|
How is the sun treating you?
Are you feeling better?
|
|
|
|
I know xulescu addressed this as a misperception, but how can something that happened after the announced attack have contributed to the attack.
It was not misperception. Fluffypony, like I earlier, got pissed off and used aggressive words. That was a communication mistake. But that doesn't change your point. |
|
|
|
|
This cat and mouse game became very tiring and you are gasping for air. I do not get pleasure from "winning" this argument. I think all interested partied can learn something from this exchange and will terminate it here.
Your model very conveniently ignores a 4-blocks-per-minute minute. Put that in, take the last minute out and redo the math. Otherwise explain why it only seems that your model very conveniently ignores it.
Edit: Finally, do you have confidence in your model or not? This affects your confidence that something changed, according to your model.
|
|
|
|
... he said basically until there is an attack there is no attack ... AnonyMint, this is, trivially, the case.and he is going to the beach. ...and kept thinking about the problem. Walks help the mind get unstuck. What does geographics have to do with it? Specifics please. Since you ask about geographics specifics, the way nodes connect to each other cause orphans and "opportunity cost orphans" due to delay. The real world is far from uniform. This is informative for a modelWhat's interesting to me is that he said "roughly", he said "at present time" and said it three months ago. I hardly see how this could still be accurate.
You make sweeping ass-u-me-ptions. Do you actually have your mind deep in the Github commits? Sorry buddy, seems you're the one making the assumption. I DON'T assume the time interval length is independent from hash rate, etc etc etc. you ASSUME it is independent. Afaics, you've likely got the wrong model, so you aren't seeing anything. So you answer none of my questions and instead broadly say my model is wrong. Well, that is not an argument.
All models are wrong, but some are useful. If you don't understand this, go deeper into modelling. You assume your model is right. Mine allows for some uncertainty. Certainly black swans are black swans by definition.I didn't change the experiment. You framed the experiment wrong. There aren't 12 gaps, rather there were 4 time intervals of 1 minute. If you have the wrong model, you see noise. I don't have to consider permutations because there are 2-4 orders-of-magnitude between the probabilities I showed. There is the stratification I am referring to. We could get more analytical, but really isn't necessary. We will have to agree that this was my model from the beginning, and you have a different model, possibly from the beginning, and then the numbers are not directly comparable. I certainly fail to see why "blocks per minute" is more fundamental than "length of gap". They're certainly inversely proportional. You're just throwing away data.It is quite obvious that we saw a rare event. And no one has shown otherwise yet. See above for correction. At least an even rarer than that was found with little manual inspection.Block lengths don't exist, there are only timestamps (or the differential, gaps). As smoothie argued, everyone can "manipulate" them because it is not enforced in any way. It is also not very useful to do so.
You ass-u-me that is not a vulnerability. Can you prove it is not? Barring weird implementation artifacts, timestamps only influence difficulty, thus we're talking strictly about a time warp attack or another difficulty-adjustment-attack. This would be seen in the data. It is true that using the wrong model you might not see the pattern, but I said that the dataset is too small to support you noticing a change or that screenshot being relevant. |
|
|
|
Remember Risto issued the 4 - 8% proclamation of the likelihood of a BCX attack while the 72 hour deadline was ongoing. One could ponder if this might have caused BCX to not call off the threatened (unconfirmed) attack. There was some progress on trying to find any vulnerabilities, so perhaps one could ponder if he might have called off the threat citing progress and cooperation.
The reason for not just handing over the vulnerability and fix, may be to illustrate the competence and to teach the community.
Are you suggesting Risto should feel guilty for BCX's continued wrath? Were we looking for vulnerabilities with a BCX quota? The ALLEGED vulnerability. But you are otherwise an excellent devil's advocate. My more substantive post is here in case you missed it on the previous page ...
Edit: I think I know exactly what your mistake is. You assume the sample in the screenshot is uniformly sampled. But it is not, because it is chosen adversarially, to prove the point BCX was making (that is it a strange ocurrence). |
|
|
|
For the fifth time, the burden of proof rests with you for your initial claim (haven't picked up the ball yet but catching up). Edit: I think what may motivate BCX is defeating overconfidence.
I don't think the devs have been overconfident at all, on the contrary. Maybe I am. But what is BCX's problem? I don't understand how you think geographics affects the model? Isn't clustering modeled by Poisson?
