Funny address pair

[amaclin]

Look at these addresses
https://blockchain.info/address/1LKR8M54CKzZ2pnkmgLmQVzgMTP4md2gMw
https://blockchain.info/address/3M1S3tZVkEJw7zVBtn1Mq8McVyfnNMAuoX

These two addresses have the same hash d3e604621abfc263162af107834b5a04011b9751
And both addresses are spendable
The first one is spendable by private key owner (not me), the second - by everyone (thanks to Eligius)

The public key of 1LKR8M54CKzZ2pnkmgLmQVzgMTP4md2gMw is 03641eee1b2919d6a7a0dd5ea30dbfee0e687948e57e70b347eb8c32feae01d96d
and it is valid bitcoin script acceptable by bip-16

Code:

03 641eee                                                  // push "641eee"
1b 2919d6a7a0dd5ea30dbfee0e687948e57e70b347eb8c32feae01d9  // push "2919...01d9"
6d                                                         // OP_2DROP 

[micaman]

The 2nd one is a multisig address. You probably need more than 1 private key to spend those funds.

[amaclin]

The 2nd one is a multisig address. You probably need more than 1 private key to spend those funds.

No. It is p2sh address but not a msig address. I took these funds without knowledge of private key.

[TimS]

Cool! A great deal of these address pairs exist. You could search for them like you search for vanity addresses quite easily - the given example only really requires 2 bytes in the public key fixed (1b and 6d), since you can compress any public key to an 03 pub key.
It's confusing to the likes of blockchain.info (it says 3M1... has a negative balance), but they're both perfectly valid and separate addresses (although, being anyone-spend, you don't want to use the second for real stuff).

[amaclin]

Cool! A great deal of these address pairs exist.

The second address is useless, because everyone can transfer funds to himself.

You could search for them like you search for vanity addresses quite easily - the given example only really requires 2 bytes in the public key fixed (1b and 6d), since you can compress any public key to an 03 pub key.

Public keys can start with 02, 03 and 04. And there are a lot of different valid bip-16 scripts - we can use a big set of opcodes - OP_DUP, OP_DROP, OP_NOPx... Several forms of OP_PUSH...

It's confusing to the likes of blockchain.info (it says 3M1... has a negative balance),

It is a bug of bc.i

but they're both perfectly valid and separate addresses (although, being anyone-spend, you don't want to use the second for real stuff).

It is not my address (I do not know private key), but I like this idea... And thanks for some btc amount which i transfered to my account

[Remember remember the 5th of November]

I am sorry, I am not quite getting this. RIPEMD160 hash is 160 bits. What's the probability of two addresses having the same RIPEMD160 hash?

[TimS]

I am sorry, I am not quite getting this. RIPEMD160 hash is 160 bits. What's the probability of two addresses having the same RIPEMD160 hash?

Randomly? 1 in 2^160. But you can choose...

An address contains three parts: a version number, the 160 bit hash, and a checksum. Version 00 is used for Bitcoin pubkey hash addresses, and makes the address start with a 1. Version 05 is used for Bitcoin script hashes (typically multi-sig), and makes a prefix of 3.

You can use the same 160 bits to make more than one address by changing the version number. As done here where one is a normal pubkey hash, the other is a P2SH address where the "script" is the pubkey (interpreted as a Bitcoin Script). If this pubkey-as-script allows you to spend the output, then both addresses can be spent from by people with the right information.

[gmaxwell]

These two addresses have the same hash d3e604621abfc263162af107834b5a04011b9751

They have entirely different scriptPubKeys, however. The sameness is only even remotely interesting because of bugs in bc.i (and perhaps some other wallets that just ignore the version bytes).

You might want to attribute where you saw this discussion…

[amaclin]

You might want to attribute where you saw this discussion…

Sorry, I am not sure that I understand you. What discussion should I attribute?

[rapport]

You might want to attribute where you saw this discussion…

Sorry, I am not sure that I understand you. What discussion should I attribute?

Maybe, how did you find this occurrence of same hash?

[amaclin]

Maybe, how did you find this occurrence of same hash?

My script monitors for non-standard transactions and notifies me about "something new&interesting".

There were such pairs in blockchain in the past, but I have never seen both addresses of pair spendable
For example:
https://blockchain.info/address/3J98t1WpEZ73CNmQviecrnyiWrnqRhWNLy - spendable p2sh address
https://blockchain.info/address/1HT7xU2Ngenf7D4yocz2SAcnNLW7rK8d4E - the private key is known to nobody[?]

[gmaxwell]

Maybe, how did you find this occurrence of same hash?

My script monitors for non-standard transactions and notifies me about "something new&interesting".

Ah, good spotting then. It was a product of this discussion today: http://download.wpsoftware.net/bitcoin/wizards/2014-07-20.html  start at 00:56:13

I made an erroneous assumption that you'd been watching the logs and made the post as a result.

[amaclin]

I made an erroneous assumption that you'd been watching the logs and made the post as a result.

I assume that you are watching the blockchain

Code:

petertodd:	lol, 3PhyP1FWB8KgwrNgUPbjP7iv6YfoH3GhK7 is kinda hilarious....
02:47:50	gmaxwell:	what this?
02:48:41	petertodd:	gmaxwell: e.g. 70c8405bd0ec10bea49b78a819dfbf46c1082e7e620588f9da65a90b71e52bbd which spends an output of it...

and discussing my transactions  

Code:

02:32:37	gmaxwell:	kanzure: looks like someone has implemented p2sh by sending as pay-to-pubkeyhash. e.g. just ignoring the version byte.

posted by me on july, 02:
https://bitcointalk.org/index.php?topic=673737.msg7633818#msg7633818

[rapport]

Maybe, how did you find this occurrence of same hash?

My script monitors for non-standard transactions and notifies me about "something new&interesting".

Thought you would have a script.  Is it open source?

[rapport]

Maybe, how did you find this occurrence of same hash?

My script monitors for non-standard transactions and notifies me about "something new&interesting".

Ah, good spotting then. It was a product of this discussion today: http://download.wpsoftware.net/bitcoin/wizards/2014-07-20.html  start at 00:56:13

I made an erroneous assumption that you'd been watching the logs and made the post as a result.

If you only discussed today, and OP made the post a few days ago, how could he have seen the logs back then?
Or was there a related earlier discussion too?   Thanks for the reference on today's discussion.

[amaclin]

If you only discussed today, and OP made the post a few days ago, how could he have seen the logs back then?
Or was there a related earlier discussion too?   Thanks for the reference on today's discussion.

Of course I located these things after the transactions were sent and confirmed into the blockchain.
So, I was several hours behind of wizards and their discussion

Thought you would have a script.  Is it open source?

It is not a script but a set of different methods from dumping all addresses to sniffing bitcoin traffic.
No, it is not opensource, but I would like to share my knowledge with everyone.
Just excuse me, my English is bad.

[rapport]

No, it is not opensource, but I would like to share my knowledge with everyone.
Just excuse me, my English is bad.

Thanks for sharing.
I understand now what is happening.

I think instead of "These two addresses have the same hash d3e604621abfc263162af107834b5a04011b9751" this would be clearer:

"These two addresses originate from the same hash d3e604621abfc263162af107834b5a04011b9751"


The first makes it sound like hash(addr1) == hash(addr2) == d3e604621abfc263162af107834b5a04011b9751
Whereas this discussion is about https://en.bitcoin.it/wiki/File:PubKeyToAddr.png  The hash you are talking about is the 20 bytes, and since different prefix byte can be used, this means the same hash can create different addresses.

edit: formatting

[amaclin]

Sorry for misunderstanding. English is not my native language. I prefer Java and C++ 

[rapport]

No problem.  Everyone else seemed to understand what you meant