By the way, there is no excuse for the cost to be quadratic.  That is one of the many crocks in the BitcoinCore implementation, that will take more crocks to work around.  Like the Segregated Witnesses proposal,  malleability and its partial patches, blockchain voting to increase the limit, etc..

There you have another possible failure mode for Bitcoin: runaway code crockification (RCC).  As the code gets more complicated and ugly, fewer competent people will be willing to work on it.  Their place will be taken by incompetent pople, who will add even more crocks -- and so on until the code will fail and there will be no one capable of fixing it in time.

Just a possibility; but after seeing the malleability problems,  the Fork of July fiasco, the "fee merket" plans and the RBF hack, the Seg Wit proposal -- I fear that the RCC may be already underway...

That is small-blockian FUD.  (Rusty is a Blockstream employee, BTW.)

There is an easy fix for that problem: limit the number of inputs and outputs to a sane value.  There is no real need to have more than (say) 16 inputs or 16 outputs.  If one needs more, just use several transactions. 

The problem is that the cost grows like N^2 for N inputs.  Thus the cost for a single 256-input, 1-output  transaction is about 256^2 = 65536 times the cost C for a single-input, single-outout one.  The same effect can be obtained with 17 transactions with 16 inputs each.  The total bytes will be only a bit more, and the validation cost will be 17 x 16^2 = 4352 times C, or 5% of the big one.

IIRC, BitcoinXT already has a limit on the number of inputs (a lot more than 16, but a lot less than 20'000) that solves that problem.  That fix should be in BitcoinCore too; but I would not be surprised if Blockstrea blocked it too.

That is true.  One should not choose between small-blocks or big-blocks just because "Satoshi said so".

However, looking at why bitcoin was created helps understand why it has certain features, what consequences they are expected to have, and whether it is a good idea to change them.  

Would it be a good idea to modify a car so that it floats, in order to get a vehicle that travels on water?  Obviously not: because every detail of the car's design, from the four rubber wheels to the placement of the license plate, was determined by its original goal of traveling on roads.  Those details are totally wrong for the goal of traveling on water.  If you want a vehicle for that purpose, you should design one from scratch, starting from that goal and letting it guide all choices along the way.  The result will be totally different, of course: a boat instead of a car.  Even if your goal is a vehicle that can travel on both roads and water, the result will not be at all like a car.

And you had better justify why you are designing a new boat, rather than buying an existing one.  Just because a new car is better than all existing cars, it does not follow that adapting it to float on water will yield something that is better than all existing boats.

In this case, reading Satoshi's writings can help us see and understand bitcoin's limitations and capabilities.  It cannot scale enough to replace VISA, and it will make a lousy medium for large settlements, because it was not designed with those goals in mind.  

If we understand the reasoning that led to certain details of the design (like the 1 MB limit and the abrupt halvings of the reward) we have a better chance of predicting what would happen if we changed them.  

Those who want to reform bitcoin so that it replaces VISA or ACH should put bitcoin aside and start the design such a system from scratch, choosing at each step the gears and rivets that are better suited to those goals.  But, first, they should justify why the world needs a better option for those goals and why they think that they can design one.

(That said: in fact, I believe that, as a software engineer, Satoshi, was much better than Gavin and Mike, who are much better than all the Blockstream developers -- who are totally incompetent and irresponsible in that regard.)


So you think that 1 000 000 traders could have designed bitcoin?  

The market is only the average of all the trader hunches.  For a purely speculative asset like bitcoin, the hunches cannot contain any intelligence, and the average is still a random value.  The bitcoin market is like a headless pig flying in circles, chasing its own tail.  (Hm, wait, I think I must work some more on that metaphor.)

A fellow computer nerd, I presume? 

It is not clear what Satoshi thought of bitcoin in Oct/2010.  At that time, the project was already giving signs of drifting away from the goal that he stated in 2009.  (Who knows why he left the scene abruptly, shortly therafter; but one theory is that he was smart enough to see that the protocol would fail to achieve that goal, and lost interest in the project.)

