= 27d567bc2d48d9f40e9ccc20ea370b15)
John_Maxwell_IV comments on Bitcoin Cryonics Fund - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (67)
As far as I can tell, the recent rise in Bitcoin prices is overwhelmingly due to speculation, and is probably a repeat of the 2011 bubble on a larger scale. If you google "bitcoin bubble", there's lots of info on this.
I'm not a quant, but if you're holding bitcoins, I'd recommend selling off half of them and then rebalancing (keeping the value of your MtGox account 50% bitcoins and 50% dollars) going forward, which basically allows you to make money off of volatility in Bitcoin (while also having the potentially useful side effect of stabilizing its price and making it more viable as an alternative currency). (You can read about the theory behind rebalancing here.)
That was my own view back when it was around $60, but now that it's ~$233, I've been reconsidering (because, well, my predictions were wildly wrong, and that seems like the sort of thing which ought to make you rethink things).
What drivers do we have? Well, there's:
Many of the catastrophic Bitcoin failure modes have ceased to materialize:
real markets can sustainably spike, and it is not an infallible sign of a bubble. One of the arguments long since made in favor of buying and holding stocks, and not attempting to time the market, is that the bulk of your stock market gains over the 20th century comes from the spikes on just a handful of days; I've seen as few as 10 days cited as giving the lion's share of appreciation. (I wonder if this were predictable just from EMH? If all spikes failed to hold, that seems like an easy money-making algorithm: any time there's a spike, short it.)
More generally, if Bitcoin were the Real Deal, we would expect large price increases as people learn of it and it directly gains value from increased use, an ecosystem slowly unlocking the fancy cryptographic features of it like all the kinds of smart contracts it enables, etc. This is the same problem as startups: how do you tell Pets.com from Facebook? In retrospect, one is a fool's bet and the other a genius investment, but we all know how much retrospective assessments are - and ultimately, in 2013, Amazon.com sells plenty of dog food, with 26k hits for
"dog food".It strikes me as being exactly the sort of thing you saw in industrial economies on strict gold standards before central banks. When you have an expanding pool of goods and services (either via industrial expansion, or the market for a type of money increasing) being chased by a money supply that is growing significantly slower, money deflates rapidly, and some people decide to hold onto their money rather than spend it as they think it will increase in value. This effectively shrinks the money pool, making the effect even stronger. Then the speculators come in, and it gets hairy and unpredictable. Eventually the money value collapses, and it starts being used for currency again.
You should EXPECT massive booms and crashes as a new currency with a fixed rate of expansion grows in use (or in speculation-attractiveness) faster than its supply does.
So maybe the bubble is still growing... I don't see how that should make you reconsider much. I wouldn't be super surprised to see bitcoin hit $1000 or greater, but I also think it's likely to lose 50% of its value or more at some point in the next 12 months.
I don't think bitcoin is a bad long-term investment. Maybe not even at the current price, for 1% of your portfolio or whatever. But I do think the recent rise has been overwhelmingly speculation-driven, and a panic is pretty likely. I think Bitcoin's price as a commodity isn't necessarily related to its use as a currency, either. The more volatility in bitcoin, the more merchants (like reddit) will convert out of bitcoin after completing a transaction. We can do this kind of fundamental analysis all we want, but realistically I think speculation has been and will be the overwhelming factor affecting bitcoin prices... and speculation is random and hard to predict, hence the rebalancing recommendation. (Has there been any investigation in to bitcoin transaction volume by speculators vs real usage?)
But it sounds like you're a lot more informed on this issue than I am. I'm also sick right now and therefore less intelligent than usual :P
$133 as of now. (Bitcoincharts.com is down, but I think I saw it at around $260 this morning.)
Mt.Gox is giving a high of $266 and a low of $105 right now, so in that sense it's lost ≥50% of its value already. But that's exaggerated by cherrypicking the high & the low, and the exchange rate's hovering around $170 now. We'll have to wait and see whether it returns below $130ish and stays there.
The longer a 'bubble' goes on without popping, the less it looks like a bubble and the more like a permanent price increase.
The only one I can think of is the Shamir blockchain paper, which since it's using the blockchain, is a rather naive and guaranteed-to-be-wrong approach.
cough
Heh, apparently I have absorbed Sumner's thinking to the extent that I almost wrote his post for him:
Ah, the nay-saying begins. "Oh no, it's only up 5x over the last month and not 10x!" Remember that the predictions I was talking about blowing were that it would peak in the $30s or $40s; I and and anyone like me was wrong, still are wrong, a fall to 'just' $100 does not make me feel much better about having blown them, and I will still have learned that there are enough people willing to buy into Bitcoin for any reason that its exchange rate can be pushed all the way to $260.
More generally, just as a large price increase does not prove that Bitcoin will ultimately not be a bubble and will go to $100k, a large price decrease does not prove that Bitcoin will ultimately be a bubble and go to $0. We of all people should know to not apply double standards and hindsight bias.
