Maybe this is making blockchain.info choke. I get 404 on many requests to http://blockchain.info/unconfirmed-transactions
Last time i looked there were over 2000 unconfirmed transactions.
Topic: z (Read 3027 times)
So does more tx mean it takes more time/power to solve a block?
In theory yes, in reality no.
Every tx added to the block requires one additional has for the merkle tree but the amount of computational work in creating and validating a block is a rounding error compared to the quadrillions of clock cycles spent (collectively) looking for a hash which solves the block.
Let's see, 50 BTC mining reward / 1322 transactions = 0.0378 BTC which would be the fee per transaction in the future in order to keep miners motivated to secure the network, I think that's acceptable.
OK how about if bitcoins were worth significantly more than they are now. Lets say they are worth $20 to $50 each. Do you want to pay $1-2 per transaction. People bitch now about a few cents extra for credit card or a cash discount.
It's quite impossible for bitcoin to exceed, or likely even approach, a condition that would result in transaction fees exceeding those of credit card companys; for no other reason than bitcoin users would switch back and forth as it suits them, and the more it costs the fewer people it suits.
That said, higher numbers of transactions per block also spreads the burdens. The more (paying) transactions that a block includes, the more miners are going to stay after the block reward decreases. No miners are required to include free transactions, so if there are more paying transactions in the queue than could fit within a single block, free transactions are simply going to have to wait till the rush passes. That's fine for donations, or transactions between trusting parties, but higher transaction values & those with limited mutual trust are simply going to have to pay the market transaction costs to get things done faster. If it's cheaper in transaction costs to buy your new car on credit, people will do it. If it's cheaper to pay the bitcoin transaction fee and wait an hour, people will do that. It's futile to try to guess how such contrived scenarios would work out, because the market would never let things get that far.
This will not come to pass.
Way too complicated indeed.
I disagree. Bitcoin is complicated, not just this pruning/compression solution. I'm not saying that the linked thread is "the best way", but I assure you there will be no simple way to do it.
However, the solutions are usually very simple in concept: represent the current state of the network as only the remaining unspent coins, you can throw away everything else. The details are in the algorithms for computing these unspent-output-trees, and sharing/verifying answers with other nodes.
I used to think this was a pipe dream, because it would cause a hard fork to implement this securely in the network (because you'd have to modify the headers), but the linked thread brings up the fact that it can be done in a parallel blockchain with merged mining. This means that the pruning/compression would be strictly opt-in: users who don't care only use the regular network. Nodes that want the compression/pruning, can subscribe to the second chain and only download the most recent unspent-output-tree and latest blocks.
With this in mind, it's actually feasible. And if there's a strong enough incentive -- like ensuring decentralization by enabling home computers to still participate in the network as full nodes -- then something should be done, even if it's not exactly this solution.
I had a similar reply started but this says it perfectly. A "rolling balance" alt-chain is a very interesting idea and complementary use would present a lot of potential benefits. What would need to be worked out is how to pay miners to mine the chain independently of block rewards and transaction fees paid by the bitcoin network, assuming another alt-coin is undesirable.
On the other hand, if the difficulty of a balance chain is 1000x higher maybe it could spit out a single indivisible "bitdiamond" every time a block is solved and do so indefinitely. That would appease the deflationary-spiral folks too...
So does more tx mean it takes more time/power to solve a block?
Nominally, yes. Practically, no.
So does more tx mean it takes more time/power to solve a block?
More TX mostly means more HDD space for storage, and a little more CPU usage and disk usage for verifying the blocks when downloading the chain.
OK how about if bitcoins were worth significantly more than they are now. Lets say they are worth $20 to $50 each. Do you want to pay $1-2 per transaction. People bitch now about a few cents extra for credit card or a cash discount.
If BTC was worth 10x as much fees 1/10th as high would pay the miners the same amount (in fiat terms). Since miner's costs are 99.9% fiat based the nominal amount of BTC is irrelivent.
Today miners get paid ~$250 per block. In this block that worked out to $0.18 per tx. Given Bitcoin is in its infancy it is certainly possible the network can reach equilibrium at 10x as many tx and 2x higher block rewards.
Something like ~15,000 tx per block and $500 reward =~$0.04 per tx.
Let's see, 50 BTC mining reward / 1322 transactions = 0.0378 BTC which would be the fee per transaction in the future in order to keep miners motivated to secure the network, I think that's acceptable.
OK how about if bitcoins were worth significantly more than they are now. Lets say they are worth $20 to $50 each. Do you want to pay $1-2 per transaction. People bitch now about a few cents extra for credit card or a cash discount.
This will not come to pass.
Way too complicated indeed.
I disagree. Bitcoin is complicated, not just this pruning/compression solution. I'm not saying that the linked thread is "the best way", but I assure you there will be no simple way to do it.
However, the solutions are usually very simple in concept: represent the current state of the network as only the remaining unspent coins, you can throw away everything else. The details are in the algorithms for computing these unspent-output-trees, and sharing/verifying answers with other nodes.
I used to think this was a pipe dream, because it would cause a hard fork to implement this securely in the network (because you'd have to modify the headers), but the linked thread brings up the fact that it can be done in a parallel blockchain with merged mining. This means that the pruning/compression would be strictly opt-in: users who don't care only use the regular network. Nodes that want the compression/pruning, can subscribe to the second chain and only download the most recent unspent-output-tree and latest blocks.
With this in mind, it's actually feasible. And if there's a strong enough incentive -- like ensuring decentralization by enabling home computers to still participate in the network as full nodes -- then something should be done, even if it's not exactly this solution.
Let's see, 50 BTC mining reward / 1322 transactions = 0.0378 BTC which would be the fee per transaction in the future in order to keep miners motivated to secure the network, I think that's acceptable.
+1 interesting observation 
Let's see, 50 BTC mining reward / 1322 transactions = 0.0378 BTC which would be the fee per transaction in the future in order to keep miners motivated to secure the network, I think that's acceptable.
and we are soon going to enter a phase where transaction fees are necessary for miners to mine profitably;
once transaction fees are about 5% of the total block reward, miners who mine empty block will not cover their costs.
Not everyone needs to use the Satoshi client. Nobody uses the original BitTorrent client either. It's clear that running Bitcoin as a full node is going to require a bit more but I don't see that as a serious problem.
We need some optimization though and I think that SatoshiDice is the perfect incentive to start doing something about it.
As far as it being a problem currently, it is really not. Our blockchain is growing faster but the blocks themselves still have a lot of room for transactions on average. None of the blocks are hitting the maximum size.
We need some optimization though and I think that SatoshiDice is the perfect incentive to start doing something about it.
As far as it being a problem currently, it is really not. Our blockchain is growing faster but the blocks themselves still have a lot of room for transactions on average. None of the blocks are hitting the maximum size.
I see no reason why the network won't be able to handle the transactions. But the hardware required will be more and more consequent...
The network should be fine, but I'm increasingly worried
about the satoshi client and the fact that it needs the whole
blockchain locally.
Agreed. There's a thread addressing this exact issue. Although I don't understand exactly how it works, it seems like a very good solution.
Here's the link: https://bitcointalksearch.org/topic/proposal-a-second-chain-for-scalability-83645
satoshidice....
will ironically be the death of bitcoin
Jump to: