Simplifying fee structure

[domob]

I would like to propose a simplification of the fee structure. Note that I'm talking about ordinary transaction fees here, not costs of names. (Only as much as transaction fees are part of the name cost.) In particular, a big difference between Namecoin and Bitcoin is that Namecoin has a "special rule" of requiring extra fee for each output less than 0.01 NMC. This, for instance, caused some surprise to the Armory port that is going on right now.

Do we really need that rule? As far as I understand it, it was introduced to combat "dust spamming" in very early times. Bitcoin doesn't have this problem because they forbid (by making transactions "non-standard") the creation of too small outputs (the famous 5400 satoshis) right away. I propose to remove the special fee rule in Namecoin, and instead switch to a rule similar to Bitcoin: Transactions with very small outputs (could be 10,000 swartz, just the same rule as Bitcoin uses or something else - remember, the dust spam is actually 1 swartz in size) won't be accepted (relayed / mined) in any case (no matter what fee is attached). They would, of course, still be accepted when in blocks.

Are there strong reasons why we shouldn't do this? I think it makes it easier for all kind of people working with Namecoin if rules are mostly similar to Bitcoin as long as this has no disadvantages for Namecoin.

[virtual_master]

I would like to propose a simplification of the fee structure. Note that I'm talking about ordinary transaction fees here, not costs of names. (Only as much as transaction fees are part of the name cost.) In particular, a big difference between Namecoin and Bitcoin is that Namecoin has a "special rule" of requiring extra fee for each output less than 0.01 NMC. This, for instance, caused some surprise to the Armory port that is going on right now.

Do we really need that rule? As far as I understand it, it was introduced to combat "dust spamming" in very early times. Bitcoin doesn't have this problem because they forbid (by making transactions "non-standard") the creation of too small outputs (the famous 5400 satoshis) right away. I propose to remove the special fee rule in Namecoin, and instead switch to a rule similar to Bitcoin: Transactions with very small outputs (could be 10,000 swartz, just the same rule as Bitcoin uses or something else - remember, the dust spam is actually 1 swartz in size) won't be accepted (relayed / mined) in any case (no matter what fee is attached). They would, of course, still be accepted when in blocks.

Are there strong reasons why we shouldn't do this? I think it makes it easier for all kind of people working with Namecoin if rules are mostly similar to Bitcoin as long as this has no disadvantages for Namecoin.

It sounds reasonable to have Namecoin compatible with Bitcoin on the network protocol level and keep the differences only where is necessary to support name related features.
This is especially trough if it hampers implementing Namecoin support for important software like Armory.
On the other side we should put in weight very carefully if removing spam protection.
What is the actual protection if somebody want to move 50 000 X 2 swartz in one block(50 000 spam transactions) ? And what would be the spam protection in the same case if the Bitcoin rule would be valid instead the actual one ?

[domob]

On the other side we should put in weight very carefully if removing spam protection.
What is the actual protection if somebody want to move 50 000 X 2 swartz in one block(50 000 spam transactions) ? And what would be the spam protection in the same case if the Bitcoin rule would be valid instead the actual one ?

In the current situation, each output costs the "base fee" (I think 0.005 NMC for miners and 0.001 NMC (?) for relaying nodes) in transaction fees that must be paid. In other words, it costs an attacker a large multiple of the amount of dust created. For the Bitcoin rule, such a transaction simply would never be mined or relayed by default. I don't see a problem with spam attacks in either case, but maybe I'm missing some cases.

Note that this transaction could be spammed onto the blockchain "for free" (except the moved coins) in both cases if mined by the attacker.

[domob]

Proposed patch: https://github.com/namecoin/namecoin/pull/201

[indolering]

I don't like the idea of having to trust miners, the dynamics will change once mining rewards asymptote. But if Bitcoin is handling it through the "non-standard" check, I'm okay with sticking to their security model.

[domob]

I don't like the idea of having to trust miners, the dynamics will change once mining rewards asymptote. But if Bitcoin is handling it through the "non-standard" check, I'm okay with sticking to their security model.

What exactly do you mean with "trust miners" here? The only thing that miners could manipulate is that they could mine transactions with "dust spam outputs" but without tx fees. I think it makes sense to allow that, as it keeps the strict consensus rules simple and clean. Enforcing transaction fees with them would mean that we would have to do a fork every time we want to change the tx fee rules.

Also note that this is not related to cost of names (as mentioned above already). Miners can not cheat with them.

[indolering]

What exactly do you mean with "trust miners" here? The only thing that miners could manipulate is that they could mine transactions with "dust spam outputs" but without tx fees. I think it makes sense to allow that, as it keeps the strict consensus rules simple and clean. Enforcing transaction fees with them would mean that we would have to do a fork every time we want to change the tx fee rules.

Yeah, I'm fine with trusting miners on this because it's a fairly minor "attack" that we could plug pretty quickly if we had to.