[hashcash] Re: how many bits / how to negotiate (Re: Re: stampcreation std. deviation)
- From: "Eric S. Johansson" <esj@xxxxxxxxxx>
- To: hashcash@xxxxxxxxxxxxx, CAMRAM <camram-spam@xxxxxxxxxx>
- Date: Fri, 13 Aug 2004 16:21:01 -0400
John Honan wrote:
> As a reference point; http://www.hashcash.org/papers/proof-work.pdf
(page
5) Calculates spammer profitability breakeven point at about 50 seconds
per email.
Microsoft CSRI paper at:
http://www.microsoft.com/mscorp/twc/privacy/spam_csri.mspx uses a slightly
different method, but calculates the breakeven point at 60 seconds.
I'm a bit too overloaded right now to actually go read the papers. But
I promise I'll do so in the next few days and give you my analysis. One
suggestion however if they start measuring the cost of hardware, they
are heading off in the weeds. In this world, hardware is effectively
free with respect to all the other costs of the system. It's important
to measure but it's vanishingly small in impact. now, for further grist
for the mill, I was using a model based on response rate.
The best analogy to the spam marketplace is direct mail campaigns. Any
direct-mail campaign measures success by response rate. The more money
you want to make means the more pieces you have to send because response
rate is going to remain the same (i.e. 1-5 percent). Conversely,
anything that reduces your volume, cut into your revenue in a one-to-one
relationship. For example, if you were to cut the volume of a mailing
in half, the revenue would be dropped by one-half.
The same relationship holds with spam. The response rate however is
something like .01% but the honest truth is nobody really knows except
the spammers and they aren't talking so any numbers people use to
populate the model are most likely made up estimates[1].
hashcash stamps primary effect is that of a rate limiter. They control
how fast you can send something given a measurable quantity of CPU
resources. So if you take the spammers send rate and drop it to
something 50 times slower, then you cut their revenue stream by the same
amount. The slower you make the stamp, the bigger the drop in the
revenue stream. This is why I'm shooting for the 15 second stamp which
works out to something like 700-800 times reduction in volume for a
spammer which means they have 700-800 times less revenue. If they can
survive on that kind of reduction of revenue, and they have a much
better business that any of us have ever imagined.
that's the best model I can come up with for the impact on spammers. It
fits basic economics of an equivalent system (i.e. direct-mail) and
that's unfortunately the best we can do today with the knowledge we have
hand.
If folks can poke holes in this model, I would more than well, it
because it will help build a stronger model.
---eric
[1] And if you don't think numbers can take on a life of their own, look
at the unverified story on estimates for worldwide volumes of
money-laundering. Originally, it started out as 300 million back in the
1980's but has been amplified by politicians until now it's something
over 300 billion. The reality is, they have no idea because they cannot
measure the real money-laundering volume. (unverified story courtesy
black unicorn)
--
Speech recognition in use. It makes mistakes, I correct most
Other related posts:
