Encryption and risk

[markw-FJ05HQ0HCKaWd6l5hS35sQ at public.gmane.org]

> On Tue, Oct 6, 2009 at 9:43 AM,  <markw-FJ05HQ0HCKaWd6l5hS35sQ at public.gmane.org> wrote:
>> "decades" is an interesting measure.
>
> and what is necessary if protecting State secrets.
>
> And may be what is needed for  digital signature in commerce, if the
> contract might be subject to verification that far in advance, even if
> not for basic secrecy.

What I mean by my comment is that "decades" of processing is a sliding
scale. I haven't worked it out, but assuming moore's law is a legitimate
constant (it really isn't but I digress)

I have CPU capacity of X, in 10 years of processing I have executed 10X of
CPU years. However, 1.5 years later, I have 2X. In another 1.5 years I
have 4X. In 1.5 more I'll have 8X, so on and so on.

So if I start process Y on January 1, 2000 it will complete D units by
December 31, 2010. One "decade" worth of work.

I could start the same program in July 1, 2001 and have that same amount
of work done by July 1 2006. (5 years)

I could start the same program in January 1, 2003 and have the same amount
of work done by July 1, 2005. (2.5 yeas)

If I started it on July 1, 2004, I could be done by October 1, 2005. (1.25
years)

It makes no sense to think of processing that takes that long because it
almost doesn't even make sense start such a project. A "decade" of
processing power is not a rational measure. The technology changes too
quickly.

>
>
> --
> Bill
> n1vux-WYrOkVUspZo at public.gmane.org bill.n1vux-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
>