I have issued a few more POLY.10.1 bonds. The bot will now put bonds up to sale (at 107% face value) as long as the total outstanding bonds have face value less than 200 BTC. This is subject to change.

That's why Bayesian statistics is superior to frequentist statistics. All the data gives us is about 10:1 likelihood ratio for 1.1D over 1D, which doesn't mean much when the prior probability of 1.1D is so low.

That makes more sense. But even then, not much evidence.

I didn't read the entire analysis but I disagree with your conclusion that round lengths are inherently longer, there is hardly enough evidence to rule out bad luck. Especially if there is no plausible explanation how can round length mean be higher.

I've been thinking for a while about offering a deterministic negative mining bond, but the designs I've come up with so far imply a specific value for the positive bonds, which isn't very elegant.

There's a very simple way to offer people to short a mining bond - write a put option for the bond, with the strike price decaying with the coupons. Maybe I'll do that...

Mpex also have the interesting offers of put and call options for a given number of hashes, rather than hashrate. That can be used to bet that the difficulty will increase, though not specifically that the market value of bonds will decrease.

I'm assuming Brian is one of the people who preordered. Naturally people who have preordered (including myself) will know a bit more about the process, such as the fact that a concrete request for a deposit hasn't come yet.

I think the quoted number of 31 preorders is very believable. As we said, nobody is complaining about funds held up in escrow because no funds were sent to escrow.

1. BFL has an incentive to exaggerate the announced specs, to discourage potential competitors and to encourage customers to buy current-gen hardware which can be traded-in. While BFL will certainly be capable of manufacturing the devices at the stated performance and marginal cost if and when they finish development (which could be in a long time), it will be a dubious business decision to offer at those prices. Sound pricing would be beating the competition decisively but no more than that.
2. LargeCoin has stated they're using sASIC which is the "least custom" ASIC. From BFL's words it seems they've put in more money and developing a more advanced type of ASIC. Also, their release time will be later in the future so the may be aiming for better process technologies.

It's not easy to to figure out which addresses belong to a given person. Case in point, I don't think it was ever done.

That's not true. The main problem isn't reusing addresses, it's merging outputs.

Block finding times are random.

The probability that within an interval of 10 minutes at least 7 block will be found is about 1/12000. Which means that it's very likely this has happened many times before.

Cryptography is mathematics.

Even if "cryptography" is assumed to mean "encryption", the public should be well aware of the important role of encryption in online commerce and banking.

It is when in the beginning of a sentence.

I agree of course but that's not really evidence, ThePiachu's master's thesis was in Bitcoin and he still didn't properly capitalize "bitcoins".

I wonder if "bitcoinology" should be capitalized?

Scientific fields aren't capitalized.

Encryption is a one-to-one transformation for which the inverse can be computed by, and only by, someone who has the proper key. It is used to conceal data from everyone but intended recipients. That isn't used anywhere in the Bitcoin protocol, but it is used in wallet encryption in the client.

Hashing is a transformation that is in general not one-to-one, and which ideally cannot be inverted by anyone. The ostensibly random nature of such transformations is used in making block finding artificially difficult (and directly proportional to computational power).

"Cryptography" isn't capitalized when in the middle of a sentence.

"Cryptography" is a general field encompassing encryption, digital signatures, hashing and more.

Wonderful. While academically it seems the intended focus was security analysis, it is comprehensive enough to be used as a reference general introduction to Bitcoin, to those willing to take the time to study it seriously.

A few nitpicks:

1. "Bitcoin" and "satoshi", as units of currency, should only be capitalized if at the beginning of a sentence. "Dollar" isn't capitalized either. Bitcoin as the name of a project, a protocol, a software, an ecosystem and a community is of course capitalized. Also, I'd avoid using "bitcoins" in reference to the system - e.g., instead of "(b)Bitcoins offer much lower cost of transferring money" I'd say "Bitcoin offers".

2. You use http://pastehtml.com/view/awb1vg03r.html as a reference for currently used mining pool reward systems, but it's badly out of date, and even for its time it was vague at best - SMPPS pools are listed as PPS, different varieties of score-based methods are lumped together in a single category, etc.

I agree but personally I don't feel the upside is worth the cost of less direct entanglement with the underlying hardware purchase. But I certainly don't deny that MOORE has a place catering to people for whom it is. In particular, it expands the space spanned by the available investment opportunities, giving sophisticated investors more flexibility in matching their portfolio to their future projections.

Regarding the psychological effect, it's mostly a product of what metric one is looking it. Currently GLBSE shows a graph of an asset price over time, and seeing a graph which is always in decline indeed looks bad. IMO the relevant metric is current price + coupons paid per unit so far. This gives a more accurate, intuitive picture of the viability of an asset, should look the same for functionally equivalent assets, and will show an upward trend for any successful investment.

PureMining is a deterministic bond, and it will continue to be 1 MH/s per bond. I don't increase it when something good happens any more than I decrease it when something bad happens, the contract is binding both ways.

Some issuers have offered to pass on gains from the BFL trade-in program. This makes sense given this disruptive event, I wish them the best of luck and investors who are confident in BFL's ability to deliver would do well to consider these offers. But that's not the market segment I'm in, I just want PureMining to continue faithfully filling its role of a deterministic bond long after the uncertainty of the current time is behind us.

BFL's announced specs are very impressive, but on the other hand they don't have a great track record of delivering what they promised on time. I will certainly take their announcement into consideration when deciding on the pricing of any future issued bonds, and investors should take it into account when deciding how much they are willing to pay.

I think it's very interesting but ultimately I don't see the point. The issuer of a bond with a hardcoded appreciation in hashrate is speculating on what the future cost of hashrate will be, and he will have to compensate for this risk in his pricing. In a fixed-hashrate bond the issuer can much better estimate his cost of covering the bond, and can thus offer it at a more affordable price. Also, since the chosen appreciation rate is arbitrary, it stands in the way of standardization and commoditization.

I prefer the scenario where the issuer just offers more bonds if he wishes to expand utilizing new hardware opportunities, and investors can buy them if they wish to maintain or increase the value of their investment.

The miner will scan the whole range looking for a valid hash. Only one time in ~1.6 million this scan will result in finding a valid hash. If it doesn't, the miner will tweak the header, typically by increasing the timestamp or the extraNonce.

Bonds such as MOORE which have hardcoded hashrate increase are interesting but they're basically speculation on the part of the issuer, as he can't know what will be the actual decrease in the cost of hashrate (and any chosen value will be arbitrary and nonstandard). With fixed-hashrate bonds, the issuer knows how much it will cost him to provide a given hashrate, and the reduced risk allows him to offer the bonds at an affordable price. If he wishes to expand he can always issue more bonds, and if the investor wishes to maintain/increase the value of his investment he can always buy them.


I've also commented here on what a better buyback provision could look like.

Buyback based on initial price doesn't work. An issuer who has been paying coupons for 2 years and needs to buyback, shouldn't be paying what he originally got now that the value of a MH/s is only 40% of the original value.

A buyback shouldn't be a free pass to stop paying coupons. It should be painful to the issuer, and profitable to investors, only to be used in a few specific circumstances such as:

1. The issuer is legally forced to cease being involved in the bond.
2. The issuer is disabled/dies and his heir doesn't want the hassle of operating hardware and paying coupons.
3. The issuer's hardware suffers catastrophic failure and he wishes to cut his losses.

I think 120% of 30-day maximum price, as in PureMining, is fair. It would require the bond value to take a sustained, massive hit in order for the buyback price to be less than the original price. And if there is such a hit, it would imply that the bond isn't as profitable as the investor initially thought, which means he would take a loss buyback or no buyback.

Another useful metric for deciding buyback price is what I call "projected lifetime earnings". This is what the bond would produce forever if the difficulty remains constant (but block reward halves). This is easy to calculate - (number of blocks until next halving + 210000) * (current reward per 10 minutes), and currently it's 1.0375 BTC. This value, combined with a projection of future difficulty changes, provides a measure of the actual lifetime earnings of the bond. If the bond is currently traded at, say, 0.4 BTC, it would mean the market thinks the value of the bond is about 40% of the projected lifetime earnings - and so a fair buyback price would be 50% of the projected lifetime earnings. This allows the buyback price to decay together with the bond, without the investors being at risk of an artificial decrease.

The optimal buyback would be something like "110% of 30-day maximum price, but never less than 40% of projected lifetime earnings and never more than 60% of projected lifetime earnings". This guarantees both the investors and the issuer are safe, but is fairly complicated.

This can be very powerful if it can be pulled off. But it seems to me that the network effect barrier will be extremely challenging.

The bot is fairly simple for now (mostly just upholding my contractual duty to buy bonds at face value), didn't take long to develop, just had to wait for me to come back home from miluim.

Yes. That's because they're all sold out . If someone sells to it it will put (some of) them back up for sale.

It's buying too, but to do that somebody has to sell .