Giving It Away Was a Deliberate Answer to a Real Problem
A payment network with no holders is a chicken and egg problem, and the first fix was to give the chicken away.
A payment network is worth using when the person you want to pay will accept what you are sending. That sentence contains the whole difficulty. Nobody accepts a unit nobody holds, and nobody bothers to hold a unit nobody accepts, and the two conditions sit there refusing to go first.
Every new monetary system has to break that deadlock somehow, and the options are narrower than they look. You can compel acceptance, which requires a state. You can back the unit with something already accepted, which requires an issuer and a reserve, and puts a company inside the settlement path in the way that e-gold and DigiCash both discovered. Or you can hand the thing out until enough people are holding it that the question of who accepts it becomes answerable.
The third option is the one the early network took, and the artefact it produced is the faucet.
What a faucet actually is
Mechanically it is trivial, and the triviality is the interesting part. A faucet is a web page holding a balance, a form asking for a receiving address, and a gate cheap enough for a person to pass and expensive enough that a script cannot pass it a thousand times a minute. You solve the gate, you paste an address, the page sends you some units, the page’s balance goes down.
That is the entire machine. There is no protocol involvement, no special rule in the consensus code, no allocation schedule. A faucet is an ordinary user with a wallet and a public form in front of it. Anybody could run one and the operator had no more authority than anyone else, which is precisely why it could exist at all: distribution was a thing a person could just do, rather than a thing an issuer had to authorise.
The gate matters more than it looks. A giveaway with no gate is not a distribution mechanism, it is a single automated withdrawal by whoever writes the fastest loop. The gate is what turns “give it away” into “give it away to many separate people”, and many separate people is the entire objective. This is also the point where the mechanism is weakest, because the gate can only ever distinguish effort from automation, never one person from the same person returning under a different address. Nothing about a faucet can establish that its outputs went to distinct humans, and no faucet operator was ever in a position to claim otherwise.
The one I can actually read
The instinct in this genre is to write from the received story. The received story about faucets is short, warm and unsourced, so here is an archived page instead, captured on 5 August 2010, run by Gavin Andresen and signed by him in the page copy.
It is titled “Get Bitcoins from the Bitcoin Faucet”. It asks the visitor to “solve the ‘captcha’ then enter your Bitcoin Receiving address and press Get Some”, with captcha in its own scare quotes, which tells you roughly how much adversarial pressure the operator expected. It carries a heading reading “What’s the catch?” and answers it:
No catch– I want Bitcoin to be successful, so I created this little service to give you a few coins to start with.
There is a second section, headed “Hey! I thought you were giving away FIVE!”, explaining that the faucet is “running dry due to very high demand” and that a donation from a third party would make it more generous again. A donation address is printed on the page for exactly that purpose, with a note that a refill may take up to thirty minutes to show up in the available balance.
Three things follow from those two paragraphs, and none of them is sentimental.
The stated motive is adoption, not charity and not marketing. The operator wants the network to work, and a network that works needs holders. That is a distribution argument, and it is made out loud on the page rather than reconstructed later by somebody explaining what the early community must have been thinking.
The supply was other people’s, and it ran out. The faucet was refillable by strangers, and it depended on being refilled. Whatever else it was, it was not an issuance mechanism. It could only redistribute units that already existed, which means it was structurally downstream of whoever was generating them and could never outrun them.
Demand for the free units was described as the problem. “Running dry due to very high demand” is a sentence about people wanting the thing. It is not a sentence about people wanting to buy the thing, and the difference is the whole limitation of the mechanism, which is the next section.
I am not reproducing the balance figure or the per-visit amount printed on that capture. A single snapshot of one page’s balance is not a distribution statistic, and quoting it would dress an incidental number up as a measurement of something. How much moved through that faucet in total, and to how many separate people, is not recoverable from what survives.
The original page is long gone and the service it ran on has changed hands and shape since. What I read was a Wayback capture, and everything above is restricted to what is visible in that capture on that date. The page was edited over its life, so a claim about “what the faucet did” is really a claim about what one snapshot said, and I have written it that way on purpose.
This is the general condition of the period rather than a quirk of one site. As the piece on the forum as the only record argues, the documentary base for these years is a handful of privately operated web properties that nobody was archiving as a duty. A history of early distribution written without checking which pages still resolve is a history of whatever happened to survive, presented as a history of what happened.
What free distribution can and cannot do
A faucet creates holders. That is a real accomplishment and it is also the entire accomplishment.
It does not create demand. Somebody holding a unit they were handed has no reason to want a second one, no reason to refuse to part with the first, and no information at all about what either is worth. Demand had to arrive from somewhere else entirely: from people who wanted the units for a purpose, or who wanted them badly enough to give up something for them, and neither of those is a thing a giveaway can manufacture.
Nor does it establish a price. A distribution event has no price in it, because nothing was traded. It produces a population who hold, and price formation begins after that and separately, in bilateral arguments between two people about what a fair swap looks like. A faucet moves the network from zero holders to some holders. Everything that makes a currency a currency happens in the next step.
And it cannot demonstrate its own effect. There is no counterfactual. Nobody ran the network twice, once with a faucet and once without, and no measurement exists that separates units acquired free from units acquired any other way in what their recipients then did. Anyone who tells you the faucet was decisive, and anyone who tells you it was a rounding error, is telling you a preference.
The precedent, and the honesty gap
Later eras on this site are dominated by arguments about token distribution: who got units, on what terms, before whom, and what was disclosed. Those arguments inherit the structure of the faucet and almost none of its candour.
The faucet’s design was fully legible from the page. One operator, one balance, one form, a stated motive, a public donation address, and a plain admission that the supply was finite and borrowed. You could read the whole mechanism in a minute, and you could verify the address it paid from.
The later versions of “hand it out to bootstrap the network” are typically structured as allocations decided in advance, distributed against criteria published afterwards if at all, and described in language that presents a distribution decision as a discovery about who deserved one. That is not a claim that the later mechanisms were dishonest, which is a claim about intent that this site does not make. It is a claim about legibility, which is checkable: the faucet’s rules fit on the page a user was looking at, and most of what came later did not.
What was expanding, what was contracting
What expanded was the number of people holding a unit, and that expansion was deliberate, cheap and completely disconnected from the thing that usually produces holders, which is somebody buying something.
What contracted, and this is the part that is easy to miss, was the share of holders who had any idea what they were holding. Early participants who obtained units by running the software themselves had at least been through the process of understanding what the software did. A faucet recipient had solved a puzzle and pasted a string. The population grew and the average understanding inside it fell, and both curves were consequences of the same design decision.
Who could tell at the time is the honest question, and the answer is nobody, in either direction. The operator could not measure whether the giveaway was converting recipients into users, because the only instrument available was a balance going down. That was one of the earliest moments in this history where a number was available, was watched, and was measuring the wrong thing.