What kind of smart contracts does it enable?
The single best resource seems to be https://en.bitcoin.it/wiki/Contracts but it only covers a fraction of the possibilities; for example, I corresponded with Hearn about timestamping for time-lock crypto but that hasn't been added.
Excellent! I had no idea the bitchain included all those extra fields. So much potential!
Yeah; people don't realize how Bitcoin can do so much. It's not (just) sending some irreversible transactions around, as useful as it is to replace Paypal/WU. It's like every financial cryptographic construct ever got mashed up in an academic paper, except it's not a paper, it's a real system in use worldwide and the constructs can be actually used by real people for real things. Of course none of them are easy to use, many aren't even enabled - but I think the situation is like the early Linux kernel: it's a rare geek who understands why you would want to hack on a kernel, a rarer techie who can compile kernels and mess with kernel settings, and an even rarer techie who can write a kernel module to do something, but the potential and power are there.
You have just created "a rare geek who understands why you would want to hack on a kernel". These features are screaming out to be used and may facilitate some of the projects that I have in mind (including some plans for small-scale crowdfunded research).
Do you have any other resources on bitcoin to recommend? In particular I a resource that lists definitively which features are enabled and ready to use in current clients? (I hate accidentally building or planning things that rely on features not yet in the mainstream software. It feels like crippling myself when I try to backport.)
No idea. I have a vague understanding that colored coins are a work in progress and may be added at some point in 2013, but I've never seen anyone lay out a list of the scripting language elements with the enabled and disabled ones specified. Since this is such an obscure area of Bitcoin, you probably will have to read the source to find out.
Time for me to do an expected value calculation. Is it worth becoming an expert on bitcoin? Not at all implausible. Apart from the intrinsic value of personal interest there seem to be things that the world needs that bitcoin could help with (an actual prediction market that doesn't suck for example). Some things that the world needs it is possible to create ways to be paid for doing. It aligns well with my aptitudes.
A very tempting rabbit hole to jump down.
a distributed prediction market and distributed exchange with built in escrow would do wonders for making bitcoin an attractive financial instrument.
Creating a bitcoin prediction market is a great idea.
One thing I've been wondering recently is why more people don't create Bitcoin alternatives. Many of the alternatives that already exist have ridiculously high market capitalizations, with what appears to be minimal marketing and almost zero rationale for their existence. If you could create an alternative that had "baked in" solutions to some of the problems people complain about re: long-term adoption of bitcoin (limited divisibility, difficulty of transferring in/out of USD, volatility (somehow fix this? e.g. if the protocol was somehow robust to high-frequency trading, perhaps that would deal with volatility as HFT people traded using volatility pumping strategies), performance problems as the network grows large, general awkwardness of waiting for confirmations and stuff), or had some kind of killer app that bitcoin didn't have, and then put an actual marketing push behind all this (and bribe the existing exchanges to accept it or find a way to piggyback on the existing bitcoin infrastructure), it seems like it could be a great way to make money.
I have been looking into the protocol for how this works. It doesn't require anything additional to be added (to the bitcoin daemon). To be practical it does require that clients are made that process them conveniently, this is what is the work in progress and does seem to be at the 'usable in a stable form some time this year' point.
The above distinction matters to me because it influence whether it is something I can just use if I find customers who want it and I am willing to do some coding to facilitate it versus features that require creation of a Bitcoin Improvement Proposal then the politically influence of a miners to start adopting it. For example zerocoin (the solution to anonymity and privacy) would require huge political advocacy to get into play in the existing bitcoin currency.
Indeed, my research has failed to turn up such a document too. The information that is out there is also significantly out of date unless pieced together rather carefully. A summary of the availability of the features that are of interest. I am reasonably confident in these findings but not certain. (90% confidence.)
Done. Now I'll have to read the source even more to work out how to make clients that use the (available) features to serve my purposes.
Well, good luck. I don't know a lick of C++ or real crypto coding, so as interesting as I find it all to read and watch and write essays about, don't expect any help from me. I hope you succeed in getting some of the features in more general use.
But also risks there being some minor problem with the unpublished Zerocoin proposal which de-anonymizes everyone who ever uses the functionality. Since laundries/mixes currently seem to be working pretty well, it's better to be conservative and get it right, than push for it to be implemented right now and in 2 years, watch every transaction be de-anonymized.
On a related note do you have any thoughts on ripple? It seems to offer huge potential convenience if 'distributed payment processing', which could work well to enable bitcoin to be easily spent and traded. From what I can tell their business model is based on making an essentially free, incredibly convenient service and if it takes off it will produce value in their own currency of which they will (allegedly) be distributing 80% for free. I don't precisely understand how it magically operates as a distributed currency exchange and payment processor but it seems plausible (if enough people want it).
