<gingeropolous> "<sech1> 1000 miners = 85 TB/..." <- Pools don’t use cdn, bandwidth is cheap now and costs a fraction of your estimate

* > <@gingeropolous:libera.chat> <sech1> 1000 miners = 85 TB/month, or $500 to pay for a CDN >>> wait wait. With this estimate, minexmr would have a $6k / month CDN cost

Pools don’t use cdn, bandwidth is cheap now and costs a fraction of your estimate

<merope> I'm thinking in terms of trying to destabilize the stability of the mining network

<merope> The reward penalty, on its own, has the side-effect of decreasing the mining incentive, thus reducing the maximum network size that could be supported by it

<merope> So a strong sustained growth would decrease the security of the network at a critical moment of increased usage

<merope> Luckily we have the small "penalty-free" zone that allows a miner to include a few extra txes without incurring in a net loss, so small sustained growth is possible without any repercussions on the mining security

<merope> I'm planning to take a deeper look into this in the future (once I'm done with the current projects), but I thought I'd throw it out there in case anyone has any inputs on the topic <--- Yes I do

The miner actually profits from the penalty. This is because fees more than pay for the penalty and this profit increases as the rate of increase of the blocksize increases

The key is that the lowet paying transaction in the block is the one that pays the hiest marginal penalty rate. This follow from just assuming a free market between miners and users

To put it another way why would a miner mine at a loss?

So the miners add transactions to the block in order of fee per byte. The highest paying transactions first

so when the marginal increase in the penalty for adding a tx is greater than the fee of the tx the miner stops

So the lowest paying tx actually pays the highest marginal penalty.

In the extreme case of doubling the blocksize the miner net 3x the block reward for mining the block. 4x in fees less 1x for the block reward

This corresponds to the maximum fee

endor00 ^^

The miner in the extreme case gets 4x the block reward in fee but not block reward. So at tail emission 2.4 XMR vs 0.6 XMR for mining the block

<sech1> "he can only criticize" <- "Lead, follow or GET OUT OF THE WAY!"

Oooh right - for some reason I was implicitly assuming that profitability follows block growth, when in fact block growth follows profitability. No higher fees to offset the penalty -> no bigger blocks

"The key is that the lowet paying transaction in the block is the one that pays the hiest marginal penalty rate." That means that the lowest paying tx has the smallest profit margin for the miner, right? So starting from the highest transaction fee per byte, the further you go, the less incentive you have to include it in a block, right? (If not, could you expand this a little further?)

(Otherwise I'm not exactly sure what you mean by "highest marginal penalty rate")

merope it's actually straight-forward: if adding a transaction to your block template increases the reward, just add it. And add transactions starting from the highest fee per byte.

with this algorithm, the last transaction to be added has the lowest fee per byte of all transactions in the block template, but it pushes block weight above median the furthest

Right

So paying a total of 4x the reward in fees is the minimum amount to make it worthwhile for a miner to double the blocksize

why 4x? The penalty can't be higher than the block reward

so you need block reward worth of fees + a bit more

But that "bit more" only gets you a small increase, because otherwise you could drop some txes and gain more from the reduced penalty

Hmmm

Then I'm not sure I fully understand this: "In the extreme case of doubling the blocksize the miner net 3x the block reward for mining the block. 4x in fees less 1x for the block reward"

A block filled at about the median block size gets you a block reward R, assuming tx fees are negligible.

Filling it to 2x median gets you 0.

So in order to fill in without losing, the second half of that 2x median block must give you at least R in total tx fees.

Since you use higher fee/byte txes first, it means the entire set of txes you put in that 2x block have at least the same bang for the buck.

So you get 2R from tx fees, plus R base reward.

I don't know where the 4 comes from.

Maybe the above is wrong, but it seems to check out.

Oh, right: my second part is wrong. You get two halves at R, but lose the block reward. So you get... R.

So yes, I agree with you, I'm missing the point too, nvm.

I think I see where it's coming from: the penalty is not linear.

So while you'd get R if each half got you R in fees, there's a better sweet spot with a partially filled second half, so you'd stop there.

You'd need to calculate where that sweet spot is. It must be twice as better then.

twice better

<moneromooo> I don't know where the 4 comes from. <---- The 4x assumes honest miners that are acting in their enlightened self interest. This makes the miners an adversary to the attacker

It comes from the fee rate at maximum growth rate of the blocksize namely 2x. This is the maximum fee by the way

Pricing big bang attacks as an add on to a 51% attack is a different issue. The cost for the big bang add on on top of a 51% attack is the block reward for the attack blocks so 1x

The key is that without 51% the honest miners profit at the expense of the attacker at the rate of 3x the block reward for an all out attack. They get 4x in fees but forfeit 1x the block reward to the penalty

dangerousfreedom, u can totally join the fun

Hey UkoeHB and sethforpivacy, I'd def like to work on Seraphis dev when the research is done. If you have any specific tasks, you need on the research side of Seraphis, lmk.

sethforprivacy this seems cool: monero-project/research-lab #100 where is this at?

it caught my eye in the list of open problems you shared earlier this week