There is no sign that he ever intended bitcoin to be a replacement to the traditional payment system (cash, credit cards, bank wires, etc.), which, as he admits in the whitepaper
 Surely he was sensible enough to realize that, while bitcoin could scale to VISA size in the distant future, it could not compete with it in all the comfort features that the traditional systems offer but bitcoin lacks.  

So, he did not intend bitcoin to be used for buying coffee or groceries, sure. But he did not intend it to be used for buying cars, houses, or space shuttles, either.  Or for large-volume settlements between banks or big corporations either. Or to be a general-purpose shared database for things like BitDNS.  

Bitcoin was designed to be It was meant to render a service, not to make anyone rich.  If it had an adoption goal, it was only to be available whenever two parties needed to make an internet payment, and a trusted intermediary was not available or desirable -- and that need was strong enough to overcome the limitations and inconveniences of the system.  For normal law-abiding people, such situations should be few and far between.  Even today, the system would be quite capable of handling such traffic.
 
If we assume that he was still clinging to his original vision, that message quoted by @AlexGR does not mean what it may sound today.   First, in his vision, there should be only two kinds of players: the miners (that he called "nodes"), and the simple clients that did only limited validation, followed the majority chain, and trusted the miners for the full validation.  (There was no provision in the protocol for the non-mining relay nodes that are now claimed to be the Guardians of the CryptoRevolution, which in fact  are the sort of middlemen that bitcoin was supposed to get rid of.)  

Moreover, Satoshi did expect that, as the volume increased, mining would be limited to entities with a stake on the network's wellbeing.  But he obviously assumed and hoped that there would be thousands of independent miners with comparable hashpower, scattered all over the world.  That assumption was needed to exclude the risk of a "majority cartel" -- a subset of the miners holding a majority of the hashpower, who conspired or were forced to act against their immediate financial interest. Such a cartel could completely block the network, and therefore could blackmail the other miners and clients into accepting arbitrary changes to the protocol.  Majority voting weighted by proof-of-work is essential for the protocol to work at all; therefore, if such a cartel forms, there is no way to protect the protocol from its abuses.

From the whitepaper:
 Note that he was not even assuming that the transctions would propagate through all miners.  

Quoting the first version of the bitcoin.org website:
But it is clear that he never intended for some clique of developers to put a limit on the size of blocks to force all cients to compete for block space by raising their fees.  That would not be an "open market", but a centrally-planned market reminiscent of the Soviet economy (only dumber).  He clearly intended that miners would operate as any business in a free market.  Namely, each miner figures out his optimum fee (the fee that maximizes his net revenue), and then processes all the transactions that he gets that pay that fee -- expanding his bandwidth and servers as needed.

To be sure, that part of his plan seems rather fuzzy.   Why would miners volunteer to process an arbitrary amount of anonymous transactions for free?  How would clients get to know the cutoff fees of each miner?  How would the miners find their optimum fee (that depends on other miners's decisions?  And so on.  I don't know whether he ever clarified that part of the plan.

Although Satoshi comes out as a competent software engineer and a fairly sensible person overall, his knowledge of economics was clearly limited, with the misconceptions that one would expect from a computer scientist (like myself, I must confess).  See for example his belief that a good currency should be inflation-free, his decision to reduce the reward by abrupt halvings intead of a gradual decay, and his failure to foresee the speculative bubbles and the inevitable concentration of mining.

I care... but I have no news either.

Suppose that you have 10 BTC and 5000 USD in each of the exchanges X and Y. Suppose that the price has been rock solid at 500, with plenty of liquidity and tiny spread, in both exchanges.  Your total worth is 20 BTC and 10000 USD.

Price drops in exchange X, and you see an ask of 1 BTC at 490 there.  You buy 1 BTC at X for 490 and sell 1 BTC at Y for 500.  Suppose that it raises the price at X back to 500 BTC.

You now have 11 BTC and 4510 USD at X,  9 BTC and 5500 USD at Y.  Your total worth is now 20 BTC and 10010 USD -- you made a 10 USD profit.

That is the basic idea.  But you can buy and sell different amounts of BTC at each exchange.  If you buy where the price is lower and sell where it is higher, you can increase your net worth and push the prices towards equality.  Note that you don't need to transfer BTC or dollars, as long as you have reserves at both ends.

I don't know much about arbitrage (the players seem to be a rather discrete bunch).  This is what I can infer and guess:

Arbitragers are racing against each other.  As soon as they see an arbitrage opportunity -- an "ask" at one exchange that is significantly lower than a "bid" at the other -- they must quickly execute the trades, before some competitor does it and destroys the opportunity.  

Typical delays seem to be tens of seconds or less.  There is not enough time to actually send the coins from one exchange to the other, not to mention fiat.  So arbitragers must keep fiat and BTC balances in both exchanges.  

While the price is wandering randomly, the trades tend to go both ways, so the accounts only need occasional adjustments.  When there is a big rally pulled by one of the two exchanges, however, the trades will happen mostly in one direction, so the arbitrager may run out of reserves.  

For example, if Huobi is leading a rally and Bitstamp is lagging, the arbitrager keeps buying BTC with dollars at Bitstamp and selling BTC for yuan at Huobi -- until he runs out of dollars at Bitstamp, or of BTC at Huobi. In the first case he must withdraw the yuan, convert them to dollars, and deposit the dollars at Bitstamp.  In the second case, he must withdraw BTC from Bitstamp and deposit them at Huobi.  

In the first case, he is out of the game for a couple of days (unless he has some faster channel to convert the currencies).  In the second case he will be out for an hour or more, including the 20 minutes for the two blockchain transactions and maybe another 30 minutes for extra confirmation of the deposit.  (Someone who claimed to be an arbitrager said that he used litecoins to transfer bitcoins between exchanges, because of the much faster block time.)  If all arbitragers between the two exchanges run out of funds, the price difference will start to grow, until they can top up the accounts.

Effective arbitrage also requires fast access to the order books at the two places.  For that reason, I suspect that the main arbitragers (at least between the Chinese exchanges) are the exchange owners themselves, or certain privileged clients.  The owner can see the order book several seconds before anyone else.  If the two exchange owners cooperate, they can beat any independent arbitrager.  (Unlike stock exchanges, nothing prevents owners of bitcoin exchanges from trading at their own shops, for front-running or arbitraging.)


Could be.  Whether a price situation is an arbitrage opportunity depends not only on the currency exchange rates, but also on the cost of currency conversion and transmission, and on factors that affect the effective value of each currency to the arbitrager, like the one you mention.

There is practically no reliable info on the bitcoin economy, in particular on the flow and ownership of bitcoin by country. (This is a serious problem for would-be investors.)

We can only note that more than 67% of all new bitcoins are mined by Chinese pools, which probably comprise mostly Chinese miners; and that bitcoin has practically no use inside China, except as an instrument of speculative trading inside the exchanges.  Until last October, variations of trading volume at those exchanges did not seem to be reflected in the USD transaction volume, which may mean that there was little deposit and withdrawal at those exchanges. 

There is efficient arbitrage between the Chinese and non-Chinese exchanges. If Chinese miners sold their coins only in Chinese exchanges, that would tend to depress the price there.  Then the arbitragers would immediately move those excess coins to non-Chinese exchanges, until the prices got equalized.

So, I would guess that it does not matter where the Chinese miners sell: the net effect is that a large fraction (if not most) of the bitcoins mined in China are eventually bought and hoarded by non-Chinese investors.

Are you asking me?  I do not knowingly "invest" in pyramid schemes, just as I do not gamble. (Last time I did, it was some US$ 200 into the Brazilian branch of Amway, in the early 1990s, because of pressure from a family member whose feelings we did not want to hurt.)  

But I find the scams and scammers the most interesting part of bitcoin. Mark Karpelès, Josh Zipkin, Josh Zerlan, Josh Garza (hm, is there a pattern there?), Danny Brewster, Craig Wright... Each case is more incredible than the previous one: if made into movies, they would be dismissed by critics for the outlandish plots...