I do not assume independence. You do. I cannot assume Poisson, I look at the data and see it isn't. 4 times in 1.5 hours. 3 months ago.
Anecdotal evidence against your claim. What's interesting to me is that he said "roughly", he said "at present time" and said it three months ago. I hardly see how this could still be accurate. (Picking up the ball) Picked the most recent 12. http://chainradar.com/xmr/blocks242773 14-10-02 05:35:56
242772 14-10-02 05:35:56
242771 14-10-02 05:35:42
242770 14-10-02 05:34:35
242769 14-10-02 05:32:59
242768 14-10-02 05:31:30
242767 14-10-02 05:30:29
242766 14-10-02 05:28:55
242765 14-10-02 05:26:41
242764 14-10-02 05:25:48
242763 14-10-02 05:25:24
242762 14-10-02 05:22:34That looks roughly to be one occurrence of 3 in one minute, one occurrence of 2 in one minute, and 7 occurrences of roughly 1 per minute (slightly longer than a minute so my summary is not a precise model). p = (13 / 3!e) × (12 / 2!e) × (11 / 1!e)7 = 0.001%So that is within a factor of 5, and note my model above isn't incorporating the effect of the slow blocks. So that anecdotally confirms your claim. Within a factor of 5 to what? What do you compare it to? However the math above is wrong because for a perfect distribution the probability would be even less. p = (11 / 1!e)12 = 0.0006%Instead we shouldn't be be comparing 12 gaps. Rather for the example above there are 9 intervals of one minute, so the probability for a perfect distribution over 9 intervals is as follows. p = (11 / 1!e)9 = 0.01%Thus the example we were considering was only 4 intervals of one minute. So let me test your claim again as follows. 242773 14-10-02 05:35:56
242772 14-10-02 05:35:56
242771 14-10-02 05:35:42
242770 14-10-02 05:34:35
242769 14-10-02 05:32:59
242768 14-10-02 05:31:30That looks roughly to be one occurrence of 3 in one minute, and 3 occurrences of roughly 1 per minute (slightly longer than a minute so my summary is not a precise model). p = (13 / 3!e) × (11 / 1!e)3 = 0.8%Sorry that fails your claim. Let's test another. How does this fail my claim? You changed the experiment in the middle of the experiment. Again, what do you compare that to, to conclude it fails a margin?242767 14-10-02 05:30:29
242766 14-10-02 05:28:55
242765 14-10-02 05:26:41
242764 14-10-02 05:25:48
242763 14-10-02 05:25:24That looks roughly to be one occurrence of 2 in one minute, and 3 occurrences of roughly 1 per minute (slightly longer than a minute so my summary is not a precise model). p = (13 / 2!e) × (11 / 1!e)3 = 1.25%Sorry that fails your claim. 0.8% and 1.25% seem within my margin, and they're not even the values for the initial experiment and not even directly comparable (you compare 6 gaps to 5 gaps).Now you see why independent verification is important. Ball in your court. What is my mistake? So you ignore permutations and variants, change the experiment twice and still don't manage to get it your way? Also, put error bars on your numbers. Your confidence will drop significantly if you did that. Are we sure there are no symptoms already?
Well let's put it this way. My argument so far is that there aren't. Suppose I was wrong and there were symptoms. Smoothie's argument was that it wouldn't matter anyway. If I was wrong, BCX claimed they would bring chaos. They didn't. In fact I again openly invite any Global Mod, Badbear or Theymos to Permaban me if I am Moneroman88.
This is a vacuous claim. You could be Moneroman88 in many ways in which none of those would know. My theory is BCX has the ability to manipulate either block lengths or the timestamps for minutes at a time.
Block lengths don't exist, there are only timestamps (or the differential, gaps). As smoothie argued, everyone can "manipulate" them because it is not enforced in any way. It is also not very useful to do so. |
|
|
|
|
For the fourth time, the burden of proof rests on you to prove material change. You cannot prove lack of material change in a stochastic environment because you need infinite data. That is not the case with proving change.
Of course the probabilities are not 0.5. But they also don't matter much. Since both my semantic simile and your argument assume independence, order does not matter. Thus all permutations are in the same class of rarety.
