Gold collapsing. Bitcoin UP.

[NewLiberty]

Re-read what you just wrote. You (and Craig) are still falling for the gambler's fallacy. You just don't see it yet.

(And for the record, the paper was publicly available on SSRN [he has now removed it] and Craig also knew I was critiquing it on this forum).

No, you are missing it entirely still. "Everybody" knows how you intended the question to be read, however in the context of the discussion it was an equivocation.

EV was measured at N-1 and unchanged by anything, unless the assumptions are also including the other two statements, which changes the perspective of the question to "the clock on the wall".
If you take the second two statements, then it would be gambler fallacy because the perspective shifts to omniscience, and 15.

You aren't telling anyone anything, that they don't already know, and frankly it is a bit insulting that you persist in pretending that there is only one interpretation of the question. There are likely a lot of folks that do not see how the gambler fallacy would apply, and make the much simpler mistake that you are suggesting Craig and I are making. That is not what happened.

Your answer is well understandable. It is not wrong given the equivocation. A naked reading of "the bet" would be obviously 15min. You do not need to convince anyone of that, and certainly neither me nor Craig.

If you do not see how he will perceive that you tricked him, and betrayed him, you will not find reconciliation. He is going to think the worst of you, and that it was a setup.
If you are able to see it as he saw it, and own the fact that he was tricked (even if unintentionally), you might even collect on "the bet" and more importantly resolve the rift created betwixt you and he by "the bet".

 

[Peter R]

Please read my post right above yours. It gives a further explanation.

I'm sorry that you're upset about this but the answer is t = 15 min. I think Craig, yourself (and some others) are angry because you think I'm trying to trick you somehow. Craig even mentioned that I had cheated. But this is not a trick question.

Many people were fooled by this. @bitcartel wrote a selfish mining simulator where he did the same thing (he assumed the expectation value was unchanged by the fact that the SM found the next block). And so he got numbers more like Craig's. I think he is close to "getting it" now however. He is approaching it with an open mind.

There is nothing wrong with making mistakes or not understanding something. The first time someone asked me the question I asked about the expected time between blocks if I check at a random time, I said "10 minutes." I thought I understood what it meant for mining to be memoryless but I still didn't get it at a deep enough level to see why the answer was "20 minutes." I bet there are still aspects of memoryless processes that I don't understand.

If you want further evidence that Craig is indeed making the mistake I claim he is making, just look at the first sentence of the paper he removed from SSRN:

By "events" he means finding the next block. He is implicitly making the statement that mining is not memoryless right here. These are the first words out of his mouth. But because the Gambler's Fallacy is counterintuitive, he still doesn't see it.

 

[AdrianX]

Can someone explain to me ELI6 the evidence why mining is memoryless, roling a dice I understand is, but doing a random culculation that is never repeat is as far as I can see evidence that mining is not exactly memoryless.

 

[NewLiberty]

At the end of the day, Selfish Mining is not an effective strategy in Bitcoin, nor in any altcoin. It has never happened, because the SM loses money, even if slightly less than other miners do. It is a losing strategy economically up until the point in the future that the ONLY profitable activity on the planet is Bitcoin Mining (never). Getting more than your "fair share" is meaningless if you get less with the strategy than without it.

Thus it is like doing physics in space, without gravity. It might be a useful way to teach basic principles, but it really has no practical purposes until we don't have a planet to stand upon.

The selfish mining paper will fall, as will all the papers that rely upon it. If not by Craig's papers than others will do so. Emin was given an early opportunity to do it himself. It was his opportunity that has passed. If you did this as some sort of virtue signaling out of loyalty to Emin, then your greatest fault may just be being too loyal to a friend, to your own detriment, and to Bitcoin's detriment. That is a value judgement for you to make. I don't know how close you and Emin are, it must be very close indeed.