But why do you think that the MMM theory is wrong?  Have you checked the story of the previous ponzis by Sergei Mavrodi, in Russia and China India?  Check the amounts involved, and compare to the bitcoin market cap...

PS. Even IF the most recent bitcoin price rally (started 2015-11-24) is indeed due to the Chinese crackdown on UnionPay abuse, it may not cause intervention on the Chinese bitcoin exchanegs.

When the Chinese citizens used the UnionPay route, they took yuan out of the Mainland economy, and some of those yuan ended yp in foreign hands.  From the perspective of the Chinese central bank (PBoC), that is bad.  For one ting, those yuan could be ammunition for speculative attacks against the currency; in order to fight such attacks, the PBoC would have to spend some of its reserves of foreign currency.  Anyway, the net result would be the Chinese population becoming poorer relative to the rest of the world.

With the bitcoin way, hoewever, the yuan will remain all in the Mainland.  The Chinese citizens will take the "worthless" bitcoins, cheaply made in China, and sell them to foreign buyers for foreign currency.  Some of that foreign currency will find its way in the PBoC coffers, some will be used by the Chinese to buy property abroad.  Either way, the net result will be the Chinese people becoming richer relative to the rest of the world.

If this analysis is correct, the good news (for traders) is that the PBoC is unlikely to intervene and restrict thet trading of bitcoins at the exchanges.  On the other hand, the price may drop once the Chinese who are buying bitcoin in China for that purpose will start selling them in the "Western"  exchanges. (Maybe this is the cause of the recent drop to ~$430).

By the way, this chart may be evidence that the rallies of the last three months are due to real demand, rather than manipulation in the exchanges.  Note that the total transaction value (in USD/day, excluding return change outputs) in the blockchain, which dropped at the end of 2013 and has been nearly flat for 20 months, climbed suddenly since 2015-11, together with the trading volume inside the Chinese exchanges.  This is consistent with the MMM and bitcoin-exporting theories: hoarders send their BTC to the Chinese exchanges, a different set of people withdraws them, and deposits them to foreign exchanges.

I have read those reports and commented on them several times.  It is fairly certain that there was a Willy Bot buying tons of coins at MtGOX.  However, comparison of prices during that rally tells me that it was not led by MtGOX, but rather by the Chinese exchanges: chiefly Huobi and OKCoin, that had opened in Beijing only a few months before, and also BTC-China in Shanghai.  That is confirmed by articles in mainstream media (not bitcoin sites) that described firsthand who were the new bitcoin traders responsible for that huge surge of demand. 

So, Willy Bot was probably doing arbitrage: buying cheap bitcoins at MtGOX to sell in those Chinese exchanges, where the price was significantly higher.  Quite likely it was buying them with non-existent dollars, with or without Mark's involvement; and that may be the immediate explanation of what happened to the missing 660'000 BTC.

It is possible that this rally is due to a Willy Megatron in the Chinese exchanges, buying coins from other clients with non-existent money; and that those exchanges are about to go the MtGOX way.  But a consequence of the bot's operation was that Mark did not have enough money or BTC at hand to honor the withdrawals of those clients who sold their coins to Willy, and therefore had to stall withdrawals.  I have not seen any claims that OKCoin or Huobi are doing the same.  But let's watch.

PS. Apparently the recent price movements were not related to the news about Craig Wright.  

The guy has been established to be a fraudster who planted lots of false evidence about other things, and his knowledge of computer science seems to be very limited.  Therefore, the evidence that he is Satoshi is likely to be faked as well.

It seems that the main target of his scams was the Australian government.  He created a tangled web of companies that closed various "research" contracts with each other, amounting to more than 100 M AUD; and then he used those contracts to get maybe 50 M AUD or more from government research incentive programs.  He may have defrauded private investors as well.  

