Gold collapsing. Bitcoin UP.

[chriswilmer]

@Peter R Cool! I hope there will be a good summary of the conference somewhere. I'm swamped with end-of-semester stuff right now... I'm cautiously optimistic about how all this will turn out

 

[cypherdoc]

oh gaud, Todd is up...
[doublepost=1449371629][/doublepost]ha! Todd bringing up a competitive motivation for SPV mining vs the relay network. now just who has been advancing that theory for many months now?
[doublepost=1449372159,1449371326][/doublepost]@Peter R

i have to believe your talk would have been orders of magnitude better than this crap we're hearing from Todd.

in fact, i'm sure of it, given the quality of your Montreal talk.

 

[theZerg]

I occasionally get anonymous tips, and a few I've received recently are relevant to this observation:

About a week ago, I received one about "panic at Blockstream about no one wanting to use bitcoind to run their business". Then, two days ago: "Lots of large Bitcoin companies are uncomfortable with the direction that Blockstream has been taking Bitcoin Core recently, and are looking hard at alternative implementations."

In light of those comments, what you're saying about the schedule, makes the conference sound like a last ditch effort to stem the bleeding of their userbase.

Show them BU if you can. I spent weeks on the Articles in the hope that Bitcoin companies would see how this structure encourages a collaborative environment where decisions can get made (I mean it is better for a decision to go against you so you can execute your contingency plan rather than no decision to be made at all) and where a minority cannot hijack the process.

And they don't have to offer $ or engineering resources up front. They can propose a feature in a BUIP and if it gets accepted then they can do the engineering and the code will get merged without subsequent decision making. This structure is much more friendly to external development than one where you first write all the code and only then does the project decide if they will accept it.

 

[Peter R]

Great comment to Peter Todd: "We in China mine differently...everyone want to protect Bitcoin...not attack...that is why we are here."
[doublepost=1449373397][/doublepost]The next speaker, Jonathan Bier, is jonny1000 on Reddit. He is a friendly and polite guy and we had lunch together in Montreal. He asked me to review his slides last week (which I did). He told me (in a friendly way) that he's going to "prove my theory wrong." Let's see...

 

[Justus Ranvier]

I wonder if Peter Todd has ever been in the position of making business decisions for an organization equivalent to a medium-sized company or larger. Something with more than $10 million annual revenue.

It's one thing to hypothesize about game theory models in a vacuum, and it's an entirely different thing to exit the ivory tower and actually go out into the world to test how well your models compare to reality.

Cryptographers are right to call out non-qualified individuals for over-confidence in their own field; I wonder how many of them realize the same applies to them in other fields?

 

[cypherdoc]

Great comment to Peter Todd: "We in China mine differently...everyone want to protect Bitcoin...not attack...that is why we are here."
[doublepost=1449373397][/doublepost]The next speaker, Jonathan Bier, is jonny1000 on Reddit. He is a friendly and polite guy and we had lunch together in Montreal. He asked me to review his slides last week (which I did). He told me (in a friendly way) that he's going to "prove my theory wrong." Let's see...

i'm not hearing any proof. he even admits his is conjecture.

i have to believe your talk was censored b/c it didn't tow the Blockstream party line.

 

[Peter R]

I think Jonathan gave a well-balanced and open-minded talk. He promoted BIP100 but he removed my biggest objection to it (that the voting wasn't based on a straightforward 50% majority). Well done, Jonathan.
[doublepost=1449374266][/doublepost]"i have to believe your talk was censored b/c it didn't tow the Blockstream party line" -- @cypherdoc

I agree. I now have a third informant indicating that this was true.

 

[JVWVU]

Still freaking clueless why we cant got to 4 or 8 MB at halving and work on a well balanced scaling after that for 2 years.

Blockstream is clearly pigeon holing this.

 

[solex]

The IBLT presentation was a real window into how efficient block propagation can become.

Correct me if this impression is in error, but it seems the conclusion is 1MB blocks can be propagated in 20KB messages with a 95% chance of successful decoding by the average full node.

My first thought is to forget about weak blocks at the beginning, leave that for a subsequent phase. Let's see whether IBLT can reduce bandwidth usage with a simple implementation, done sooner than later.

Also, it is very interesting that full node mempools are much more similar than compared with the contents of mined blocks. This means that miners are stuffing in loads of out-of-band tx as a matter of normal operation. The whole success of a PoW secured blockchain is undermined if the tx which go into the blockchain are not subject to a certain level of consensus. Garbage in, garbage out. Ideally, all tx which are in mined blocks should be seen by the whole network in advance, and had a chance to be present in the majority of node mempools. It is a cryptocurrency functioning most efficiently. This makes apparent another strength of IBLT, that it can enforce consensus on unconfirmed tx, making for a cleaner blockchain.

 

[theZerg]

@solex can u expand on the evils of out of band txns. What if I told u blk size limit might encourage the practice?

 

[cypherdoc]

great talk by Toomim.

very encouraging for 101.

 

[solex]

@solex can u expand on the evils of out of band txns. What if I told u blk size limit might encourage the practice?

I'm not a miner, so I am not privy to all the various reasons why out-of-band tx are added. A classic example though is pools paying their hashers. Logically, the main reason that tx are added to a block, that are not pre-broadcast, is that they carry insufficient fee and would not be propagated. Also, they can circumvent general policy, an example is the "bloat" tx which were sent a few months back with so many inputs that the block took 25 seconds to validate on most nodes, Originator intentions were good (mopping up UTXO) but these transactions exceeded the relay limits observed by the majority of nodes.