All Emin's SM paper shows is that Selfish Mining is an effective strategy using a model of Bitcoin abstracted from the reality of Bitcoin in the real world. At best it is a useful teaching exercise, but it WAY oversteps the bounds of rationality to suggest that there are any changes to be made to Bitcoin to accommodate it. That it goes so far as to make such suggestions, means that it is also dangerous to Bitcoin if taken seriously.
It would be like saying we should change car tires so that they stick to roads because in zero gravity they will fly off the road on a curve.

I saw this when it was first released, and remember thinking. "Oh my, this is a really bad career move for the author." I knew it would get shot down. I'm not in academia and have no use for academic credentials, if I did, I probably would have taken shots at it myself, it seems such an easy target. When Craig made a big deal out of shooting it down, I told him he was wasting his time. No one takes it seriously. Then he showed me all the citations it has, and it started to make sense why he was wasting his time discrediting it. To me it never held any credit or worth as it was obviously false.

[doublepost=1501562637][/doublepost]Mining is not perfectly memoryless, but nearly memoryless to the point where it might as well be entirely memoryless.

There are two basic reasons for this, as well as some other more esoteric ones, but these two should suffice.
1) The nonce space is very large compared to the amount of nonces burned through, and there is no guarantee of a matching nonce in the space anyhow.
2) The block a miner is hashing against changes whenever a TX in it is changed, added, swapped for a higher fee TX or etc, this destroys the tiny amount of meaningful memory there might be.

This is why the EV of finding a block stays the same each time it is measured. The cool part mathematically comes with the poisson distribution, (or negative binomial for SM issues), and is why the math shows that it doesn't take forever even though the expectation is always moving away with each passing second.

 

[Peter R]

@NewLiberty

I don't think we see selfish mining for mainly two reason:

1. It requires the selfish miner to maintain the attack for a long time in order to recoup the losses he suffered before the difficulty reset. I suspect a potential selfish miner would be worried that the honest miners might react with a defence strategy before he recoups his losses.
2. For practical values of gamma (I think anything above gamma = 0.5 would be difficult to achieve), profitable selfish mining requires the selfish miner to control a lot of hash power (25% in this case) and I think miners who control this much hash power want to "play nicely."

What I think _is_ a really interesting question is what those "defence strategies" might look like. Can we prove that honest miners could detect a selfish mining attack and then take defensive action? Or maybe something like subchains would be sufficient to prevent selfish mining, by driving up the orphan risk for when the selfish miner publishes his private chain.

I think it would be more fruitful to acknowledge that Eyal and Sirer discovered something very important. In fact, it was so counterintuitive that it was overlooked by people intimately involved with protocol development for 5 years. To this very day, many people still don't believe that it works and argue for weeks claiming the math must be wrong!! The selfish mining paper is a landmark paper that will be cited and read for decades to come.

Now we should build on top of this work by, first, accepting that it is real, and then trying to show how the honest miners could defend against it.

 

[NewLiberty]

Except that SM isn't real, and doesn't work until at the end of the world when the only activity is Bitcoin mining.

It fails at changing the Nash Equilibrium. It is a losing strategy for the SM, they never catch up in profitability no matter how long it is engaged, it only gets worse. Difficulty adjustments aren't going to save it.
Getting a bigger piece of a much smaller pie, is a losing strategy. Unless it is the only pie there is.

On detection and countermeasures
It is detectable at the first block released by the SM.
If the SM is hiding a block they have solved.
What happens in a standard query?
The SM advertises they are doing the attack because the orphans are not lost. Bitcoin miners maintain both forks. Miners only build on one... but they save both.
So. The SM states no block... then releases one with a prior time
The evidence of the attack is in the first block.
[doublepost=1501564604][/doublepost]If the SM instead of losing that money, invested it in mining, they would be more successful than selfishly mining... and that is what miners do in the real world.
[doublepost=1501564697][/doublepost]If SM was overlooked, it was because it is unimportant. There is nothing to build on, there is nothing to fix. Until the end of the world.