Reading about it right now, I would analogize it to an automated barter system, the hawala money transmission system, pre-modern banking's system of drafts and correspondents like the Medici Bank, and Szabo's bitgold minting.
If I'm understanding the wiki documentation right, you can basically think of Ripple as a Bitcoin in which everyone can issue as many bitcoins as they want but each user produces a different kind of bitcoin. The only reason anyone is willing to hold a wedrifid-coin is because they know you in real life and expect you to pay up, or the reason someone is willing to hold gwern-coins is because I have in the past upheld my agreements to, say, order something off Amazon.com for them. I might decide to accept wedrifid-coins and now the wedrifid-coin holders have the option to either shake you down in real life or buy something off Amazon via me; and if someone else decided to accept gwern-coins, a holder of wedrifid-coins could convert to gwern-coins and then convert to someoneelse-coins.... Automated barter of personal IOUs, almost - like those circles you hear of involving scores of people in which A gives something to B, B gives something to C, and so on to Z who gives A what A really wanted (most famously used in kidney transplants; I need a transplant, but my brother is not compatible with me, so he gives his kidney to B, and B's son gives a kidney to C, and C's father gives me a compatible kidney).
Since you're deciding whose coins to accept directly or transitively, Ripple seems to pass the security buck to your decision there. This obviously makes Ripple's job much much easier. Since you'll be deciding basically on social or extra-network factors, Ripple also doesn't bother with any worries about anonymity. (Their list of features re mining, if I'm understanding the architecture right, less than impressive in this light.)
There's also a centralized aspect which is nasty and probably part of how they plan to monetize it.
So yeah, interesting. I don't feel compelled to trade in my one bitcoin for any Ripple stuff, though.
I think it operates pretty much the same way a PGP web of trust operates. You specify a few public keys who you trust, and you can bootstrap from there.
So (I assume) the way it can operate as a distributed currency exchange is if there is someone with Yen out there and someone with USD out there and either Yen-guy or USD-guy trust each other directly or their is a web of trust between them. So RandomJoe can sell Yen-guy coins for stuff and buy USD-coins then USD guy will give actual cash to him.
Yes, I think that's how it could go down. Or Yen-guy and USD-guy could exchange currency directly, or they could do it hawala-style indirectly (for example: Yen-guy could hand out yen to anyone wandering into his Tokyo shop with USD-guy coins, and then USD-guy, to settle his bills with Yen-guy, can fly in and visit Yen-guy with a case of wine which Yen-guy can drink or sell).
I wonder if a form of coin-locking could be used to create simulated versions of other currencies. Depending on the demand and level of trust in the given currency, a given bitcoin sum could be locked in such a way that it gets automatically distributed to all holders at some future point (if they don't choose to hold onto the simulated currency instead) in proportion to how much they have. The choice to hold or drop the simulated currency would be a way of measuring its value against the bitcoin.
I can think of various ways to facilitate this using the techniques I am considering (or considered and discarded) for the creation of a distributed prediction market. The details of the precise implementation would depend on precisely what your goal was. I can't fully determine what you are aiming at from what you have said so far? More detail? A use case?
What I have in mind is something a little bit like the zerocoin proposal, in the respect that it's an alternate coin which can be translated to Bitcoin and back as part of how the system works.
But instead of being fixed with respect to the bitcoin it would increase or decrease in relative value in response to market signals. There would be an exchange rate determined by dividing the number of spendable units of alternate currency against the number of bitcoins locked behind it.
So what you need is a series of rules for controlling that number that tend towards economic stability. Which is itself kind of tricky. I had thought locking the coins by a time-lock could do it, but I haven't thought of a way to base an exchange rate on that. Empirical work with simulator or a smart math person with a whiteboard might turn up something.
The following uses incentives to accomplish something similar-in-principle:
Alice creates 100 dollarcoins at a cost of 1 bitcoin. She transfers 1 to Bob, he can turn it into 0.01 bitcoin.
The dollarcoin has a specification that includes things like interest payments. If Bob waits 30 days after receiving a dollarcoin he gets 1% of its nominal bitcoin value in interest. He gets another 1% payment 30 days later, and so forth. But if he spends it right away or converts it to bitcoin there are no payments.
Claire decides to use the dollarcoin specification to mint some dollarcoins. She has to pay a 1% minting fee, so 1.01 bitcoins to make 100 dollarcoins. The backing rate becomes 200 dollarcoins to 2.01 bitcoins.
Thus the more dollarcoins get minted the more valuable/highly backed they are, initially. If people spend them quickly, or convert them to bitcoin quickly without collecting interest, they continue to grow in bitcoin value. On the other hand if people hold onto them over time, they become less valuable as the savers get paid for keeping them.
Criticisms: It's complicated, and the numbers for interest rates and minting fees are pulled out of thin air. Presumably there's some range that works well, and could be tested by implementing multiple currency specifications side by side.