More questions about Echolot (still trying to be reliable)

[Joel M. Baldwin]

--On Tuesday, September 02, 2003 9:16 AM +0200 Peter Palfrader 
<peter@palfrader.org> wrote:
>
> . . . snip . . .
>
> FYI, the echolot in CVS, which will be the next release, will handle
> this a little bit different.  Stats from 12 days ago don't count as
> much as yesterdays.
>
> day    |  0  |  -1 |  -2 |  -3 |  -4 |  -5 |  -6 |  -7 |  -8 |  -9 |
> -10 | -11 weight | 0.5 | 1.0 | 1.0 | 1.0 | 1.0 | 0.9 | 0.8 | 0.5 |
> 0.3 | 0.2 | 0.2 | 0.1

My only concern here is that if a ping was sent out less that a
reasonable period of time ago based on the ave latency of that
remailer then the weight would be 0.  I.E. don't count pings in
the stats if they were just sent out.  I would think that a
reasonable delay before counting the ping would be ( 1.5 * AveLatency ).

I'm tired of seeing 99.X% stats with a line of +'s and a single
8 or 9 at the end!

> I think this is better.  Comments?
>
> Peter
> --
> Cannot verify the sig?  Update 94C09C7F from subkeys.pgp.net.