Furthermore, if only counts of "short" vs "long" gaps matter, then instead of x seconds times 8 + y seconds times 4 you also need to include small deviations. Such as, for example, x-1, x+1, x times 6 + y times 4, and all the permutations of each of these. So you are integrating over all partitions on 12 elements, which is a gigantic set when you generalize enough to learn the blockchain in any meaningful way. The blockchain has too little data to believe your statement with even 60% confidencence. We're talking 0.1 sigma deviations here and a combinatorially monstrous set, with only a quarter of a million of data points. Not gonna happen.
On the contrary, my position is that given the amount of entropy there is in the blockchain so far, and adding time dependence to the combinatorial mess (because independence is false), we cannot say that there is something wrong with meaningful certainty.
Finally, over all this academic modelling exercise that we went through, the reality is, as I mentioned in my first post and smoothie detailed, that even if you were right on the modelling, what we know about how timestamps work and how they are somewhat adversarially arbitrary for you as the modeller, your conclusion holds no epistemic water.
You are right on the Maths. You are wrong on many levels on your modelling. Even if you were right on the modelling, you are still wrong on what conclusions you can draw from the results.
Your argument is "something could be wrong". My counterargument above is "even if that was the case, you don't have enough entropy to draw that conclusion". Smoothie's corrolary is "even if you are right, it doesn't make much of a difference".
Thank you for responding calmly earlier and compelling me to articulate my position.
Edit: answering your edit
|
|
|
|
I have gap distributions and how these distributions change in time, I have autoregressors trained on the historical blockchain data, I watch 40 notes geographically around the planet and THERE IS NOTHING WRONG.
Now data or GTFO.
I don't see any data. Open source it. For the third time, the burden of proof rests with you. Moreover, I do not intend to open source my data because (1) most of it is simply the same blockchain that you can view yourself (2) I use some of my indicators for trading purposes (3) there is no reason you would believe any data dump I would release as supposedly from my nodes around the world. EDIT: Let me rephrase what you propose:
I have a fair coin and BCX asserted he can control the outcome of independent throws. So, I throw the coin 10 times and get 0001011010. Since the throws are independent, that should happen only once in 1024 throws. THERE MUST BE SOMETHING WRONG.
Correct. We would need to add up all the probabilities of every permutation, which is what I noted in my prior message. However, your analogy is inapplicable. Do you know why? Of course my analogy is inapplicable for a number of reasons. I do not pretend that is a model of the problem, but only exemplifies what you agree here (that absolute probabilities are meaningless). But since I'm trying to get on the same wavelength with you, I'll ask: "Why is that?" |
|
|
|
I believe the math above is incorrect, because each 1 minute trial is independent (which is one of the requirements for a Poisson distribution). Thus we have four consecutive events, two are 4 blocks in a minute and two are 2 blocks in a minute. Thus the probability is as follows.
p = (14 / 4!e1)2 × (12 / 2!e1)2 ≈ 0.000795%, i.e. an occurrence expected roughly every 125,794 minutes which is every 87 days!
If my correction is correct, we do have evidence of something rarely occurring.
(edit: fixed quote)Man I'm tired of your fucking antics. Do you understand probabilities? Do you understand why your assumptions of independence go out the door? Do you understand that the absolute probability of a complex event is meaningless? What you're doing is either intentional FUD, delusion or skipping or meds. I had an enormous respect for you when you had the AM handle. Ever since the BCX saga began you've been (publicly) nothing but a reckless agitator, shouting EUREKA for every bullshit possible omen. I have gap distributions and how these distributions change in time, I have autoregressors trained on the historical clockchain data, I watch 40 notes geographically around the planet and THERE IS NOTHING WRONG. Now data or GTFO. Why are you so pissed off? If you have a refutation, then present it. Ad hominem doesn't add information for the readers. Note you are so emotional, that you didn't even pay attention to what I wrote. I said my correction is to bring it into agreement with the requirements of a Poisson distribution. Note I didn't make any proof that a Poisson distribution applies here. So you are building an ad hominem strawman. Note what I wrote is misleading because there innumerable other rare events, and when all those probabilities are summed, then the probability of any one of them occurring is much less rare than I stated above. One objective is to refute the dismissal upthread stating that the above event could occur every hour. Please show me such an event occurring every hour.another 4 blocks in the last minute.
