virtual currencies

The recent discussion about high denomination banknotes has extended into the merit of paper cash.

In so far as paper cash is needed, does it need to be done the old school way? For buying goods in legal market, probably not, an alternative would be to use “bearer numbers” (implemented as QR-code or barcode) that could be printed on paper if people want that.

The way it would work would be something like:

  • ATMs are replaced by an online banking facility that debits the client’s account, credits the bank’s account at the central bank (or whoever is the barcode cash clearer), which in response issues a new random number associated with the amount. The number is printed in computer readable form on a ticker the punter puts in their wallet.
  • In a shop the punter shows the number (the ticket) which the cash register redeems with the central bank computer. A new number is issued for change and printed on the receipt. These numbers can then be spent at other shops.
  • People can also simply give away the printout to others, as long as the recipient trusts the giver not to spend a copy before them.
  • Splitting a ticket would require an online app similar to the shop’s cash register.

This reproduces something similar to paper cash but without the need for actual ATMs with a stock of high value banknotes, or a banknote printing and processing infrastructure.

The principal disadvantage is that it requires all non-trusted-parties transactions to be online, to check the value and validity of the number with the centralised issuer/redeemer computer system.

The issuer could be a Bitcoin-like system — you can indeed do all of the above with Bitcoin — though current blockchain technologies add a delay to transaction authorisation that’s impractical for most shop-style settings.

It could be argued this is less anonymous than paper-cash because all redeeming transactions are logged (like in Bitcoin).  Technically that could be done with current cash as well: banknotes have serial numbers that could be tracked to produce interesting meta-data — a government could easily mandate the use of some scanner widget in the cash registers of all legal businesses.

The current models used to fund digital media are not very satisfactory.

Old Newspaper likes firewalls, with content behind a monthly subscription. Like in the good old dead tree times. While those with a conservative ageing demographic can sometimes get a sizeable subscriber base, it does not seem to be the future when it’s so easy to move elsewhere in a world of plenty of free content. Attention seekers and new entrants will always be happy to produce free content when the cost of publishing itself is so close to zero.

Digital Natives like the advertising funded model, free to access at the expense of privacy, degraded user experience and an incentive to produce “Social Media leaders’ 25 ways to produce Click Bait headlines” content.

I hereby propose a new model: charge for comments. There are lot of trolls in the world who have a lot of rage, seem very motivated, and produce a lot of low quality and oft duplicated content that would be nice to see less of.

Several models can be envisaged:

  • Charge per comment: each post is charged a small fixed amount.
  • Deposit system: the user pays a fixed amount, once, that gives a right to a number of daily comments. The amount, or some of it, is forfeited if one or more posts are moderated away. Users who do not post objectionable content can get their deposit refunded when they decide to stop posting – for them posting remains close to free.
  • Attention auction: for popular articles with many comments, you could auction the most visible places in the comment page. Surely some people would pay a penny to lift their pearl of wisdom to comment page 1. This can be combined with the deposit idea.

This should help deal with spam as well as trolls incidentally. Payment in digital currency can preserve anonymity if so desired.

One thing I think Bitcoin, or a similar blockchain-based token system, may be useful for is to replace centralised social networks, or more broadly messaging systems.

A core if oft forgotten feature of a social network is how it manages spam and other forms of network abuse. One solution may be to exploit altruistic punishment, a tendency people have to want to correct bad actions even if it’s not in their direct private interest to do so. In familiar terms, people seem to enjoy pushing the “dislike” button even if they get no personal benefit.

How could one devise a cryptocurrency transaction type that cab be used to harness this effect?

One possible way is what I’ll call “altruistic destruction”. The basic idea is to have a transaction that can cancel some of the recipients’ funds, that is enriching everybody else, through the reduction of the monetary base. If this is free to the sender, this is open to abuse — though whether such abuse would be common could be a subject to interesting experiments — so a compromise might be simply that both the sender and recipient destroy some tokens. As a base case a simple matched ratio may do. The altruistic destruction transaction semantic would be as follows.

When Alice sends to Bob an altruistic destroy for N tokens,

  • Alice’s account value decreases by M tokens.
  • Bob’s account value decreases by M tokens.
  • where M = min(N, Bob’s balance)

M copes for the case where Bob’s has fewer token than N, assuming the token system does not allow debit balances.

In paper terms, this would be equivalent to burning a bank note with the name of a miscreant on it, where through some magic burning the named banknote would also make a banknote of the same value vanish from the miscreant’s wallet. Economically this reduces the total amount of banknotes in the system thus making holders of the remaining banknotes richer (like simply burning one without magic does) everything else being equal.

For practical use in a spam control system, this would have to be implemented with policies that uses minimum balances as condition of message transmission — a distributed deposit system of sorts.

There may also be other application of that transaction type. The ratio of destroyed tokens between sender and recipient may also be toyed with, but 1:1 may be special in that the cost of inflicting damage is equal to the damage individually, while the social benefit is leveraged: destroying 1 of one’s tokens produces 2 tokens’ worth for the community, and the punishment’s social value is free on top of that.

Having not made divination posts for a while, let’s make a new prediction, that the price of Bitcoin will go down markedly in classic currency terms in the next 3 to 6 months.

Transaction acceptance means net bitcoin selling

The reasoning is pretty simple: one of the main happenings in the Bitcoin world recently has been increased acceptance at legitimate vendors like Dell, Overstock, Expedia (for hotels) and others. Bitcoiners see that as a success and, ironically, some Bitcoin holders see it as doing something of civic value to buy at vendors that accept Bitcoin.

But what happens when someone buys a laptop from Dell with Bitcoin? There is no sign of Bitcoin becoming used in the business supply chain, where it doesn’t really solve any material interesting problem, so Dell will immediately convert the Bitcoins to dollars to pay their employees and suppliers. Indeed one of the reason vendors are accepting Bitcoin is that payment processing intermediaries make that easy by doing it for them and “removing the Bitcoin risk” from the vendor: they price in dollars (or another classic currency) with the payment gateway, and get dollars paid by classic wire transfer.

What happens on the buyers’ side: will people buy Bitcoins to buy Dell laptops? It seems unlikely. Dell is probably more trustworthy than most operators in the Bitcoin ecosystem, and you don’t want anonymity or pseudonymity when buying a laptop: you want them to know your address and deliver the laptop there. So there’s basically no reason to pay using a classic method to get Bitcoin and then use those shortly at a legitimate vendor. So who’s gonna buy Dell laptops with Bitcoins? My guess is mostly people who for some reason or other had accumulated Bitcoins, legitimately or not, and now have an opportunity to spend it on something more useful than Bitcoin t-shirts, and also see it as a “good” thing for the ecosystem.

And then came momentum

As it happens Bitcoin has been on a downward trend since its peak for several months, and has also seen a further boom in “mining” activity. Bitcoin mining, like its real world namesake, is one of those activities which tends to be structurally loss-making as it’s very hard for rational investors not to be outnumbered by innumerate optimists. Volatility and a downward trend may accelerate some of these past optimists giving up on accumulating mining proceeds, and exiting through transactions, which might provide some comfort (product + civic value) even if nursing a net loss.

Raw momentum regardless of mining (“it’s going down because it’s going down”) will also tend to put a further downward pressure, so basically we have a fundamental trend towards sale of Bitcoin created by transaction acceptance, amplified by momentum and mining dynamics.

Absent any new net buyers of Bitcoin, it’s hard to see how it could go anywhere but down on a simple demand and supply basis. It’s hard to think of a source of new net buyers. Most people who may be interested in Bitcoin at this stage — where it has acquired some maturity but is still an emerging underground tech — have probably already jumped in and are sustaining steady-state flows. I just don’t see any new major demand source coming in to offset transaction-driven sales.

Show me the money

So to quantify the prediction for verification, let’s say that BTC/USD will touch US$ 250 (approximately halve in value from now) for at least one day at reputable exchanges before April 1, 2015.

(Disclosure: I have a few pennies on downward Bitcoin bets at shady blockchain betting operator — not a recommended counterparty or investment. Note that if I win I get my payoff in devalued Bitcoins, which, if my fellow gamblers don’t discount that effect correctly, as I expect them not to, will be a pyrrhic victory.)

Estimates of the Bitcoin holdings of the network’s anonymous designer, Satoshi Nakamoto, seem to indicate he owns a minimum of 5-10% or the current issuance. He doesn’t seem to have “spent” them as the addresses known to belong to him have not seen activity — this can be checked in the blockchain, Bitcoin’s public transaction ledger.

If he’s not thrown away the keys in a spring clean, he effectively controls the currency. His holding is a multiple of daily volumes on all exchanges, so he could easily intervene on the markets to cap the price (like a regular central bank) and introduce a fixed or capped USD/BTC rate. He wouldn’t even need to spend much, just announcing “I’m Satoshi Nakamoto and I’ve decided to cap the rate at this value” would do it, as nobody has anywhere near comparable firing power so a successful challenge seems unlikely.

The market cap of Bitcoin-derivatives being even smaller, by extension he can also control the entire alttcoin movement.