 

[Peter R]

@NewLiberty

Yes, it could very well be true that in practice the attack would result in a loss, once you factor in counter measures that honest miners may take, or perhaps changes in the price of bitcoin due to loss of confidence.

I am not disputing that. In fact I agree. And I'm sure Emin Gün Sirer would agree too!

 

[NewLiberty]

Where the SM paper completely lost credibility when I read it years ago, was in the recommended changes, which do absolutely nothing to resolve the problem and instead make it worse.

The SM is going to lose every race to propagate because it releases 2nd, unless the paper's recommendations were followed, of randomizing the block to build upon. This would give the SM a 50% chance of winning that race! It is the exact opposite of a good recommendation

When I saw that, it showed me that the paper was nothing more than a disingenuous and obviously stupid hit piece on Bitcoin. One of a huge number of Bitcoin-is-dead FUD nonsense. I knew it would be discredited. I had no idea it would take this long.
I just assumed the authors were looking to create a buying opportunity for themselves, but the paper was bunk.
[doublepost=1501566062][/doublepost]I really assumed that everyone knew this and was laughing behind the back of the authors, as I was, and enjoying the buying opportunity. I had no idea the academic community thought that it had any merit at all.

 

[NewLiberty]

@NewLiberty

Yes, it could very well be true that in practice the attack would result in a loss, once you factor in counter measures that honest miners may take, or perhaps changes in the price of bitcoin due to loss of confidence.

I am not disputing that. In fact I agree. And I'm sure Emin Gün Sirer would agree too!

SM fails even with no countermeasures.
It is not a paper that is written about Bitcoin. What it is written about does not exist in anywhere in the world. Maybe it works in EminCoin, but there's nothing like that in reality.
Again, it is like doing physics problems without gravity and without friction and telling car manufacturers that they need to make changes in the way they design cars.
The really dangerous thing is that some development teams DID take it seriously, and I had assumed that they did so only because it aligned with their own nefarious agenda. I never really believed that people took it seriously. When I see developers taking it seriously, I tend to assume either
1) that they have some agenda other than what is best for Bitcoin and want to use it to do something that advantages them, or
2) that do not see economics as an important part of Bitcoin design. Either would disqualify them in my evaluation as someone I would want to be involved in Bitcoin development.

So now I am pretty much flabbergasted that it has come up again. I don't get it. My most generous explanation is that personal loyalty has blinded you to all of this, and that you are OK with sacrificing everything you are doing for a friend, even though the friend is a naked emperor.
[doublepost=1501567836][/doublepost]On top of all that, it is entirely not my place to be involved in this discussion. Getting between the two of you can only be bad for me. I'm not going to be making any friends by sticking my nose in this. It is no-win for me. I do hope that the two of you can resolve it like gentlemen before too long. I do not know what that resolution would look like, it seems all but impossible. I'm reaching out to you here mostly because I have a pretty good idea how it will end up if you and he don't resolve it, and I don't like how that comes out.

So looking at my own motivations on getting involved, all I can think of is that we shared a dinner together in Netherlands and toasts and drinks and merriment and camaraderie. I cherished that night.
More of those would be good. Life is long, but opportunities can pass swiftly.

 

[NewLiberty]

Please read my post right above yours. It gives a further explanation.

I'm sorry that you're upset about this but the answer is t = 15 min. I think Craig, yourself (and some others) are angry because you think I'm trying to trick you somehow. Craig even mentioned that I had cheated. But this is not a trick question.

Many people were fooled by this. @bitcartel wrote a selfish mining simulator where he did the same thing (he assumed the expectation value was unchanged by the fact that the SM found the next block). And so he got numbers more like Craig's. I think he is close to "getting it" now however. He is approaching it with an open mind.

