What scaling problem exactly are you referring to here? ^
Bloat in particular. There was an attack exploiting this generation of bloated mixings and the response was to increase fees so as to make the attack less feasible.
Isn't this an "attacker economist" response to the problem of generating bloat from the mixing?
...
It gives an example of how the math and probabilities (keyword) of the level of mixing increasing decreases the probability an attacker would be able to successfully link addresses together on the monero block chain in a non-liner fashion.
In other words, the system is not 100% robust in itself. It is not designed to be 100% safe. It is designed with the assumption that someone doesn't have a lot of funds to sybil the shit out of the system.
In DJB's block post he states the following:
...
The "attacker will probably fail" approach. People taking this approach say things like this: "We're designing this system so that the cost of breaking it is larger than the resources available to the attacker."
...and this is precisely how sybil attacks are dealt with in Monero.
Yes you could say that about the fees being raise. But how is that a problem? Can you specifically detail how raising the min fee is still within the "attacker economist" approach? As it wont be a problem for an annoying-person to spam the monero network if they choose. The block size will adjust accordingly up and down.
Monero is counting on the possibility that the resources of the attacker will not be sufficient to sybil attack and unmask those trying to be anonymous.
Quite the contrary it isn't about the
"possibility that the resources of the attacker will not be sufficient"....
it is...
"the probability that the attacker will not have enough resources to successfully execute the attack...ever"Those two statements mean very different things ^.
Another snippet from DJB's post concerning an attacker having enough resources:
Can an attacker actually carry out 280 or 290 or 2100 operations? Here are some back-of-the-envelope numbers that help put attack possibilities into perspective. Mass-market GPUs use state-of-the-art chip technology, are optimized for floating-point operations, and perform about 258 floating-point multiplications per watt-year. The number of floating-point multiplications that the attacker can carry out in a year with this technology is limited by the number of watts available:
226 watts (284 mults/year): the power substation for one of NSA's computer centers.
230 watts (288 mults/year): the power available to a botnet that has broken into millions of computers around the Internet.
244 watts (2102 mults/year): the power actually used by the human race at this instant.
256 watts (2114 mults/year): the power that the Earth's surface receives from the Sun.
257 watts (2115 mults/year): the power that the Earth's atmosphere receives from the Sun.
The idea is to make sure that an attacker doesn't have sufficient resources to even attack.
The difference between that approach and the "economist attacker" approach is that the attitude of it "being good enough" as opposed to "being overly sure an attack could never take place" is executed.
It's like standing near a cliff and hoping the cliff doesn't collapse and you are at the edge within a certain "safe" distance.
Distance of the "economic attacker" approach: 20 feet
Distance of the "attacker will probably fail" approach: 20,000 feet.