It's a safe assumption that a tx which most nodes don't see before it goes into a block has a higher probability that it wouldn't have been relayed or mined by anyone else, ergo, tx which are seen by everyone else are more likely to be valid real-world business, and not spam, advertising, or otherwise "less desirable" tx.

I had not considered that the block limit affects these, just that IBLT helps reduce them.

 

[cypherdoc]

the whole concept of bigger blocks somehow exacerbating the Chinese advantage, ie mining centralization, is terribly misplaced. there is so much fiat money waiting on the sidelines outside of China to see if Bitcoin will allow itself to scale. with bigger blocks, there would be an explosion of mining pools outside of China leveraging their bandwidth advantages. that would be a very good thing resulting in much greater decentralization of the network. and the Chinese miners could compensate using block construction outside of China according to jtoomim.

the entire progress of such an important new revolutinary monetary system that depends on pushing the fringes of technology cannot be held up by shitty bandwidth connections behind the GFC enforced by a Communist gvt. Bitcoin needs to technologically continue to outrun all gvts worldwide.

 

[Zarathustra]

Greg is out if we have big blocks:

"that kind of system just isn't interesting to us"

Indeed, not interesting to Blockstream, if the stream wouldn't be blocked anymore.

raisethelimit :

"Looks like the price rallied right around the time you made this comment. But what do market know, right? That's why you and Blockstream want to micromanage them. You know better than all the foolish miners and node operators"

 

[awemany]

@awemany : I was travelling all last week and didn't have time to comment, but I wanted to mention that I really liked your work trying to quantify the cost of writing transactions to the Blockchain. It didn't get the attention it deserved on Reddit, but I think it could with the right presentation. I hope you continue to develop these ideas!

Thank you! But as you can see it is really an almost trivial calculation. That said, it can be extended a little (for example, include (repeated) IBD cost, loss of HDDs, cost of online vs. offline storage). I sincerely doubt one will arrive at realistic figures that are more than a magnitude higher.
And with UTXO commitments, I can only see the cost to operate a full node falling.

In any case, I would expect something like this to be the starting point informing a productive discussion about needed decentralization and node cost.

In a discussion on reddit, I told gmax that I'd be fine with a 1000 full nodes scattered worldwide, I consider this enough decentralization also in the end game.

He explicitly said that this is a Bitcoin he doesn't want to work for (so basically indirectly admitting that he's steering it off-course right now). At least a negative data point. But it should be easy for him to state his needed level of decentralization.

It is a gaping hole in their whole argument - concrete, realistic numbers.

I am also pretty sure that Gavin is not going to be too far off with BIP101 - whilst ensuring exactly what BS supposedly wants: Let simple people be able to run full nodes still.

On that note, we were talking about a related idea in the old gold thread about how it is actually cheaper for the network to write a backlog of spam to the network than to reject it:

Yes, I saw that back then on the original thread, but thanks for reminding me. This is indeed a calculation that is very similar.
[doublepost=1449394672][/doublepost]@Zarathustra:

Another data point. A thousand full nodes worldwide is not enough for him.

 

[Zangelbert Bingledack]

Middle of the scaling conference and Gmax is posting giant comments in random minor threads on the alternative subs. Also look at the content: he's been paying close attention to recently comments and criticisms such as Charlie Lee's litecoin holdings, twisting Satoshi's phrasing to make it look like it supports his position - generally a masterpiece of damage control:


But why?

 

[albin]

How are Core devs supposed to talk to miners?

It's apparent on forums / reddit / social media you're only allowed to engage Core devs using their words, cliches, and memes, otherwise you just get lectured about how stupid you are for using certain words in an unapproved way.

Now these guys are supposed to talk to a bunch of guys of varying technical background from a completely different culture, who have largely not participated at all in any of the English language discussion media?

Sounds legit.

 

[albin]

I think "adversarial thinking" is becoming a new meaningless buzzword.

The comments in the coindesk article (http://www.coindesk.com/chinas-miners-heat-up-blocksize-debate-at-scaling-bitcoin-hong-kong/) from Poelstra's talk are very troubling to me (emphasis added):

“A lot of society is about limiting adversarial behavior,” Poelstra said. “Online things are anonymous, pseudonymous and difficult to trace. If it’s possible to hurt the system, someone will do it. We can’t assume they’ll be caught.”

In his remarks, Poelstra discussed how, under these conditions, even variables that are statistically unlikely need to be considered seriously given that it is expected that the systems in place today will be built on in the future.

“When we move from traditional cryptographic assumptions to more nebulous region of incentives, economics and trust ... you assume people know each other or won’t try to screw each other. This is not the world bitcoin lives in,” he continued.

Is this part-and-parcel of a movement to further trivialize and denigrate the contributions of anything outside of strictly CS and applied mathematics?

 

[cypherdoc]

@albin

keep reading Andrew Miller's thread. he seems to one CS major, at least, who understands a key difference btwn the fields of distributed systems and cryptography; hard solutions vs probabilities. and the melding of the two through Bitcoin along with economics and game theory.

 

[Justus Ranvier]

I wonder how much of the drama is based on certain individuals' desire to stave off the realization of their worst fear:

Bitcoin is actually protected by people instead of math.

This is not a view that presents a problem to the market economist - their entire field is based on the premise that individuals only need to be allowed to act freely and they'll create a self-organizing and self-optimizing economy due to the natural effects of incentives.

From that point of view, Bitcoin's protocol rules are tools for removing ambiguity that cheaters use to exploit in other monetary systems and so gives the cooperating players a decisive advantage.