There is nothing wrong with making mistakes or not understanding something. The first time someone asked me the question I asked about the expected time between blocks if I check at a random time, I said "10 minutes." I thought I understood what it meant for mining to be memoryless but I still didn't get it at a deep enough level to see why the answer was "20 minutes." I bet there are still aspects of memoryless processes that I don't understand.

If you want further evidence that Craig is indeed making the mistake I claim he is making, just look at the first sentence of the paper he removed from SSRN:

By "events" he means finding the next block. He is implicitly making the statement that mining is not memoryless right here. These are the first words out of his mouth. But because the Gambler's Fallacy is counterintuitive, he still doesn't see it.

The gamblers fallacy isn't counter-intuitive. At least not for me I played many years of dungeons and dragons and developed an intimate relationship with probabilities. It was quite intuitive before I hit puberty.

I've not read Craig's paper yet, don't know which events that passage refers to. I imagine that piece of the introduction is leading to a discussion of the independent and conditional probabilities or HM/SM distribution being a negative binomial rather than pure poisson, (but I couldn't say without reading it), just because those where the things being discussed when you brought it up.

I'm certain that you are teaching us more about who you are, than anything to do with probability calculation. Check your hubris please.

 

[Zarathustra]

Except that SM isn't real, and doesn't work until at the end of the world when the only activity is Bitcoin mining.

Is your position that selfish mining cannot be mathematically modelled just as the economy cannot be mathematically modelled? I think the laplacian demon is the only one who would be able to calculate it. The processes are 100 percent deterministic, but orders of magnitude too complex to calculate.

 

[Peter R]

@NewLiberty: I don't think you realize it, but the more you write the more evidence you're leaving that your understanding of selfish mining -- and bitcoin mining in general -- has a real blind spot. I'm going to stop responding to you now because I don't think this is constructive any longer. You appear to be getting more angry rather than trying to understand.

Think about it: Right now you are angry at me because I think that selfish mining is real.

 

[Zarathustra]

@Peter R
"Think about it: Right now you are angry at me because I think that selfish mining is real."

Is it real? Does it happen?

 

[Bagatell]

Check your hubris please.

Amen.

 

[satoshis_sockpuppet]

While I think, that Selfish Mining is far away from a real world problem (simplified world), I haven't seen any substantial criticism in this thread. Also, selfish mining criticism and the bet between Peter R and CSW are two different topics. CSW's paper is not good (and he pulled the paper, so he seems to think the same..) and he definitely lost a bet.
And I'm disappointed to see you, @NewLiberty trying to argue from authority. Peter made clear and falsifiable statements, that you argue like this

If you do not see how he will perceive that you tricked him, and betrayed him

just shows me, that you've seen, that you are wrong but can't admit it.

Is CSW now the untouchable master mind?

The Bitcoin whitepaper was very well written, to the point, without unnecessary palaver but detailed where necessary.
The stuff CSW/nChain put out so far is of very, very bad quality. That is just how it is.

 

[go1111111]

So now I am pretty much flabbergasted that it has come up again. I don't get it. My most generous explanation is that personal loyalty has blinded you to all of this, and that you are OK with sacrificing everything you are doing for a friend, even though the friend is a naked emperor.

It's interesting to note that "naked emperor whose support comes entirely from personal loyalty" describes Craig perfectly. There are many people who agree with Peter about the selfish mining paper who strongly dislike him. I don't know anyone who agrees with Craig's selfish mining claims who isn't one of his fans.

Has any technical person with a public reputation ever publicly defended Craig's technical arguments? If so can someone link to this? If not, isn't this odd, given that Craig's technical arguments are publicly derided by so many technical people?

 

[lunar]

Hard Fork incoming

T minus 6 blocks and counting

 

[throwaway]

Idea: should block 478559 be made a checkpoint?

 

[adamstgbit]

this is exciting!
i wonder how much hash rate is looking for the first block.

 

[Dusty]

I almost fell of the chair laughing when a friend of mine coined this new definition paraphrasing the well known AB's one: