Unless the tx fee is so high which could compensate the orphans, miner will start to reject tx like what Eligius doing. Eligius is simply putting the burden on other miners and increase their orphan rate.
Although I agree SD is abusing the network, a successful BTC network should be able to handle 1000x of current volume. If SD is a stress test, the network is failed. It means the network cannot grow further.
Without changing the current infrastructure, there are 3 options: 1. Miners will allow only a limited number of tx in a block, but this may leave some tx never confirmed and BTC is dead; 2. Increase the tx fee by 10x or even 100x; 3. A revised tx fee formula to punish people using coins with <3 confirmation.
All these solutions, however, would be short-term. If more people use BTC (which most of us want to see), the same problem will emerge again. If this could not be solved, this will be a major bottleneck of the whole system.
Actually, the TX fee used to be much higher (0.01 per KB vs 0.0005).
It was reduced to 1/200th it's original size after last summer's bubble. At this point, txfees are almost meaningless because of that decision. To process credit cards, you're looking at percentage fees, in addition to $0.05 to $0.30 per transaction, depending on volume and your system's ability to run the transactions in batches and the volume of transactions you process daily. Right now, the txfees are pointless. They don't even amount to a penny per TX in most cases. That's why I actually agree with what Luke has done.
It is not SD's fault, they are following the rules that were changed because the developers were being pressured by the community last summer when TX fees were "becoming too big" because the price was $20-30 for a few weeks. However, even at the peak value and 0.01 per KB, they were almost always cheaper than what would be charged by any credit card/payment processor.
If the fees were still 0.01, this wouldn't be a problem. The fees would be adding enough to each block to make up for the higher risk of orphans. Of course, it would also probably kill SatoshiDice because the fees would be a much larger portion of the bets. Not that Bitcoin should be kept in this state for SatoshiDice. I don't ever recall Bitcoin being marketed as a micropayment system.
And even if fees were bumped back up to 0.01, this doesn't mean everybody is suddenly paying high fees. The network is designed to give priority to large transfers of highly confirmed coins. The only people that would feel the change are those sending coins just after receiving them, or sending extremely small amounts of coins [or paying with a "bag of pennies" in change].
It's only 1/20th, not 1/200th
Transparency of the BTC tx fee system is actually quite low. Although the client will provide a "standard" rate, you will never know if it will be accepted or not. I think the major pools should publish their fee schedule/rules. Fee should be decided by the market, not the devs.
BTC may not be marketed as a micropayment system, but it has a good potential to be one. Tx fee of 0.01BTC may work a this moment. Pushing the tx fee to 0.3USD (0.01*30USD at the peak) will just kill any sub-dollar tx, or force them out of the blockchain (what SD suggested to do). (OK, may be I should start a BTC micropayment service)
Is it possible to improve the protocol, so we could tolerate SD-type tx while keeping the fee low?
