re: the spam attack, I think we should model out the dynamic fee adjuster so that it cools down much slower than it ramps up

argh not dynamic fee, dynamic block size limiter

like if it cooled down 4x slower than it speeds up then an attacker / abuser wouldn't be able to yoyo it to reduce their costs

IIRC ArticMine commented on this somewhere, but I can't find it now.

<p​reland:monero.social> My question that I had thought of earlier today is: does the block size increase linearly?

<p​reland:monero.social> If so, it should probably be changed to something more exponential, as the number of transactions happening follows trends exponentially more than linearly

<r​ucknium:monero.social> preland: It depends on how much people are spending on transactions. High-fee txs make the block size increase faster. If everyone is paying the minimum fee (20 nanoneros/byte), then according to spackle's analysis the rate of increase is about f(x) = x^x. Very nonlinear.

<p​reland:monero.social> Hmm

<r​ucknium:monero.social> That's what we've seen on stressnet. I don't have an exact equation for what we've seen empirically on stressnet, but getting from the base of 300kb per block to 600KB takes a while. Then when we reach 5MB, it is quick to get to 6MB.

<j​effro256:monero.social> Fluffypony: I disagree. It should cool down very quickly like it does so that spikes (natural or otherwise) in tx volume are treated correctly: like spikes. The block size limiter should not stay high for temporary spikes as this opens the door to more abuse

<j​effro256:monero.social> I much prefer yoyoing to permanently large blocks

<j​effro256:monero.social> Also, yoyoing does not reduce cost for the attacker since the most expensive part of a spam attack is pushing the block reward into the penalty zone

<g​ingeropolous:monero.social> "the long tail" is how i always thought of it. And the blocks wouldn't be permanently larger... sure, during a spam attack they might, but the blocks will only fill with transactions that are available.

<g​ingeropolous:monero.social> and during a spam attack, it might allow normal users to get the experience they expect

<g​ingeropolous:monero.social> spams gonna happen. to me its just a matter of how usable the network remains during the "spam"