On Wed, 3 Apr 2002, Adam Spiers wrote:
> > Here's my todo list:
> >
> > - proper numerical x axis, one tick per day
>
> Done. I was expecting this to get rid of the general linearity, but
> there's still no evidence of it slowing into a shallower curve.
> Scary! As Ton (I think) observed, we should reach zero-length
> solutions sometime next week.
Don't you think the curves should look more like steps?
After all, when you go from, say, 121 to 119, this is not a linear
progress. It's more like a jump. Or a fall.
So maybe the curves should look more like the following, don't you think?
-----+
-----|------------+
| |
120 - | +---- Ringo
-----|-----+
+-----|----------- John
110 - |
+----------- Paul
| | |
then later now
This means you'd probably have to use two scores for the same moment in
time. If the score goes from s0 to s1 at time t, the data to feed the
program should be : (0, s0), (t, s0), (t, s1) and (now, s1)...
A little perl filter might help... ;-)
--
Philippe BRUHAT - BooK
Financez le développement de Perl, avec YAS et les Mongueurs de Perl !
http://www.mongueurs.net/association/actions/pdg2002.html
