[oleganza]

Except you can lock your bitcoins in 5-of-9 multisignature transaction among 9 of your friends in San Francisco, London, Paris and Moscow with an agreement to not sign the transaction unless you are safe and can talk directly to them.

[reinhardt]

Because obviously everyone considering bitcoins has a dozen cryptogeek friends spread out in half a dozen countries.

[patrickaljord]

It's easy to get friends this way with irc and the internet you know. Besides, even if they are only spread around the country or a couple of countries it already makes things way harder. Regardless, passwords, even regular banking passwords are not meant to protect you against people that are willing to abduct you and torture you so I've always found this xkcd kind of unfair. Most passwords are meant to protect you from regular hacking attacks or internet thieves, not your local crazy mafia gang.

[kaoD]

Too bad the Bitcoin protocol is hard-wired to accept at most m-of-3 transactions. More than 3 signatures is considered non-standard and rejected by peers/miners.

[sirsar]

The key, however, can be split an unlimited number of ways using Shamir's Secret Sharing Scheme or a similar protocol.

http://en.wikipedia.org/wiki/Shamir's_Secret_Sharing

[kaoD]

Nice! It isn't quite the same since the spending party will know the private key forever, while m-of-n is a per-transaction signature. Still very useful.

[vbuterin]

You can push directly to Eligius if you want anything up to k-of-20; http://github.com/vbuterin/pybitcointools lets you do that with the eligius_pushtx command.

[oleganza]

Bitcoin is not hard-wired to limit N to 3. Default client discourages "non-standard" transactions by not relaying or mining them, but if you mine them yourself or have someone to mine them, they will be valid and accepted. For all practical purposes, non-standard transactions are just taking longer to be included in the chain and typically required to have a non-zero fee (while the regular payments can be often mined for free).

[qbrass]

Those 9 friends are most likely other bitcoin filled pinatas. So the expense of tracking them down will lead to more paydays.

[oleganza]

Those 9 friends are:

1. In very different locations. Running around the world is going to be quite expensive.

2. Using different security measures. If they find one friend for his mistakes in maintaining privacy or security, same trick isn't going to work with some others.

3. Friends will also lock their stash in X-of-Y transactions with some other people, so finding them won't immediately increase potential gain.

I'd say the cost of running after individuals grows quadratically while the potential revenue only linearly.

4. Not all your funds will be locked with the same 9 people. 10% will be, while other 10% will be locked with some other group. To get 100% of the stash you'd effectively have to kidnap and torture maybe 20 different persons in different countries all over the world.

5. People learn. Once 2 or 3 folks are captured this way, all the rest will reshuffle their funds elsewhere and use better security measures. So you'd have to catch all at once, otherwise money will always leak right through your fingers.

Bitcoin is really, really a leapfrog technology with security incomparable with anything you had before. Previously known 2/3-factor schemes (including Shamir's Secret Sharing Scheme - SSSS) always required a single non-compromised machine to bring all secrets together. Bitcoin n-of-m transaction can be securely signed by N potentially compromised machines provided they are not all owned by the same operator.

See also global economical implications of the Bitcoin security: http://blog.oleganza.com/post/67872772342/bitcoin-and-gold