---

[Financisto]

@unchi

Suggestion: show the generated private key also in compressed format.

You can get the code from bitaddress.org

[1714891098]

1714891098

[1714891098]

1714891098

[1714891098]

1714891098

[RGBKey]

Brainwallets are just another form of storing your bitcoin. Humans are known to be notoriously bad randomness generators, but if you can find a good random phrase that you can remember, that hopefully you didn't come up with yourself, that's not out of a book or anything, it can be a novelty way to keep your coins without needing to save a file with your private keys.

They can confiscate your computer but they can't confiscate your thoughts.

[lorylore]

This is an update, thank you unchi, you are very kind. It's really hard to remember 12 different words of this kind.
What about to release an update for print a good design of the printed page?

That's a brilliant idea, and I will definitely implement it.  Thank you.

I like that you like my suggestion, finger crossed for the next update. The activity of visitors will increase definitely. 

[prodigy8]

This is an update, thank you unchi, you are very kind. It's really hard to remember 12 different words of this kind.
What about to release an update for print a good design of the printed page?

That's a brilliant idea, and I will definitely implement it.  Thank you.

I like that you like my suggestion, finger crossed for the next update. The activity of visitors will increase definitely. 

UPDATE

Print button has been added!  It just prints out your keys and QR codes.  Plain and simple. 

The difference with bitaddress/org is that here it uses less printer ink. Keep it up

[lorylore]

This is an update, thank you unchi, you are very kind. It's really hard to remember 12 different words of this kind.
What about to release an update for print a good design of the printed page?

That's a brilliant idea, and I will definitely implement it.  Thank you.

I like that you like my suggestion, finger crossed for the next update. The activity of visitors will increase definitely.  

UPDATE

Print button has been added!  It just prints out your keys and QR codes.  Plain and simple.  

Awesome, one more suggestion i was thinking something like this:

[guitarplinker]

Awesome, one more suggestion i was thinking something like this:
snip

He could probably add an option for a fancier image like that. I like how it is now, because like prodigy8 said, it uses very little ink when being printed compared to your suggested design.

[lorylore]

Awesome, one more suggestion i was thinking something like this:
snip

He could probably add an option for a fancier image like that. I like how it is now, because like prodigy8 said, it uses very little ink when being printed compared to your suggested design.

Ok then I think adding 2 buttons: 1st let be as it is now "Simple print" with a description that it uses little ink and the 2nd button with a design which use lots of ink inc. color. The design should be good even if the printer is black and white. This is just an idea so the OP can implement it.

[prodigy8]

UPDATE

You should see some significant performance improvements now.  I am using a streamlined scrypt implementation that has a smaller memory footprint, resulting in faster wallet generation and better support for older hardware.

I can see the difference it is much faster than it was before. And try to keep the website simple and let us use less printer ink

[ryanc]

If you're going to print it, just use BIP38 paper wallets. If you're worried about it getting lost, make multiple copies.

[lorylore]

Awesome, one more suggestion i was thinking something like this:
snip

He could probably add an option for a fancier image like that. I like how it is now, because like prodigy8 said, it uses very little ink when being printed compared to your suggested design.

Ok then I think adding 2 buttons: 1st let be as it is now "Simple print" with a description that it uses little ink and the 2nd button with a design which use lots of ink inc. color. The design should be good even if the printer is black and white. This is just an idea so the OP can implement it.

Thanks for the suggestions. I will look into making the print page a little prettier, but I think I'd like to do it through CSS and avoid using images.

I think a great design something unique only for bitcoin addresses, yes avoiding images is good too, it can be sometimes not printed in the right way. I wish you good luck.

[ryanc]

The documentation on the website as to what algorithm brainwallet.io uses is inaccurate.

It says:

key = scrypt(passphrase, salt, N=2¹⁸, r=8, p=1, dkLen=32)
keypair = generate_bitcoin_keypair(sha256(key))

It's actually:

key = hex(scrypt(passphrase, salt, N=2¹⁸, r=8, p=1, dkLen=32))
keypair = generate_bitcoin_keypair(sha256(key))

Why are you using uncompressed keys?

[Financisto]

@unchi

Suggestion: show the generated private key also in compressed format.

You can get the code from bitaddress.org

This would be easy to implement, but I'm not sure if there would be enough demand for it.  I prefer to keep the website as simple as possible to avoid confusion.  What's the general consensus on this?

I guess it would be more like an add-on to your project.

Not an issue at all...

Some discussions about that: https://bitcointalk.org/index.php?topic=129652.0

[ryanc]

Why are you using uncompressed keys?

For compatibility.  

With what? It seems like pretty much all tools have been supporting compressed keys for quite some time.

[Tstar]

I seriously never pid attention to brainwallet and never knew what it is ..
but why would someone use the wallet, any significance ?

[ryanc]

For those of you complaining to people linking to my slides/blog posts about brainwallets - I'm currently testing support for brainwallet.io in brainflayer. My limited benchmarking gives an estimate of about 75k 750k passphrases guessed per dollar on Amazon EC2 spot instances.

Edit: I am bad at math.

[lorylore]

For those of you complaining to people linking to my slides/blog posts about brainwallets - I'm currently testing support for brainwallet.io in brainflayer. My limited benchmarking gives an estimate of about 75k passphrases guessed per dollar on Amazon EC2 spot instances.

I have read all of your slides and i was surprised,
We are thrilled to know the results at the end, i think it will be harder than brainwallet.org since brainwallet.io use salt???

[ryanc]

For those of you complaining to people linking to my slides/blog posts about brainwallets - I'm currently testing support for brainwallet.io in brainflayer. My limited benchmarking gives an estimate of about 75k passphrases guessed per dollar on Amazon EC2 spot instances.

Wow that's better than I thought it would be. Thanks for taking the time to do that. For comparison I'm curious to know the benchmark results for brainwallet.org. Could you provide that as well?  

Never mind, I saw in your presentation that the answer is 560 million passphrases per $1.

I would rephrase your benchmark statistic for brainwallet.io to say "75k passphrase-salt combinations per $1". You would spend a lot more than $1 trying to crack one passphrase because you would have to go through every possible salt.

"Better" as in you expected it to be more or less expensive? Cracking benchmarks are typically understood to imply the numbers are for a single salt, if salts are used.

[Financisto]

For those of you complaining to people linking to my slides/blog posts about brainwallets - I'm currently testing support for brainwallet.io in brainflayer. My limited benchmarking gives an estimate of about 75k passphrases guessed per dollar on Amazon EC2 spot instances.

Congratulations for also researching and testing this tool.

That's a positive effort and all community should benefit from it.

[HostFat]

For those of you complaining to people linking to my slides/blog posts about brainwallets - I'm currently testing support for brainwallet.io in brainflayer. My limited benchmarking gives an estimate of about 75k passphrases guessed per dollar on Amazon EC2 spot instances.

It would be interesting to change these settings to see which are enough to get near 1 passphrase per 1 dollar (or even lower)
N=218, r=8, p=1, dkLen=32

[ryanc]

For those of you complaining to people linking to my slides/blog posts about brainwallets - I'm currently testing support for brainwallet.io in brainflayer. My limited benchmarking gives an estimate of about 75k passphrases guessed per dollar on Amazon EC2 spot instances.

It would be interesting to change these settings to see which are enough to get near 1 passphrase per 1 dollar (or even lower)
N=2¹⁸, r=8, p=1, dkLen=32

The issue is that a legitimate user has to spend the same amount of work as the cracker per passphrase, so there is a practical limit on how much work the KDF can do. With scrypt specifically, scrypt(N=2¹⁸, r=8, p=1) uses 256MiB of memory - scrypt(N=2²⁰, r=8, p=1) would be 1GiB. If you wanted scrypt to take longer than that, I'd probably suggest something like PBKDF2(iter=64, prf=scrypt(N=2²⁰, r=8, p=1)) which would take several minutes to run and is probably close to the upper bound of what anyone is willing to put up with.

You could also force some extra randomness into this by generating say, four hex digits as part of the salt and telling the user to write it down. Lose the digits and you have to brute force them - time consuming but possible. The cracker, though, doesn't have them and has to try them all in addition to whatever other salt there is...