Discussion about the transaction fees

the amount depentent fee has the downside to split a large amount into a lot of smaller ones when you want to send lets say 1M nubits. this will have a spam-like result!
i think ppc has a correct anti-spam fee

1 Like

You can reverse the curve right? so it will be much more efficient anti-spam fee than the fixed 0.01NBT.

But lets for now stick with the constant Relative fee (With a minimum 0.01NBT) and keep it as low as possible to have the competitive edge.

We currently have about 20 transaction daily and average 600K NBT daily trade volume, so lets compare the two schemes assuming this pattern will continue:

    Daily destroyed NuBits (NBT)     ||        Fixed Fee (NBT)    ||      Relative Fee (%)

                                        0.2                                                         0.01                             3.33E-005
                                      25.6                                                        1.28                             0.0042
                                    204.8                                                     10.24                             0.0341
                                    819.2                                                     40.96                             0.136
                                  6553.6                                                  327.68                             1.0
                               52428.8                                                2621.44                             8.7

As you can see achieving the same amount of profitability with fixed fee is Impossible.
We need to vote on 40 NBT to have the same profits of as little percentage as 0.14%!

Even if this pattern radically changed i doubt that the Fixed fee profitability could ever get close to the Relative fee.

May be i am wrong about this comparison, but it all depends on how NuBits would be mainly used as;

Main UsageProfit model

Stable store of value … Small inflation or Reverse parking rate
Arbitrage facilitator … Percentage fee
Digital transactions facilitator … Fixed fee per transaction

In this Article it seems that Paypal have about 7M daily transaction, while having about 300M daily transaction volume.

If NuBits acted like a money transfer facilator and managed to achieve paypal level, then a Fixed fee of 0.1 NBT will be enough to make Nu very profitable as a Percentage fee of 0.1% could do, and i don’t think that 0.1 NBT will have a significant effect on Nubits effeciency at all.

I hope Nu the best.

i am in favor of the fixed fee. is there a motion about it to vote?

I’d like to reopen a discussion about fixed fee versus variable fee. I think @Raythma’s point is valid that a fixed fee is not useful as a revenue source, while a variable fee is.

Also, as fiat is normally something like 3% fees, we are easily still competitive with even 0.5% fees.

1 Like

Continuing the discussion from the last post @Creon made here: Another paid article about BitShares?

I don’t expect 7.5% monthly to be competitive in the long-term. Your excellent work pioneering liquidity pools is helping to ensure that new entrants will soon begin providing liquidity at cheaper rates if they are able to.

The level of transaction fees can certainly be debated, but I absolutely believe 0.10 NBT per transaction would be a competitive transaction rate at this point in time. NBT are not yet being used for microtransactions or purchasing $0.50 US bags of candy in department stores. It appears to mainly be used for hedging large amounts of USD against Bitcoin trading. The value that NBT is providing these users is (I believe) far in excess of what we are charging them to use our product.

Analysis of the average NBT transaction size would give us a better indication of what an appropriate fixed rate should be if percentage-based fees are technically infeasible.

2 Likes

So you’re saying that for now it might be best to charge a higher transaction fee, since the majority of people are currently using NuBits to transfer larger sums of money. Therefore, they wouldn’t even notice the $0.10 cent fee in comparison to the amount of money they’re putting into their transactions?

If that is true, it would buy us some time to develop other services or ways of earning profit for the network. As our liquidity operations become sustained more and more through these other income sources, we could continuously lower the transaction fee until the point where we’re competitive with other cryptos for microtransactions. Is this what you have in mind @tomjoad?

Its more 20 cent we are talking about than 10 cent. This is already quite a sum and would bite me if I see it on the bill of my $1.10 coffee order. On the other hand, the current design won’t ever allow such microtransactions with Nu, so maybe we also should try to target other groups like @Sentinelrv suggests, i.e. people who want to store value in a decentralized way with a stable value and only to transfer it in “exceptional” cases. A bit like a savings account.

For those people 20 cent per transaction are surely fine. The problem however is, that I made my calculations with the current transaction volume. What effect would a 20x tx fee have on this? And if sending NBT around is our revenue, how do we encourage that? The daytrader who keeps all NBT on the exchange will never pay this fee.

More accurate data is probably required, which I am not able to produce right now. How long will it probably take until we have the tx fee as voting parameter? Since this involves a hard fork, I assume it needs to wait for the other protocol changes to be implemented as well…

I wasn’t suggesting anything. I was only trying to interpret what @tomjoad was saying. I believe he means our current user base is using NuBits to transfer larger sums and that these people wouldn’t mind the larger transaction fee. They would only mind the larger fee if they were trying to transfer smaller sums of money, purchase a cup of coffee, or perform microtransactions.

I was trying to say that targeting people that transfer larger sums with a higher transaction fee would tide us over for a while, but it should only be a temporary rise in fee, not a permanent one. We should develop services and other ways of earning income for the network, which we could then use to pay liquidity costs. With the money to pay for liquidity coming from other sources of income, we could then gradually lower the transaction fee to the point where we’re once again targeting people who transfer smaller sums of money.

In order to continually lower the transaction fee back to normal, we would need to find enough alternate sources of income to completely cover our liquidity costs. The higher fee would not be needed anymore at this point.

Why not percentage fees, I really don’t understand.

In my opinion the .01 NBT is fine for a minimum fee to prevent blockchain spam. For the staking system, however, we should be voting on percentages not flat fees.

A fixed relative rate will discourage large transactions just as a fixed absolute rate discourages small transactions. I don’t see how the rate table shown by @Raythma, which is a similar approach like in parking, can be applied here. If Nu charges 5% for 100k but only 2.5% for 50k, well, then I just send two times 50k. Not sure why anyone would pay the larger fee if you can just split your transaction.

