A question about transaction fee incentives

[aodh]

I'd like to ask a question about transaction fees. I've read at http://bitcoin.stackexchange.com/questions/876/how-much-will-transaction-fees-eventually-be that transaction fees are not set in an effective way by the market (in particular, they would be too small), and that this could be a problem once new bitcoins are no longer generated. I have an idea for how this might be solved, but I am not sure if this is already a known idea, or whether it will actually work.

The problems described at the above link is that there is a "race to the bottom": miners will always want to add "one more" transaction to their block if has a non-zero fee attached, even if it is too low of a fee to cover the cost of mining (since the cost of mining is largely fixed with respect to the number of transactions in the block). As a consequence, people sending transactions wouldn't have an incentive to set a "reasonable" fee. It is an instance of the tragedy of the commons.

My proposal would (maybe) fix this. The idea is: each transaction would set a maximum fee, and the miner would set a fixed fee for the block (per byte). The miner would be allowed to include any transaction such that the fee per byte times the size of the transaction is less than the maximum fee set in that transaction. All of the transactions would then be charged the same fee (per byte), and the remaining amount for each transaction would be sent to an output specified in the transaction.

In this case someone wanting to send a transaction could not make their fee arbitrarily small, since then no miner would ever want to include it (since they would lose out on the fees from other transactions). Additionally this seems more "fair" in some sense, since everyone would pay the same transaction fee.

On the other hand, I have no idea what the equilibrium fee would be under this system. Also, of course, this would be a major modification to the protocol, and so may be difficult to implement in practice.

EDIT: To clarify, my actual question is whether this system has been considered, and whether it would work/what problems there might be with it

[DannyHamilton]

. . .

My proposal would (maybe) fix this. The idea is: each transaction would set a maximum fee, and the miner would set a fixed fee for the block (per byte). The miner would be allowed to include any transaction such that the fee per byte times the size of the transaction is less than the maximum fee set in that transaction. All of the transactions would then be charged the same fee (per byte), and the remaining amount for each transaction would be sent to an output specified in the transaction.

. . .

EDIT: To clarify, my actual question is whether this system has been considered, and whether it would work/what problems there might be with it

Interesting concept.  I don't see it being adopted in bitcoin, but perhaps someone will want to try implementing an alternate currency to try the idea out.
Bitcoin deals with the "race to the bottom" or "tragedy of the commons" that you describe by limiting the size of a block.  If/when there are more transactions than there is space available in the block, then the miner has incentive to only choose those transactions with the highest fees.  If the block can only fit one more transaction, why would a miner choose a transaction for that space that has a low fee when there is another transaction available that has a higher fee that can take the same place?

[DeathAndTaxes]

Also so far there has been no race to the bottom.  Most pools lump in ultra low fee tx in the same category as free tx and they are routinely not included in blocks (much to the frustration of those wanting instant confirmatons and not paying even a penny).   Yes technically the pool is losing some revenue but the amount is a rounding error and it helps to drive avg tx fees higher.

https://blockchain.info/charts/transaction-fees-usd

So there are three things to consider:
1) There may not be a race to the bottom.  

2) There may be alternative methods of compensation.  i.e. a merchant who deals in video games may pay a pool a flat monthly fee to include all tx the merchant receives regardless of tx fees in the next block thus guaranteeing a fast response time and better customer service.

3) The change you proposed would require a hard fork of bitcoin and realistically the chances of that are essentially 0.0%.  Even less controversial changes which eventually have near unanimous support (P2SH) are difficult to get implemented.

It is an interesting idea but probably one best discussed in the context of an alt-coin.  It would also make "fee discovery" simpler.  Miners could publish their min fee requirement and clients could then give users a good idea of how likely confirmations will take.

i.e.
a fee of 1 mBTC will result in 10% of network including this transaction in the next block (~100 min estimated confirmation time).
a fee of 3 mBTC will result in 25% of network including this transaction in the next block (~40 min estimated confirmation time).
a fee of 10 mBTC will result in 50% of network including this transaction in the next block (~20 min min estimated confirmation time).
a fee of 25 mBTC will result in ~99% of network including this transaction in the next block (~10 min estimated confirmation time).

[aodh]

Thanks for the replies. I do realize that it is not realistic to expect this will be incorporated into Bitcoin. I might implement an alternate chain at some point, since there are several other ideas I'd like to test out as well. However, I don't have that much time right now.

As far as the fixed block size goes, as I understand it this is intended to be increased at some point. However, since it is not adjusted dynamically (like the difficulty is), it does not seem that this would both allow for the capacity to increase with demand and also impose enough competition between transactions to increase fees, in general.

Anyway, this may very well turn out not to be a problem in practice. It's unfortunate that this won't be tested until Bitcoin has been around for many years (until the block reward decreases to a small amount).

It would also make "fee discovery" simpler.  Miners could publish their min fee requirement and clients could then give users a good idea of how likely confirmations will take.

That's an interesting idea.

[Bitexchange]

If raising the transactions per block limit would require a hard fork, will this ever happen with bitcoin?
Or will we need an alt-chain for micro transactions?
I dont really like the idea anyway of bloating the blockchain with micro transactions, it could make the whole experiment fail miserably if it takes too much ressources to run bitcoin.
If we had a minimum fee of around 0.5 USD, that would mean around 500 USD reward per block, about the same as we have now with 50 BTC per block.
And personally i wouldnt care if a transaction costs 0.5 USD minimum.