[hashcash] how many bits / how to negotiate (Re: Re: stamp creation std. deviation)
- From: Adam Back <adam@xxxxxxxxxxxxxxx>
- To: hashcash@xxxxxxxxxxxxx
- Date: Fri, 13 Aug 2004 09:51:24 -0400
Jonathan's fast minter does ~4 Megahashes/sec or about 22 bits/sec on
a 3Ghz P4. But I concur: I think 20 bits is about right now also.
The question of how to upgrade this over time is an unanswered question.
CAMRAM does a challenge-response thing and so can negotiate upwards.
Bare-hashcash can not do this.
This is partly what the extension mechanism is for -- to allow us to
plug in a bit upgrading extension later.
I had one candidate which is some group agrees a policy (eg 1 second
on best price point desktop revised once every 6 months) and publishes
via a symmetric key "signature" like construct. (Which is eg
authenticated using a hash-chain.) Or using signatures.
But I wanted to progress the basic client to v1 format and get that
into SA3 before we explore this to work incrementally.
So the hashcash recommended from the hashcash man page is currently 20
bits. (Run it without the -b <bits> option and you get a 20 bit stamp).
I came across one security-related mailing-list that is attaching
hashcash to their security bulletin mails to subscribers.
http://www.computercops.biz There was a thread in news.admin.net-abuse
http://groups.google.ch/groups?hl=en&lr=&ie=UTF-8&threadm=41184f25%243%24fuzhry%2Btra%24mr2ice%40news.patriot.net&rnum=10&prev=/groups%3Fq%3Dhashcash%26hl%3Den%26lr%3D%26ie%3DUTF-8%26scoring%3Dd%26selm%3D41184f25%25243%2524fuzhry%252Btra%2524mr2ice%2540news.patriot.net%26rnum%3D10
where computercops.biz were trying to get their IP off spews. (Their
IP blocked because of arguably spammer-friendly/neutral ISP they are
using).
I am not quite sure how to read the whole thing, but they do appear to
send hashcash as I tried to subscribe to see and the subscription conf
msg had a valid 24-bit (version 0) hashcash stamp on it. (Of course
we don't know if that was meant to be a 20-bit but they got lucky --
an example of the info loss in not having the intended bits field).
But generally what it says in the FAQ about mailing lists is the
sender should put the stamp, and the recipient should accept the list
address as an address they accept mail as. (Long term solution being
list server verifying the stamp, and replacing with a signature).
Adam
On Fri, Aug 13, 2004 at 09:23:59AM -0400, Jean-Luc Cooke wrote:
> This policy I'd personally like to see it as "hashcash recomands X bits, but
> it is up to the mail-server/procmail user to decide how to filter mail".
>
> Reasons:
> 1) Mailing lists like linux-kernel may find ti too expensive to compute
> hashcash values for 3,000 emails 400 times a day.
> 2) lots of others
Other related posts:
