Quote from: cypherdoc on January 05, 2015, 06:14:39 AM

how does the IBLT scale with the size of the ~~block~~ UTXO set? linearly or by some other ratio?

edit: actually iirc, it scales with the size of the difference btwn UTXO set estimates across the network. iow, a UTXO set estimated to be 99% similar across the network will allow a miner to send a smaller IBLT when solving a block than a UTXO set estimated to be only 89% similar across the network. is that right?

Yes. The size of an IBLT does depend upon the average expected number of differences between the unconfirmed tx mempools of the majority of the nodes. Because each node has independent choice on which tx to accept or reject then it is impossible to be precise, in advance, about how efficient IBLT will be. Miners will always want to propagate small blocks, so there will be an incentive for mining nodes to find consensus on their tx mempools. Gavin describes it:

Quote from: Gavin Andresen on October 27, 2014, 01:27:22 PM

RE: O(1) versus O(some-function-of-total-number-of-transactions):

Yes, it will depend on whether or not the number of differences goes up as the number of transactions goes up.

The incentives align so it is in everybody's best interest to make the differences as small as possible. I wouldn't be surprised if that causes innovations to drive the actual size to O(1) minus an increasing constant, as code gets better at predicting which transactions our peers do or don't have.

It is also worth noting that IBLT can be implemented on top of other block propagation models, such as Matt Corallo's block relay system, which already saves on bandwidth.

Quote from: Peter R on January 05, 2015, 06:17:45 AM

Agreed. I didn't mean to understate the importance of IBLTs; I just wanted to point out for readers that IBLTs only improve block propagation (still an important problem) and that each TX still needs to be processed by each node (linear BW scaling with TX/sec).

Thanks for making the clarification.

Quote from: tvbcof on January 05, 2015, 06:56:06 AM

Quote from: solex on January 05, 2015, 06:10:13 AM

Yes, but "simply" is an understatement. Real-time tx cross the network uniformly, with enormous capacity already available, perhaps 1000 TPS. It is block propagation which is time critical where milliseconds count. When a block is solved the rest of the mining network is then hashing uselessly until the block is fully propagated. Incentives are perverse, against large blocks which are ultimately needed to fund the network via fees rather than rewards. Removing the bottleneck for block propagation is a huge win for scaling Bitcoin.

Doesn't matter if the transaction fees are high or low. Doesn't matter if the inflation-based reward 50 BTC or 0 BTC. Doesn't matter if a block is empty or filled with 100,000*60*10 transactions. The amount a sha256 miner nets approaches zero. And if Bitcoin valuations fluxuate that net will drop below zero. This due to the simple fact that there is no limit on supply of gear. Only a lag.

None of the Libertarian brain-trust (justusravnier, cypherdoc, etc) want to touch this one. Surprise, surprise.

The answer I was waiting for is that actually mining will be profitable for some (marginally) and not for others depending on various factors (operational costs, economies of scale, ability to leverage otherwise idle resources, etc.) What will almost certainly happen will be that alternate revenue streams will be tapped. In http-land one such tap is known by the metric called 'eyeballs' and it's why you see ads all over the place. Another tap is for intel. Some of it you see when you notice targeted ads...some of it you don't (normally) see at all... 'Where who spends what' is an especially rich vein of actionable intelligence, BTW, and one the operators can easily get once they've got control of the system generally.

The salient point is that the only outfits who will be competitive in providing this support (and currently Bitcoin is wholly dependent on sha256 mining) are the companies who are already giants. Perhaps also minor giants who sell intel to the major giants (both of whom are under the thumb of Big Brother.) Well OK fine you might say. But I've already got a Visa card and a PayPal account so yet another such thing doesn't hold much interest for me.

What does pique my interest is a solution which can be feasibly operated covertly in people's garages scattered about the world in a SHTF scenario. That's still the case for Bitcoin at the present scale an it's something I don't want to lose.

To be honest, I have not previously considered the points you make above. I have always taken Satoshi's block rewards->fees evolution at face value, and am not convinced that the majority of miners will work for zero income. Obviously this does happen to some of them, occasionally.

A real-world analogy must be the oil market. With the oil price down 45% a lot of producers are unprofitable, pumping for zero, and considering closing down. But there are profitable producers and the oil industry will continue, even if the price stays low and half of the producers go out of business.