In short, Mike Hearn and others figured out how to exploit Bitcoin's native scripting capabilities to support microtransactions. They call it microchannels, and there's a very succinct explanation here if you're familiar with Bitcoin scripting: http://www.slideshare.net/JohannBarbie/bitcoin-micropayment-...
There's even a working implementation: part of Mike Hearn's BitcoinJ, the second oldest Bitcoin implementation.
Now, you could argue that there's so far no big demand for Bitcoin microtransactions, and I'd have to agree. But to say bitcoin is not capable of them is simply false.
Edit: Also, your per tx fee is too high by at least 10x. In fact, many transactions can be sent at absolutely no charge.
Micropayments that aren't channels will be possible when fees are lower.
As it stands now there is no universal standard, which is needed before you can say bitcoin is ready for microtransactions. It's a market issue, not a technical one.
New applications are exactly why a (non-Turing complete) scripting language was included in the protocol -- so people could extend the Bitcoin to new areas.
You may not like Bitcoin, but calling the use of a feature that was built by the original developer to support one of the earliest aspired uses of Bitcoin "patchwork" is pretty unfounded.
And regarding Stellar, I'm a big fan of the Ripple technology, but very cautious about the entities behind both Ripple and Stellar. Just like Stellar, Ripple started out with promises of distributing the vast majority of the coins around the world, and yet 2-3 years out this still hasn't happened. The fact is -- unlike Bitcoin's -- Stellar's and Ripple's distribution model is highly centralized and thus potentially undependable.
In addition, Graydon Hoare (creator of Rust) has also joined the Stellar team from Mozilla. Now that Stellar supports over 2.3 million wallets, making it the largest implementation of this system, we have found certain scaling limitations of the core tech. Graydon is one of the developers spearheading efforts to make Stellar capable of scaling well beyond its current limitations.
Once these two efforts are complete in a few months, I think you will feel much more comfortable with the robustness of the Stellar protocol.
https://bitcoinj.github.io/working-with-micropayments
In short, the payer sends a signed transaction to the payee every time they want to increment the payment, but the payer does not broadcast the transaction to the network until they want to finalize it. To prevent double-spending, the payer and payee first set up a 2-of-2 multisig, so all transactions must be signed by both parties.
This lets applications send many instant micropayments, and only one transaction fee is paid.
In the end bitcoin needs to handle around 1,000 independent transactions a second to see widespread adoption and even 100,000 transactions a second is low for some use cases. NYSE for example peaks well above that.
The above is subject to malleability-resistant payment channels, which will require a patch like Peter Todd's CHECKLOCKTIMEVERIFY (https://github.com/petertodd/bips/blob/checklocktimeverify/b...)
BTW, this does not require the rapidly adjusted transaction technique.
Bitcoin will become something like the international wire transfer mechanism.
One implementation of chaum ecash is open-transactions, i suggest you check it out: http://opentransactions.org
> Bitcoin can't handle micropayments
It already does.
> The block chain is limited to about 7 transactions per second
No, it doesn't. That limitation is completely artificial and can be removed at any time. The core developers are working on scaling the system to 1000+ tps.
> and it's currently about $0.30. That's more than many merchants pay to process a credit card transaction.
No, it isn't more. And it's $0.03, not $0.30. Have you looked at the actual credit card fees? They consist of a percentage + fixed fee per transaction + fixed monthly fee.
http://credit-card-processing.findthebest.com/
> Nor can Bitcoin handle fast payments. It takes about an hour to get solid confirmation of a transaction
The transaction takes a few seconds. One confirmation takes around 8-9 minutes on average (we always mine faster than the 10 minute goal).
Compare it to a bank - transaction takes days, confirmation takes 6+ months.
Compare it to a credit card - transaction takes a few seconds, confirmation takes 180 days.
> * you're selling anything that actually costs you money, you have to wait for some confirmations*
Same with credit cards and bank transfers. Except these kinds of payments can be reversed for 6+ months, and it's a big headache for the merchants and stores.
Two use cases:
1. Micropayment channels. When you pay to a single party in frequent small chunks, you can create a transaction that locks up certain amount upfront and than you simply re-sign that transaction moving more coins to the recipient and less back to you ("change"). Recipient does not publish transaction until the session is done. If you disconnect or they decide to stop providing you with a service, they simply release a transaction with the latest balance of payment and change.
2. Micropayment channel is not always relevant. E.g. you want to make a one-time small payment. E.g. buy a drink in a cafe for $2. We can imagine another micropayment protocol similar to Ripple where IOUs are transferred, but unlike Ripple, each IOU is ensured by at least twice the amount of real bitcoins locked up in 2-of-2 multisig transaction by each pair of nodes that exchange IOUs. In other words, it's a Visa-like clearing house, but it's decentralized (mesh-like) and every penny of debt is fully insured by real bitcoins being held hostage in "joint escrow" transaction (2-of-2 multisig).
That is an inaccurate statement. Technically speaking, sending small amounts of Bitcoin from one source address to one or more destination addresses in separate transactions are inefficient, given the default size of the transaction(s) and the cost of processing. Depending on your use cases, there are solutions to this apparent limitation.
To handle one-to-one micropayments, Bitcoin microtransactions can be done in a trustworthy manner by using rapidly adjusted payments between parties[1]. Basically, funds are held in escrow and the amount owed is adjusted in a variable off-chain transaction until either a) the funds are exhausted, b) the parties decide they are done, or c) a timer expires.
Second, Bitcoin allows for making one-to-many payments in a single transaction[2]. You could send sub pennies (US$) amount of value to 1,000s of addresses in a single transaction. The cost of the transaction goes up as the size of the transaction goes up (more addresses == bigger transaction), but it remains a very economical way to send very small amounts to lots of addresses.
I'm using the latter of the two methods to send coin to virtual appliances controlling OpenStack cluster based instances, which would have an analogy of feeding parking meters in the physical world. You need to trust the payment party (entity paying for all the instances/parking spots in a single transaction), so the plan is to make the payment mechanism distributed by deploying multiple payment nodes onto the network. Payment nodes deposit pennies into instances/parking meters periodically from wallets they create and control themselves. If someone hacks a single payment node's guest machine, and steals its private key, they'd only score a few bucks. Not worth the time or effort of hacking the hypervisor, in theory.
An interesting side note here, payment nodes could also be aware of their own Bitcoin instance payment address, effectively allowing them to use a wallet they are carrying to pay for their own existence.
[1] https://en.bitcoin.it/wiki/Contracts#Example_7:_Rapidly-adju...
1. Transaction times are long (6 blocks to confirm)
2. Transaction Fees of 0.0001 BTC could likely incur for transactions smaller than 0.01 BTC. This is about a $0.03 USD transaction fee for transactions under $3.30.
My solution to these limitations is to trade BTC on a distributed, low-transaction fee, fast confirming exchange.
That is why today I launched a BTC to Stellar gateway for my p2p microlending startup at https://onecred.com
I expect to see a lot of things people never imagined were possible once bitcoin's store of value combines with a distributed exchange that is good at the things that consumers need (speed, low fees, security, decentralization).
> Bitcoin can't handle micropayments. The block chain is limited to about 7 transactions per second,
Note that this is not a technical limit, it is an arbitrarily set parameter. Gavin argues it could be increased at least 1000-fold with more professional infrastructure over time. Even with a tenfold increase, an average PC with mediocre broadband could still process every transaction ever. I'm confident that as soon as the need arises, this limit will be increased.
> and every node has a full copy of the block chain, so traffic goes up roughly as the square of usage.
That's not true, traffic goes up linearly with the number of transactions. If there are twice as many transactions, twice as much data needs to be sent around in the network. It is also linear in the number of full nodes running - but most users do not run a full node.
> There's a minimum fee for each transaction (if you want it confirmed in any reasonable length of time) and it's currently about $0.30. That's more than many merchants pay to process a credit card transaction.
Wrong again. At the moment, the recommended minimum fee is 0.03 USD, or 0 for high-priority transactions (those who move large amounts or old coins). It used to be 0.001 BTC, but that was reduced to 0.0001 BTC. Also, most merchants have to pay much more than 0.30 USD to process credit card payments.
> Nor can Bitcoin handle fast payments.
Firstly, when looking at throughput (how many times can the same Bitcoin be spent per day), Bitcoin beats all existing payment systems by a wide margin. Even with SEPA, which is praised to be very fast, you can resend the same euro to a new account at most once per business-day or about 20 times per month. A Bitcoin, however, can be moved thousands of times per day (assuming movements between trusted parties) or dozens of times per day between untrusted parties. So regarding throughput, Bitcoin is the clear winner.
Secondly, there is the time it takes until the money arrives with certainty. For credit card transactions, that time is 60 days. For bank transfers, settlement is T+2 days. That means if you send 100 USD from a Lehman brothers account to a BofA account on Monday, and Lehman goes bankrupt on Wednesday, you won't receive anything. With Bitcoin, it takes about an hour to reach that level of certainty.
The problem you are referring to is that by colluding with a miner, someone could issue a competing second transaction that invalidates the first. Fortunately, such attacks require effort and are perfectly detectable. So if not much is at stake, one can accept a transaction as soon as it has spread to the relevant miners. For example, when you order a coffee in a restaurant, it is much easier to simply walk out without paying than to try to launch such an attack.
> Meanwhile, Bitcoin for vending machines, music tracks, and parking meters isn't going to happen.
Exactly for those applications, it is safe enough. None of them is 100% safe anyway: vending machines can be fed with fake coins, music tracks can be pirated, and parking meters can be ignored (betting that the police won't check them). In all those cases the effort to cheat with traditional means is smaller than that of colluding with a majority of miners (and you need a majority for a reasonable rate of success).
Why hasn't the need arised yet? People have been saying this since the limit was introduced, and it hasn't changed.
> Wrong again. At the moment, the recommended minimum fee is 0.03 USD, or 0 for high-priority transactions (those who move large amounts or old coins).
This is before the next halving. Right now, miners get 25BTC for completing a successful block. Satoshi expected that at the time this became halved, fees to the miners would take over. So sending 0.01BTC wouldn't make sense since they'd need 12.5BTC in fees to make it profitable.
The reward will be halved in two years ( http://bitcoinclock.com/ )
How does that make sense in a peer to peer environment?
[1] http://bitslog.wordpress.com/2014/02/17/5-sec-block-interval...
Nor can Bitcoin handle fast payments. It takes about an hour to get solid confirmation of a transaction, by which time the price of Bitcoins may have changed substantially. If you're selling anything that actually costs you money, you have to wait for some confirmations. One gambling site tried accepting unconfirmed transactions. People would bet, then, if they lost, double-spend to kill the payment.
A distributed digital currency without these problems is technically possible. That may be the currency that actually becomes useful for micropayments. Meanwhile, Bitcoin for vending machines, music tracks, and parking meters isn't going to happen.