[RC5] Stats suggestion

G & E Holt geholt at t-online.de
Sat Jan 6 10:00:13 EST 2001


I work with a few databases myself, and anytime you add the ability for user
options, it generally makes it more complicated.  I like the halflife idea,
so I believe it should be a fixed value(s) to keep things fairly simple.

The longer halflives are nice as they do not show the spikes, but who wants
to wait for 4 to 6 months for that halflife to get close to accurate.  I
know many have mentioned how ppl would be upset if their initial was set to
0, but right now I do not even pay attention to my Overall key rate as I
know it is lagging behind.

If there is someone on my list that I am trying to catch, I will go to their
stats and look at their keyrate for the previous day.  So basically the
Overall key rate at this point is just a number and has very little bearing
on anything.  Actually we are ranked by Overall blocks, but the halflife
Overall key rate would allow us (and our teams) to see what we have to do to
pull in front of person/team x.  It would also stop the megaflushing that
the major teams do just to be ahead of the other team for a day.

As for the initial starting value, start us at anything you want.  We would
have to wait, but it would be MUCH better in a month or two when we have
"close" to the correct Overall rate versus now.

GE


-----Original Message-----
From: owner-rc5 at lists.distributed.net
[mailto:owner-rc5 at lists.distributed.net]On Behalf Of Jeff Lacy
Sent: Saturday, January 06, 2001 6:07 AM
To: rc5 at lists.distributed.net
Subject: Re: [RC5] Stats suggestion


----- Original Message -----
From: "Arne Heizmann" <ajh65 at hermes.cam.ac.uk>
>
> > Now, what would be *really* nice is if people could choose their own
> > halflife (or halflives) for their own records
>
> I thought of that too, but it would encourage "cheaters" to put it on a
> very low value for one day, mega-flush, and then put it on an extremely
> high value. This would make the weighted average show an unrealistically
> high key rate for quite some time.

All right, if that would encourage cheaters, then maybe have two halflives.
One that is the same for everybody and one that is whatever the participant
selects.  That way, if someone has a really high average, then just don't
look at it and only look at the average that everyone uses because the high
one is probably inaccurate.  Another way might be to limit the range of
possible halflives, like say from 10-50.  What does anyone think about this?
Just my $0.02 worth :-)

Jeff

>
> --
> To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
> rc5-digest subscribers replace rc5 with rc5-digest
>

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list