Re: [bitcoin-dev] Codex32

[Andrew Poelstra <apoelstra@wpsoftware.net>]

[-- Attachment #1: Type: text/plain, Size: 3397 bytes --]

On Thu, Feb 16, 2023 at 12:50:12PM +0100, Pavol Rusnak via bitcoin-dev wrote:
> Hi!
> 
> The BIP states that its only advantage over SLIP-0039, which has been used
> in production for nearly three years (in at at least 3 SW/HW wallet
> implementations), is that it aims to be simple enough for hand computation.
> However, the BIP also indicates that "details of hand computation are
> outside the scope of this standard, and implementers do not need to be
> concerned with this possibility." Therefore, I am curious about how
> significant this advantage over SLIP-0039 really is. If hand computation is
> not straightforward and there are no other substantial advantages over
> SLIP-0039, I cannot help but feel that this BIP is simply a result of
> not-invented-here syndrome, but please correct me if I am wrong.
>

In my view, the hand computation is actually the main benefit of this
scheme. The process *is* straightforward, but tedious enough (and the
security benefits obscure enough, though they really shouldn't be...
"computers are opaque and untrustworthy" should be a common sentiment)
that it's hard to expect more than a small absolute number of users to
actually do it.

But for the purpose of the *standard*, what is important is that it is
possible to implement and use this within a normal hww workflow. This is
important for hand-computing users who know that their coins will not
die with them (since the 'standard' has fallen into obscurity), and
important for "normal" users who have the option to seamlessly switch
over to hand computation as the BTC price goes up or the world becomes
scarier.

For what it's worth, the draft lists several benefits over SLIP-0039.
I agree that none of them are particularly strong [1], and even together
they probably wouldn't meet the threshold to take the time to write a
standard, but I assure you the motivation was not NIH :).

> Keep in mind that the encoded shares in SLIP-0039 consist of exactly 200 or
> 330 bits, both of which are divisible by 5. This makes it straightforward
> to encode them as Bech32 strings.
> 

This is true! And very convenient for people who may want to simply
"layer on" the codex32 checksum/splitting logic onto their SLIP39 words.
They can use a lookup table to do the conversion, spend years or
whataever doing hand-computation on them, and then use a lookup table
to go back.


[1] One listed reason is that "a SLIP is not a BIP". I have heard people
    speculate that this is one reason SLIP-0039 is not nearly as
    widespread as BIP-0039, even though it is objectively a far better
    standard. I'm unsure whether I believe this, but "there is no other
    BIP" does seem like a good reason for BIP-0039's continued
    dominance.

    At the very least, it means that on BIP-0039 itself we have nothing
    that we could say "supercedes" or "is recommended instead of" the
    BIP. See https://github.com/bitcoin/bips/pull/1413

    So it's something of an aside, but I think it would probably be good
    for the ecosystem (though maybe bad for this BIP's prospects :)) if
    you would request a BIP number for SLIP-0039.


-- 
Andrew Poelstra
Director of Research, Blockstream
Email: apoelstra at wpsoftware.net
Web:   https://www.wpsoftware.net/andrew

The sun is always shining in space
    -Justin Lewis-Webster


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]