By my math, with 1 block per minute mean rate, one should see 4 blocks in the same minute about once every hour or so. Is this correct? OK, my apologies for the deviation from discourse standards. Your position therefore is that IF the Poisson distribution is correct AND events of comparable probability happen rarely THEN something is wrong. I hold that the Poisson distribution is not justified for a number of factors, including variation in hash rate, geographics and clustering (nonuniformity) of block finds. I also hold that you can choose any consecutive 12 gaps and you will get a probability within a factor of 5 from your example's with confidence over 90%. And again, the burden of proof rests with you. When one argues things did not change and the other argues they have, it is the latter who has to prove the change. |
|
|
|
I believe the math above is incorrect, because each 1 minute trial is independent (which is one of the requirements for a Poisson distribution). Thus we have four consecutive events, two are 4 blocks in a minute and two are 2 blocks in a minute. Thus the probability is as follows.
p = (14 / 4!e1)2 × (12 / 2!e1)2 ≈ 0.000795%, i.e. an occurrence expected roughly every 125,794 minutes which is every 87 days!
If my correction is correct, we do have evidence of something rarely occurring.
(edit: fixed quote)Man I'm tired of your fucking antics. Do you understand probabilities? Do you understand why your assumptions of independence go out the door? Do you understand that the absolute probability of a complex event is meaningless? You start from the wrong place, do the wrong thing with that and arrive at a result that is in itself meaningless. What you're doing is either intentional FUD, delusion or skipping on meds. I had an enormous respect for you when you had the AM handle. Ever since the BCX saga began you've been (publicly) nothing but a reckless agitator, shouting EUREKA for every bullshit possible omen. I have gap distributions and how these distributions change in time, I have autoregressors trained on the historical blockchain data, I watch 40 notes geographically around the planet and THERE IS NOTHING WRONG. Now data or GTFO. EDIT: Let me rephrase what you propose: I have a fair coin and BCX asserted he can control the outcome of independent throws. So, I throw the coin 10 times and get 0001011010. Since the throws are independent, that should happen only once in 1024 throws. THERE MUST BE SOMETHING WRONG. Of course that exact sequence likely never happened again, and that proves nothing. EDIT 2: Before you set your hair on fire and run like a headless chicken again, note that (1) this is only my opinion (2) I am not affiliated with the core team and (3) I am talking my books. |
|
|
|
|
So much repressed homoeroticism.
Risto could not have accepted the bet because a 500 BTC bounty for an attack is likely sufficient to bring down Bitcoin too (at least temporarily).
TFM insists that there is an attack path that BCX could take to gain financially from:
A1. Use TFM's or some other amplification attack to deanonymize a public key somehow
A2. Make use of the simultaneous equation to solve the EC problem and expose the private key for that public key
A3. Be a significant part of the hashpool and sufficiently connected to intercept transactions on the fly and double spend them (or feasibly perform the attack many times).
Other attack path:
B1. Time warp and/or somehow isolate the exchanges; DDoS as necessary
B2. Double spend the exchanges.
State of the matter:
TFM was paid to work on A1 and smooth (IIRC) confirmed the dev team gained from their collaboration.
A2 is a very thorny problem and I'm willing to bet that the community at large considers the burden of proof on TFM's shoulders -- we have no reason to believe the extra question helps in any way, but it might. There is nothing in the literature, as far as I can tell, one way or the other. Such an attack would be much larger in scope than just Monero and would create waves in the security community.
Looking at the blockchain and network statistics, there is no reason to believe either that (1) a significant part of the hash rate is controlled by BCX or that (2) he is sufficiently well connected to attack transactions on the fly reliably. Of course, as many others have said, BCX is a rich and resourceful narcissist so A3 is not necessarily out of his reach. All I can say is that we have no reason to believe this was the case in the last day/week/month.
BCX allegedly already DDoS'ed Poloniex as a test/warning. We have no reason to believe a classical time warp attack is happening. There are other possible attacks that are silent until they hit.
Bullshit factor:
BCX's communication so far was repeatedly confused, self-contradictory, called out on this and ignored. So we can assume nothing and merely speculate. We have no reason to believe his actions are rational or in his best interest.