If the protocol doesn’t fall apart, his capacity is still notionally limited as he can’t issue new Bitcoins — he’s surrended control of the algorithm long ago. But even then, he may have enough buying power to bribe enough nodes to be able to control the network if he so wished. He could similarly sell his stash at a discount to someone else who wants to control Bitcoin, easing the exit problem he has should he want to convert his holding to real world claims.

Overall that’s more power than any central banker — who have a monetary policy committee to contend with and can be removed by their parent government sponsor — has. If estimates of his holdings are correct, this “decentralised” currency is in effect an absolute monarchy, with an absentee, and so far benevolent, king.

I incidentally had a wander around the Bitcoin ecosystem in the past few days and there are a number of entertaining phenomenons to observe.

MtGox: a slow train crash

Looking at comparative exchange rates between Bitcoins and fiat currencies, one can’t help notice that BTC are 10% more expensive at MtGox, the highest volume Bitcoin exchange, than at every other exchange. The reason is not hard to find: MtGox has problems with their banking partners so that they have limited capacity to process withdrawals, and have thus set up a rate-limiting (both time and size) queue system. So basically it takes weeks to get money out of MtGox, making the obvious arbitrage opportunity of buying cheap bitcoin from any other exchange and getting back your original capital + 10% by selling it on MtGox and withdrawing it much harder (or for people who are BTC-based, the mirror strategy of buying fiat temporarily and converting it back into Bitcoin via cheaper exchanges). Of course the arbitrage is less interesting when (1) it takes weeks, (2) is limited in size, (3) MtGox might not be there by the time your withdrawal is due.

Interestingly MtGox remains the highest volume exchange despite being functionally insolvent — any old school bank sitting on funds for a month would be put in resolution by their regulator and counterparties would run away from it as fast as possible. It would be interesting to know how much of the remaining volume is the arbitrage trade and how much is people actually needing a transfer for non speculative reasons. Some inertia might be due to the difficulty of being validated on other exchanges.

Bitcoin exchanges: a flawed idea

MtGox problems are I think revealing a more fundamental problem. The entire modern banking system is oriented towards traceability and, to a degree, reversibility. Interestingly MtGox’s euro bank sits on incoming electronic transfers for 7-10 days, presumably to cope with reversals. Any transaction chain that gets out of regular banking and back in again via Bitcoin is in principle a loophole in the world banking system (let’s assume for a moment that Bitcoin is as anonymous as its proponents claim).

The way it seems to work is that while a new Bitcoin exchange is small enough to be under the radar it’s left alone until it becomes big enough to trigger compliance problems and then loses its interface with the regular regulated money world. I don’t think that’s going to get any better, limiting the potential for Bitcoin to gain wide acceptance. It can still work as an underground currency by using bridge goods for exchange, or peer-to-peer exchange, like LocalBitcoins, which basically works around the regulations by turning private individuals into unlicensed money changers, probably breaking some rule or other but possibly escaping it as long as individual members operate on an amateur scale.

ICBIT futures contango

Another arbitrageable anomaly is the contango curve for BTC/USD futures on the ICBIT futures exchange. Liquid things normally have a flat term structure, because you can construct a market neutral position with the underlying and the future and gain the difference in a risk free manner.

Some possible explanation for this contango:

  • innumeracy may dominate, if more people believe that the future prices for liquid assets are bets about prices at expiration than there are people prepared to arbitrage this basic error.
  • the icbit exchange is too small to be worth bothering (the open interest on the BTC/USD futures is less than $400k nominal).
  • an arbitrage position that is short is subject to infinite risk if the price on icbit spikes before converging at maturity.
  • an arbitrage position is exposed to counterparty risk: the exchange, like many bitcoin ventures, is probably a one-man shop who can at any time disappear with the margin account or get it hacked/stolen.

Fees: even more expensive than regular finance

One striking thing is that bitcoin exchanges, and other bitcoin-based financial websites, despite skipping the full regulation of the regulated world, charge fees that are can be higher than the regular regulated financial industry, for example an arbitrage transaction getting euros in via Bitstamp and out via MtGox would cost:

  • Bitstamp deposit: unspecified USD/EUR cost
  • Bitstamp exchange fee: 20-50 bps (volume dependent)
  • MtGox exchange fee: 25-60 bps (volume dependent)
  • MtGox withdrawal fee: 100 bps

A speculative forex round trip with a competitive regulated money solution is much cheaper than that.

These services are just a bit of software, with a supporting geek community, it’s surprising cooperative services that don’t charge fees haven’t appeared.