He claimed that many of those multimillion contracts were paid in bitcoin; which may have been just an excuse for the lack of bank records proving that the companies had that money to begin with, and that the payments were real.  That may help explain why he started to plant clues that he was Satoshi, including backdated blog posts and a supposedly leaked contract with a deceased American colleague that established an offshore trust fund with 1.1 M BTC, to be locked up until 2020. (But there are several other possible explanations for why he wanted to be identified with Satoshi.)

In particular, he claimed to own the 15th largest supercomputer in the world, the largest privately held one.  He even gave (totally hilarious) masters-level lectures on supercomputer programming at his university, which included lab exercises with remote access to an "access node" of his supercomputer.  He also posted a video on YouTube where he discused his supercomputer and showed an intern working on some visual node management software.  However no one has seen that machine, not even a photograph of it.  On his company's website there was a letter from SGI that boasted the use of SGI hardware in his supercomputer; however, SGI denied ever having any contact with him or his company, and knew of no such supercomputer.

The final leak of "evidence" seems to have happened when the Australian Tax Office had finally realized the extent of the scam and was about to squash him.  It seems that he fled the country just in the nick or time.  Recently he somehow got himself invited to a panel on cryptocurrencies witn Nick Szabo, Trace Mayer, and other guys.  He spoke via Skype, claiming to be in London with his family.  Like the last emails by Danny Brewster of Neo&Bee, that may have been an attempt to divert the ATO investigations to the wrong place...

Looking at the daily price and volume at Huobi, OKCoin, and other exchanges, I can see

(1) A gradual "exponential" increase in volume and price from about 2015-09-20 to the peak (~3350 CNY) on 11-04;
(2) A tumbling price and volume drop between 11-04 and 11-24 (down to ~2050 CNY);
(3) A very sharp increase in volume to a record high value on 11-24, that persisted until now; and a corresponding rally in price, by a rapid steady rise broken by several large jumps (to ~3150 CNY).

There was a partial crash today, to ~2800 CNY, but it is not clear yet how it will end.

There seems to be no certain explanation for these moves.  (And it seems that no one wants to find out...)

My best candidate explanation for (1) is still the bitcoin-based MMM ponzi and copycats, mostly in China; amplified by day-trader speculation.  The dates seem to match, the ponzi can easily move that amount of money, and the gradual "exponential" growth is consistent with the demand for bitcoins spreading among a new population of  by "infection".

The crash (2) would then be the speculators dumping their coins when they realized that the primary demand had leveled off.   Perhaps the ponzi saturated by 11-04, or participants ("lucky victims" and organizers) started to sell the bitcoins that they received.  

The new rally (3) could be due to the same cause as (1); however, considering the much higher volume and the way it grew (suddenly rather than gradually), the cause is probably different.

Rally (3) may have been due to the crackdown by the Chinese government on the use of the Chinese state credit/debit card (UnionPay) to export cash for gambling and other purposes.  The abuse seemed to be pervasive in Macau, the Las Vegas of China, which has a somewhat independent economy (like Hong Kong). Reports say that Mainland gamblers would go to Macau, pretend to buy merchandise at local pawn shops with the card, then pretend to return it for a cash refund (minus the shop's commission).  That way they could bypass the card's limit on cash withdrawal.  If that loophole was suddenly closed, it seems possible that the gamblers switched en masse to bitcoin.  

A sudden jump in the price must be due to single person buying or selling a large amount of bitcoin in a short time interval.  If the above explanation is correct, those may be exceptionally wealthy people trying to move their money out, or perhaps clandestine money transmitters providing that service for many smaller clients.  Either way, the jumps seem to indicate that the price is determined by a relatively small number of players.

He tweeted that, on his supercomputer, he had tested bitcoin up to 340 GB (yes, GB) blocks.  Yet he defended keeping the 1 MB limit?  I seem to perceive a slight discrepancy somewhere in there...