The current approach isn’t bad in my opinion. Note that the transaction fee we are voting on is not the fee someone has to pay for each transaction, but is a scaling factor of the size of the transaction. If you are sending large sums then it is very likely that you will need many UTXO’s for this transaction and therefore you will also pay a larger fee, even right now.

My question is more what a good value what look like. I think we kind of agree here that 0.01 NBT is too small. The very rough calculation showed that 0.20 NBT could be sufficient, but that’s a lot. So what is the best trade-off?

1 Like

I agree he gave a bad example. What if it’s a smaller % for a bigger volume? Like 1% for 1 NBT, .5% for 10 NBT, .25% for 100 NBT, .125% for 1000NBT, and so on. .01 NBT for any transactions smaller than 1 NBT.

Honestly, I don’t think this needs to be variable, I think we just need to pick a good curve. The equation for this example, by the way, is:
Fee = 1% / [log(Volume)+1]

Tx Volume (NBT)Fee(NBT)Fee(%)

    0.01............... 0.01.............100%
    0.1................ 0.01.............10%
    1.0................ 0.01.............1%
    10................. 0.05.............0.5%
    100................ 0.25.............0.25%
    1,000.............. 1.25.............0.125%
    10,000............. 6.25.............0.0625%
    100,000........... 31.25.............0.03125%
    1,000,000........ 156.25.............0.015625%
3 Likes

I don’t see how this is relevant. The vast majority of “transactions” that matter to hedgers are handled off-blockchain on the individual exchanges.

2 Likes

It just depends on user behavior. If the user is continually depositing and withdrawing NBT from exchanges the transaction fees are relevant. If they leave NBT on the exchange then yes it’s probably not as relevant. Although, if a trader is doing this behavior then I assume USD would be be preferred over NBT?

1 Like

Perhaps, but it is easier to (conceptually) deal with 11,000 NBT in an account versus $11,000. Many exchanges don’t apply KYC/AML protocols to crypto <> crypto trading.

1 Like

a bit a la paypal.

I like @Nagalim’s approach very much, but I’d vouch for a different lower cap for the fee.
I’m not sure whether that will disincentivize small transactions too much (interfering with what the mobile wallet tries to accomplish), but as the block chain can’t deal with huge amounts of micro transactions any way, it might at least be suitable for the far future of Nu.

Tx Volume (NBT) ....Fee(NBT)........ Fee(%)
0.01............... 0.05.............500%
0.1................ 0.05.............50%
1.0................ 0.05.............5%
10................. 0.05.............0.5%
100................ 0.25.............0.25%
1,000.............. 1.25.............0.125%
10,000............. 6.25.............0.0625%
100,000........... 31.25.............0.03125%
1,000,000........ 156.25.............0.015625%

Right now @Nagalim’s fee approach might suit the situation of the Nu network better.
Paying a fee that is derived from the transmitted amount and not from the size of the transaction seems like a fair approach in terms of paying a price for a service, but what would that mean for the protocol?
As I’m no developer I think deriving a tx fee from the transferred value and not from the tx size is no big deal - I might be horribly wrong with that…

1 Like

With the release of Nu 2.0, this seems like a good time to revive discussions about transaction fees. I’m currently voting for a 0.10 NBT fee. I believe this amount would be reasonable even on a 1.79 purchase of a cup of coffee with NuDroid, and is still very cheap for those users who are using NBT to hedge against BTC in large amounts.

I think it’s important for shareholders to realize just how important our ability to set the transaction fee is. We provide a transaction service to people in the form of a secure and price stable cryptocurrency. Keeping the peg stable costs shareholders effort and money. Our business model consists of our ability to continually increase demand and sell newly created NuBits to people that have use for a dollar equivalent cryptocurrency.

Variable transaction fees give shareholders the ability to charge for the peg service we provide our users. As we raise transaction fees, NuBits are destroyed, which naturally reduces the overall supply. Decreasing the supply allows us to create and sell more NuBits to people when needed, thus bringing in profit to the network. This ability to charge for the peg service we’ve been providing for free has been missing since launch, but now we’ll finally have access to it. In a way it’s similar to BlockCredits being consumed in B&C, except with the added complexity of the peg and the fact that NuBits have more use cases.

Now, it’s also important that we don’t abuse this feature. What we need to do is find a transaction fee rate which is acceptable and doesn’t discourage the flow of transactions too much. To find this rate, it would be great if we had access to some data from the blockchain, such as the number of transactions every day, that way we could measure the effect our fees have on users of the network. I imagine stage 3 of the website was supposed to give us access to historical data like this, but it appears that it has been reprioritized in favor of parametric order books…

[quote=Jordan Lee]I asked desrever to prioritize implementation of parametric order books over the data service ChicagoSchooler is asking about. Parametric order books are needed to bring NuBit liquidity to pairs other than BTC and fiat (USD, EUR, etc). This feature will be important to the success of B&C Exchange as well as Nu.

In my mind it is questionable whether it should be funded at this time. It is estimated to cost around $10,000. The decision to fund that or not is ultimately up to shareholders. They can fund it directly with a custodial grant if they wish.
[/quote]

What I’m saying here is that we shouldn’t set transaction fee rates blind. Shareholders should have access to some way of easily viewing historical data from the blockchain, so that we can make better and more informed decisions.

2 Likes

I made a python script that looks at Tx history when I was doing the variable Tx fee analysis. It’s not super efficient (takes half a day to analyze the whole blockchain) but it should be usable for some estimates. I’ll upload it tonight.

As for what I’m setting for tx fees, I’m a fan of 0.02 nbt fee and 10 nsr fee. That way we more accurately represent the 2:1000 ratio of the nsr price.

1 Like