Optimal transaction fees

[adamstgbit]

There's a balance between driving fees high and driving a high number of low fee TX
bacily if we look at block space as a good miners produce, the pricing of that good is the TX fee.
how do we find the best price?
somthing like this:

I believe we dont need to artificially limit supply, since miners have a real cost to including TXs in blocks in the form of orphan risks

Block limit makes Supply is easily known... 1MB.
Demand can be roughly calculated with avg number of TX bytes to include every 10mins
Miners can calculate the optimal price for a TX fee
Miners should choose to not sell block space for much less than this optimal price.

thoughts? am i making any sense?? lol


@Peter R
I often quote your work http://www.bitcoinunlimited.info/resources/feemarket.pdf
I have a 2 questions
1, do you think there will still be a cost to include TXs if thin blocks incress propagation times 100X
2, did you ever plug in some rough numbers into the formula to try and determine what Q* would be ( the free market "max block size" )

 

[67 speedqueen]

@adam,
Supply/demand doesn't work like that. With stuff like money, think Network effect. The more people use it, the more valuable it becomes.
The corollary of the network effect, especially for money, is: doubling the number of users *more than doubles* the number of transactions. If you have a phone and there's no one else in the world with a phone, you make 0 * 1 (zero) phone calls per day. Give your GF a phone (double the number of users) and instead of doubling the number of phone calls (0 * 2 = 0), you [probably] get some positive number, i.e. "more than double."

Same with [non-commodity] money. If you're the only one who has it, it's pretty much useless, so 0 tx/day. Also worthless, because not commodity.
If you and Bees Brothers have it, it becomes marginally more useful, possibly one tx a month. Add more people & the network effect starts kicking in, adding not just their tx, but, since you can now deal with them *and Bees Brothers,* you also add more tx.
It's morning, coffee isn't kicking in for some reason, so if above made no sense, that's why. And totally not because I'm retarded. Because I'm not retarded, Adam.
Pretty shitty of you to even think such a thing.

 

[Peter R]

1, do you think there will still be a cost to include TXs if thin blocks [decrease] propagation times 100X

According to the theory presented in my paper, the "supply cost" due to orphaning is given by the equation:

cost per transaction = (propagation impedance) x (inflation rate) x (size of TX)

Currently the propagation impedance across the P2P network is about 17 sec / MB, the inflation rate is 25 BTC / 600 s and the average transaction size is 500 bytes. This suggests the cost to supply block space for a 500 byte TX is roughly

cost per transaction = (17 / 1,000,000) x (25 / 600) x (500) = 0.00035 BTC ~ $0.14

(The reason fees are actually LOWER than this is because miners mostly connect through the Relay Network so their propagation impedance is significantly less than 17 sec / MB and the figure above is thus too high).

I don't think thin blocks alone will decrease propagation times by 100X, but 10X is feasible. Combined with the halvening, this suggests that fees could drop to about 0.000018 BTC for a 500 byte TX (3.5 satoshis per byte) or about $0.007 per TX at $400 / BTC.

This might seem cheap, but it is priced in BTC. So if the price jumps to $4,000 / BTC, we'd be looking at $0.07 per TX (which is a lot).

2, did you ever plug in some rough numbers into the formula to try and determine what Q* would be ( the free market "max block size" )

Well, except for the last 6 - 8 months, Q* has been exactly what the average block size was at any given time (because the limit wasn't having an effect). I think @Roger_Murdock estimated that Q* is probably around 1.5 MB right now.

 

[AdrianX]

I've been letting this idea mature in my subconscious. Everyone has a different subjective value use for money. So not all transactions of the same value are equal.

There is also the idea that the money supply sould not decrease or increase (stuck transactions) based on this subjective evaluation.

If someone sends a transaction with a fee that's too low how should it be handed. Also how should an appropriate fee be judged.

The ideal solution should reflect the demand side that is valued by the users.

And the supply side by the cost of the network to relay a transaction.

Transaction fees should also send intimation to miners to allow them to prioritize transaction for maximum network value and minimize orphan risk.

I've been critical of BU blindly copying Core's fee policy so I have an interest to know what others think as I think a BUIP will be needed soon.

 

[adamstgbit]

block space is a resource, and miners including your TX is a service, that service comes with a cost. ~$0.007 / ~265 bytes ( 1/2 a cent / TX ) (( assuming thin blocks become the norm ))

and thats exciting!

it means a free market can be allowed to form around blockspace without limiting blockspace!

0.0007$ is the cost of including 256bytes into a block, but that just cost of production. this is != to the price at which they should sell block space... they should try and sell block space for a price that will bring the most profit, not the min price, my idea is to calculate the best price / TX based off of age old economic theory 101

BU should throw away Core's TX policy and take all this into account this idea .


i think miner can be made to agree its in there best interest to not include a TX much below Optimal price,

and for sure they can be made to agree its not profitable to include TX below the cost of production 0.0007$ / 265 bytes.
[doublepost=1459140083][/doublepost]i'm very excited by these theories which demo. a real cost to including a TX.
and i think we can take it one step further and calculate "Optimal price"

BU is the way to go!

i am very much onboard for limitless block space because i KNOW its wont lead to 1BG blocks, cuz the cost of producing a 1GB block is simply to high! ( given today's internet speeds )

*lol i was quite drunk last night ( easter dinner ), but i think this all makes sense! lol"