BTW he claimed to have lost quite a few bitcoins in MtGOX, enough to cause  bankrutcy of his Hotwire company (p.28)

SW allows more transactions per block, split into two parts (basically, signatures and everything else).  The 1 MB limit will apply only to the latter.  The total size of a block (which will have to be transmitted and stored in full, except to simple clients who will not need the signature data) can be up to 4 MB, depending on how many transactions are issued in the SW format and how many signatures they have. 

Peter Wuille (Blockstream; Segregated Witnesses developer) explains the shift from "1 MB is the limit" to "4 MB is OK":
https://www.youtube.com/watch?v=fst1IK_mrng&t=1h4m1s
(first two questions at that point)

Reddit thread:
https://np.reddit.com/r/bitcoinxt/comments/3vxv92/peter_wuille_deer_caught_in_the_headlights/

For one thing, development has been stalled over the last year by the feud between Gavin and the small-blockians.  Gavin has coded and tested BIP101 (the only one of those BIPs that has been implemented, AFAIK) several months ago, but has not been allowed to put it into the core.

But, apart from that, a good software developer is not one who makes lots of changes.  The best way to increase the block size was coded by Satoshi in 2010, and it is a one-line patch.  Blockchain voting is a significant but pointless complication, with many things that could go wrong (in the code and its operation). Too bad that Gavin had to add it for BIP101 in order to appease the ideologues.

The problem with Greg is that he does not really like Satoshi's design, and has his own strong ideas on how bitcoin should be.  He has been pushing for that vision for a couple of years at least; and now that he has the power, he is bent in implementing it. 

The Core devs certainly are not in "maintenance" mode.  Just the SW proposal is a major redesign of a data structure that iis shared by tens of thousands of users and hundreds pf independent applications.  One should not be adding all those new features to a production system whose users and app developers are still struggling to cope with.  In fact, with their plans for the "fee market" and the off-chain processing, they are in full "retargeting" mode  One should not try to do a major redesign of the functionality, like the fee market, without knowing whether it will work or not. 

Greg may be a good cryptographer and programmer, but he is not a good software developer. In fact, it seems that he does not know what that means, and does not want to know ...

Reading several things, including the dev mailing list (including that letter by Satoshi, and various bickerings between the devs until Gavin and Mike left. And Greg unsubsctibed.)

But almost no one reads the dev mailing list, or even the forums; and very few understand what it is discussed there.  Saying that "alerts are not needed because it was discussed in the dev mailing list mor months" reminds me of a certain incident in The Hitchhikers Guide to the Galaxy...


Actually I have been programming almost daily for the last 40 years.  And I have watched from front row some of the best software developers in the world as they worked.  As well as some terrible ones.  

That is why I think that swapping Gavin for Greg was a disaster for bitcoin...


Care to name some?  

Perhaps you mean my belief that Blockstream and Viacoin are "conspiring" to retain control of the protocol and ensure a congested network because their business is to sell tools for off-chain transactions?


You are looking at miners adoption, not relay nodes or clients, right?

I meant that, in general, the devs like soft forks because they can be deployed in "stealth mode", without alerting all the users and nodes -- and therefore avoiding "wasteful" questionings and explanations.  See BIP66 last July, and BIP65 that was just enabled yesterday.

In this case too, the devs seem to be decided to roll out SW without waiting for it to be scrutinized and approved by the community.  (No independent applications will break, of course,  unless they deserve to break.)

I don't know whether SW is good for the devs overall, but some features at least are needed for their planned "bitcoin 2.0" products.  I have yet to see why it is good for clients, or even for the relay nodes (the witnesses still have to be relayed, and blockchain pruning can be done even without SW.  IN fat, that is what the UTXO database is -- a pruned and indexed copy of the blockchain.



I am not assuming that.

I am assuming that, with a 1 MB limit, the network will each saturation in mid 2016. (Although there has been an extra increase in November, so that may happen in Q1 already.)  I am also assuming that, six months after SW is deployed, half the traffic will still be using the old format.  Then, even with half the traffic in SW format, the network will saturate anyway  in 2016, but perhaps in the second half only. 


I am totally confident that the core devs will deploy such a change in 1 month, with holidays in the middle.