Another look at fees is the cost per transaction shown on which seems to oscillate around 2-3% which seems pretty steep to me for transactions within a purely electronic currency with no services attached. I understand this includes the block fee from BTC created and given to miners, which are not paid directly as part of the transactions, but are fair to include given that they increase market cap and thus, everything else being equal, devalue existing bitcoins.

Bitcoin has gone mainstream, boomed and crashed, in just a few short weeks. If you want to understand how it works, the self-evident blog has a brilliant introductory series which goes in just enough depth to be interesting and yet capture the subtleties of the technology.

What interests me however is to try and think of ways in which it could fail. It’s based on a clever technology, but like computer security it’s very hard to design software systems or algorithms that correctly anticipate every possible misuse. It’s been surprisingly resilient so far and a remarkable success story. It’s nowhere near becoming a real alternative currency for what existing mainstream currencies are used for, but just getting considered for that role is a remarkable achievement for something created by one lonesome and anonymous hacker.

There’s probably a myriad of ways in which it could fail, so I’ll just throw a few random ideas. It is also worth noting that currently most people involved in Bitcoin seem to genuinely want it to succeed, for various ideological or aesthetic reasons. It feels like the naive Usenet of the 80s and early 90s before wilful spam was introduced into it by unrepentant green card lawyers.

The blockchain as a data store

The core of Bitcoin is a distributed ledger of all transactions ever conducted, which may “metastasise”. The obvious risk is what happens if it becomes too big, although it’s also hard to overwhelm modern computers with mere transaction information (a modern smartphone has more computing power and storage than a major bank had when bank accounting moved to electronic processing), but it may face tangential attacks or misuses.

Through steganography any data store can be used to store anything else. People may start to use the Bitcoin blockchain as a store for information they want to publicise anonymously or hold for a a long time, by generating sequences of special dummy transactions. This might cost transaction fees (to get included in the blockchain) but for that you get perpetual public distributed storage, which is a good deal! It doesn’t need cost actual Bitcoins beyond transaction costs, because the encoding transactions can be conducted between Bitcoin addresses all belonging to the entity writing the data.

Here are some applications I can think of:

  • Way to publish illegal content like child porn
  • Way to publish secret information, like wikileaks
  • Back up device for small data as a substitute for a paid for data store
  • Underground social network using the blockchain as the transport and storage mechanism for status updates
  • Anything you can do with a public more or less anonymous data store…

This is a bit limited in size, if people start backing up their HD porn collection encoded as Bitcoin transactions the network would probably be overwhelmed fast enough to require remedial action.

For small controversial content though it might attract the ire of authorities which are used to be able to take such content down… the obvious answer would be for a way to “delete” the offending “transaction” but this is totally at odds with the core design principles of Bitcoin…

Jamming the network with a penny auction

Another idea is that an hostile force which has managed to acquire some Bitcoins could try and jam the network. For instance they could offer a prize to the Bitcoin address which posts the most transactions in a given block, which would push third parties to devise inventive ways to maximise pointless transactions. The logic is that of a penny auction, which leverages greatly the impact of the prize — a lot of people may compete for a single prize, collectively loosing more in effort and transaction cost than the nominal prize is worth. If it worked well the price of transaction may go up with increased competition to be included in the blockchain, perhaps making some genuine transactions uneconomic.

Becoming an ordinary currency

Bitcoin is effectively a software monoculture — nearly every actual node in the network uses the reference implementation — so there’s some scope for the people managing this code to act, de facto, like central bankers for regular fiat currencies. As long as they can get the community to upgrade and not fork the code — arguably a protection — they control all the rules that are encoded in the algorithm as code. This can also be seen in the highly concentrated mining syndicates, who intermediate the mining process, although again the members could switch if they disagree and manage to overcome institutional inertia. One could make a parallel between them and how the network of regional Federal Reserve banks compose the Fed…

Reinventing the existing would not be a particularly unusual phenomenon. Some institutional forces are so strong that alternatives tend to gravitate back to something close to what was originally rejected, if you let them marinate long enough.

Scarcity is not scarce

This would not be a technical failure, but it could also fall into disuse because of being drowned by competitors doing the same thing — Bitcoin is a form of synthetic scarcity which is itself not scarce: anyone can start a new Bitcoin clone, either as a new root with the same code, or by creating something similar based on a different protocol. Seen as synthetic gold, the problem is shared with actual precious metals. Gold supplies may be limited, but people who see it as a store of value could move on en masse to platinum or any other rare material, or alternative sources of scarcity, and thus whatever value it has depends on ebb and flow of fashion.