Block space as a commodity (a transaction fee market exists without a block size limit)

[Peter R]

I've thought about ways to try to estimate the "lost fee income too," but I don't see how there's enough information to make a reliable estimate. Like @awemany says, I'm not sure we'd see a downward slope for the actual fees paid. There is certainly a downwards slope on the demand curve--and if we knew what the demand curve looked like perhaps we could do as you suggest--but I'm not sure how to figure that out.

 

[solex]

I'll have a go at getting the raw data for 2015 and plotting it with fee/byte.

 

[awemany]

@solex, yes, that's exactly the right approach. Graphs that show them how much money they could make by ... you know ... actually growing the market.

I think maybe we should start brainstorming here about what other ways we could use to approach the miners with some convincing graphs.

Something directly saying: This is how much you'd lose.

Another simple idea would be to look at how elastic the fee price actually is (as many people seem to use wallet defaults) and extrapolate the steadily increasing transaction rate in the future (lin and/or or exp). Do it with a 1MB cap and without and one could argue that so and so much fee income would be lost. With the reward halvening, an implemented blocksize limit increase and a (hypothetical) good bubble, I could imagine we're actually not too far off anymore from potential fee/reward parity.

 

[solex]

Initial results look interesting. This is about a 4-week period which has been very busy. Plotted are 3819 blocks (4058 excluding 239 empty blocks with no propagated tx) There is a definite curvature in the data with blocks about 600KB being slightly the most profitable measured by fees per size..

A linear trendline is nearly horizontal. A polynominal trendline does pass through 0 at about 1.7MB. Is this significant? It would seem to imply that without the 1MB some blocks would be occasionally produced up to 1.7MB.and that some fee income is lost. But is that fee income made up for by higher fees within the 1MB? Probably. Is fee-paying business already being discouraged? Probably also.

Thoughts welcome.

 

[Zangelbert Bingledack]

@solex

By "made up for by higher fees" did you mean "offset by higher fees"?

Seems total revenue would always be higher without the extra constraint, though I guess that takes into account additional transactions that would be made if fees were lower, but that we cannot see because the fee pressure hasn't allowed them to come into existence.

 

[solex]

@Zangelbert Bingledack

Yes. I could have been clearer: offset by higher fees. Total revenue for miners should be higher when uncapped, but this is offset by the extra fee income due to the hard limit pressure.

Getting a measure of the historical change in average fee (since the 1MB became a noticeable constraint) should make it possible to reduce the theoretical fee income which is not able to exist.

 

[laurentmt]

Food for thoughts: Fee/kb VS Transaction size

Aggregated per block (for 10/2/2015)

Intra block (for block 359121)

TL/DR: Small transactions pay higher fee per kb

Intuition:
Many wallets use a same formula to compute the fees to be paid ?
Fee = FeePerKb * Math.ceil(size / 1000)
with
FeePerKb = FeePerKb parameter defined in the wallet
size = size of the transaction

 

[solex]

@laurentmt
Welcome to the forum! Your analysis is always valued.

 

[Peter R]

@laurentmt

Interesting data! We've often discussed here how much of the analysis (my fee market paper included) assumes that mining is perfectly competitive, but in reality miners will likely have some pricing power. It would be interesting if miners in the future keep fees for small TXs higher (per byte) than large TXs, just knowing that the average consumer "thinks" in terms of the price per transaction (rather than the price per byte).

In the present case, I suspect that you're hunch is correct that most of what your data shows can be explained by wallets using the same fees for a range of transaction sizes.

 

[Pecuniology]

"We've often discussed here how much of the analysis (my fee market paper included) assumes that mining is perfectly competitive, but in reality miners will likely have some pricing power."

Textbook perfect competition requires that all producers face the same production costs in a single, homogeneous market. Viglione documents Bitcoin price premiums that vary inversely with Economic Freedom Index scores in different countries. This belies the assumption of a homogeneous market. Different bandwidth, electricity, and IT management costs in different countries belie the assumption of uniform production costs.

Furthermore, the existence of a relatively small number of large mining pools, the operators of which watch each other closely, indicates oligopoly, which is a small step from monopoly, if those who control a majority of the hashing power cooperate, even tacitly.

So, yeah... no.

Nonetheless, if miners more or less consistently seek to undercut each other and are limited on the lower bound by operating costs, then the assumption of perfect competition is probably safe as a first iteration, and good enough for practical purposes. On the other hand, if one is writing for an audience of economics researchers—as opposed to finance researchers—then this kind of hair-splitting might be good for an academic paper.

 

[Peter R]

@Pecuniology

Welcome to the forum!!