Thus, at this point, there really is nothing to do but wait until either (a) BCX succeeds (b) BCX confirms no/failed attack (c) everybody gets bored of the drama. Since BCX claimed Cryptsy was not considering Monero due to security concerns and they added it yesterday for voting, and now is on third place with no direct BTC votes... well really make of that what you believe. There is another Danish (?) exchange that suggested they might add USD/EUR markets soon so... again, caveat emptor and past performance does not guarantee future performance.
In any case the bullshit levels are clearly over 9000 here. The only reason anyone is really taking BCX seriously is because it's common knowledge that he's rich and brain damaged, so he needs to be treated with gloves.
Obviously, this is only my opinion and I'm talking my books.
|
|
|
|
No offense but I don't think BCX is trying to trade his reputation for a 15% discount on Monero.
His 500BTC bet offer seems to back up my assumption.
The bet is untenable. It is game-theoretically impossible for Risto (or anybody else, who we suppose helps XMR) to accept it. BCX is an asshole, but he's not an idiot and perfectly knew that. |
|
|
|
Has anything interesting happened in this story over the past 24 hours?
No man. But in Russia we say *go big or go home* so either scoop up all the ultra cheap Monero or let it be ~MM88~ Classy choice of sig. You must be Fonzie! ~x~ |
|
|
|
qft
What does QFT stand for? The only expansion I know of is Quantum Field Theory and doesn't sound right. Is it a variation on QED (quot erat demonstrandum)? EDIT: Oh, "quoted for truth". Noted. |
|
|
|
Yup. I am out of the stupid argument about pure economics and semantics.
My conclusion for this large OT-branch here is that why not put Monero into the bullish trend by adjusting the supply to the demand. If there is no demand for all the supplied coins (ie price goes down), there is no reason to supply too many coins. Those coins can be "saved" to later days when there is more demand.
Also, if it is made sure the supply is slightly below the demand, it makes the marketcap rising (some may want to dump but this need to be considered when re-evaluating the coin mintage).
For the adoption, it is better to make it constantly rising (it can be made only by adjusting the supply vs demand framework).
For a college student in an elite economic theory program, you DO lack reading comprehension. Re-read my posts and re-evaluate your responses. You either (1) didn't read the replies (2) didn't understand the replies but decided to ignore them as irrelevant (3) understood the replies and decided to ignore them as inconvenient. I will not continue this off-topic. |
|
|
|
|
I believe it is a good time to restart the conversation on Monero's tail emission. So far, the main ideas that have floated around are (in order from fattest tail to thinnest tail):
1.
constant inflation
block reward nonmonotonic: block reward decreases exponentially until tail starts, then increases exponentially
total emission unbounded, with an inflection point: concave ("doesn't hold water") until tail starts, then convex ("holds water")
2.
inflation converges to zero from here below
block reward: constant after tail starts
total emission unbounded, linear after tail starts
I would like to propose other ideas for the sake of discussion:
3.
inflation converges to zero
block reward: inverse linear decay, converges to zero
total emission unbounded, logarithmic after tail starts
4.
inflation converges to zero
block reward: inverse quadratic/cubic/... decay, converges to zero
total emission finite from here below
Bitcoin uses
5.
inflation converges to zero
block reward: exponential decay, converges to zero
total emission finite
|
|
|
|
|
Sent 100 XMR. Payment ID: 99cd3b04420824a615b08fa248f5ed48c912a12f9263668690a6a6bb0d147b31
|
|
|
|
even though I mostly agree with your conclusion as well as your position, the free market fails in our specific spot to finance the development of monero. the reason is a dilemma which cannot be solved in a decentralized, anonymous free market manner. at this point decentralization fails, this obv. does not neccessary need a state to be a regulator but it needs a central institution, which solves this specific problem
Yup. The problem of free markets is that it is not a perfect like Chicago school suggests. The free rider problem is something that the free markets cannot fix. In this specific case, however, Monero's main problem is its low marketcap and thus the hardness to get some serious dollars out of the community for development. Personally I am ready to donate as soon as my monero wealth grows in a significiant level in value (I want higher purchasing power for my moneros so that smaller donation will count more). But, the problem of free rider is just one critisism of free markets. I am not purely Keynesian or purely Chicago school because both have their problems. I don't think anyone here suggested free markets are perfect. At the same time, you are suggesting a central bank style direct intervention on the money supply. That is the main reason why cryptos exist in the first place - to prevent one entity from controlling the money supply. |
|
|
|
|
Show Posts
