That is the finding. e-gold built an elaborate structure to separate the issuer from everything the issuer did not want to be responsible for, and the structure was made of contracts and corporate charters, which are exactly the materials that a government is equipped to take apart. The ledger itself was never the target and never had to be.

DigiCash failed because nobody would carry it. e-gold did not have that problem. It had the opposite one, and the two failures together are the reason the next generation of designs looks the way it does.

What the company said it was

The same archived page continues:

e-gold was developed and deployed as an Internet payment system by Gold & Silver Reserve (G&SR), Inc., a Delaware Corporation in 1996. The e-gold roles of Issuance and Settlement were devolved to e-gold, Ltd. in January 2000. G&SR, Inc continues to serve as the operator of the e-gold payment system.

So by its own account: a Delaware operator, a Nevis issuer, a formal split between operating the system and issuing the unit, and a user agreement holding the two together. The directors page names Douglas Jackson, a physician board certified in radiation oncology, as chairman, and says that in 1995 he “conceived of e-gold as remote payment system that did not require an obligatory financial intermediary”. Barry K. Downey, a lawyer, is listed as co-founder and director.

Note the ambition in Jackson’s own phrasing, because it is startlingly close to what people were still saying about digital money a decade and a half later. No obligatory financial intermediary. And then note what the corporate history describes: an intermediary, with a charter, in the West Indies, with a Delaware sibling operating it.

The separations were real, and every one of them was legal

e-gold went further than the corporate split. It also declined to run the on-ramp. The home page, captured 4 October 2006, invited users to

Buy or sell e-gold from/to the independent exchange service who best meets your needs

and, on the next line, offered “Exchange provider guidelines” under the heading “start your own exchange business today”.

That is a deliberate architecture. The issuer holds metal and moves entries between accounts. Somebody else, unaffiliated, takes the cash. It is the same shape the industry rediscovered later, with a protocol in the middle and regulated fiat gateways at the edges, and it was arrived at for the same reason: whoever touches the cash inherits the licensing question.

The problem is that the separation only works if there is nobody in the middle to serve process on. e-gold had a chairman with a name, a Delaware corporation with a registered agent, a database, and administrators with credentials to it. Every legal wall it built terminated in a person who could be asked to do something.

No chargebacks was the product

The 2006 home page sold the system on two properties above all others. The first was speed: “Immediate payment no waiting for checks to clear or credits to be made.” The second is the one that matters here, and it was pitched in the company’s own quotation marks:

No chargebacks “Get paid, stay paid” unlike credit card payments.

Irreversibility was not an accident of the design and it was not a technical consequence somebody had to apologise for. It was the marketing. It was the reason a merchant would take e-gold instead of a card, and it was, unavoidably, also the reason certain other users would take e-gold instead of a card.

You cannot sell finality selectively. A settlement system that cannot reverse a payment cannot reverse it for anyone, and the population that most values that property is not a population an issuer gets to choose. This is not a moral observation about e-gold’s users. It is a structural one about what happens when you make a promise about the mechanism rather than about the customer.

The part that was reversible

Here is the asymmetry that makes e-gold a postmortem rather than an anecdote. Payments were final. Accounts were not.

The ledger was a private database. The operator could read it, could see the graph of who paid whom, and could stop an account from transacting. Whatever the user agreement said about finality of settlement, the administrative interface had a different set of verbs available to it, and those verbs were the ones that mattered when somebody came asking.

So the system offered its users the strongest possible guarantee about the one thing the operator did not control, and no guarantee at all about the thing the operator did control completely. That combination is what “centralised digital currency” actually means in practice, and it is why every serious design that came afterwards spent so much effort on removing the operator rather than on constraining them. A constrained operator is still an operator.

It is also the second half of an argument this site opens with. Every digital cash design before 2009 answered the double spend problem with a party who keeps the ledger and refuses the second spend. e-gold is what that party looks like once it has customers: not a villain, and not a design flaw, just a company, with an address, holding a database it can edit.

What this piece is not covering, on purpose

The legal proceedings that ended e-gold’s operations are a matter of public record and they are not restated here: not the counts, not the pleadings, not the outcomes, not the figures. A history site can make the structural point without relitigating a case, and the structural point does not depend on who was right. Whether the enforcement was proportionate is an argument, and it is somebody else’s argument.

What is worth recording instead is the state of the evidence. The archived e-gold site is a frameset, and the Internet Archive captured the outer frames far more reliably than the inner ones. The statistics page and the “examiner” page, which were the company’s own public accounting of circulation and backing, survive as empty wrappers with a link reading “Click here for a non frames version”. The substance behind them is not in the capture I could reach.

What was expanding, what was contracting

Two curves ran in opposite directions across e-gold’s decade, and almost nobody was reading them as a pair.

The first curve was capability, and it went up steadily. A working system for moving a metal-denominated balance between strangers, worldwide, in seconds, without a card network, existed and had users. On the engineering, e-gold was finished years before anyone else in this history had shipped anything at all.

The second curve was the room a private issuer had to operate in, and it went down. Through the same period, the category of business e-gold was in was being defined, and the definitions were tightening around exactly the arrangement e-gold had built: a company holding value for others and moving it on their instruction. The corporate structure that looked like prudent risk allocation in 1996 read, by the middle of the next decade, as a description of a licensable activity in a jurisdiction that had not licensed it.

Who could tell at the time? A small number of people, and mostly not the users. The company itself could see the second curve, because it began blocking accounts before anyone made it. That is documented in the sense that the capability plainly existed and was plainly used. What almost nobody drew was the conclusion: that the ability to block an account and the vulnerability to being told to block an account are the same property, seen from two sides, and that no charter drafted in Nevis makes the second one go away.

The builders who came next did not need the legal history. They needed the one sentence underneath it, which is that a system with an operator can be addressed through the operator, and no amount of cryptography applied to the payments changes where the pressure lands.