Whenever I have checked, my "runs" have followed the classical Poisson
distribution. Of course if the run is very slow or very fast, then this may
not be apparent, but in my case it is not unusual to go 1 or 2 minutes
without a call, and then get 3 or 4 in the next minute, possibly five or
more calling at once. We simply should not be surprised about this, it's
basic statistics. I had to study stats. in my Chemistry education, and
confess I mostly hated it. However, I have never forgotten the title of the
original Poisson paper, being, I believe " Goals, floods and horse kicks".
These were the events that Poisson studied, being the occurrence of goals in
football matches, natural events such as floods, and bizarrely the number of
reported deaths of cavalrymen who had been kicked by their horses! These he
believed were randomly occurring events ( with respect to time) in nature,
which could be predicted to follow his now classical distribution curve.
Perhaps we may now add callers in an HF contest!, although I am not
sufficiently expert to understand that this is scientifically correct.
Either way, our runs will always be "lumpy" to a more or less extent.
On the other had, when you have been spotted, you will know it, trust me;
the run is not just lumpy, but quite unmanageable for a short while. We all
know the cheats are out there, but proving it is quite a different matter,
even to the relatively non-rigorous standards we may expect. Comfort
yourself, these guys are only cheating themselves.
73
Clive
GW3NJW
On Oct 17, 2005, at 2:58 PM, Steve.Root@culligan.com wrote:
> I would conclude that I must
> have been spotted to account for this and make note of the time.
> After the
> contest it was easy enough to do a spot search on DX Summit to see
> how many
> times K0SR had been spotted. What I found was that I might get
> spotted once
> or twice over a whole contest. (Minnesota isn't exactly a rare
> multiplier
> and I'm not going to be loud from here). I see almost no correlation
> between my rate and any "spots". I've concluded that changes in
> propagation
> could easily account for the changes in my rate especially on a
> marginal
> band like 10 meters.
Actually, it's probably simple math.
It's been a long time since I dealt seriously with queuing theory,
but its certainly true that most Markovian (and many non-Markovian)
processes will tend to "clump" like this.
The best example is the simple Poisson process. The Poisson process
has one parameter - that in a given time interval, the probability of
an arrival is some constant. So, if your running stations at 60 q/hr,
then the probability that any station would call in a particular 6
second period is one in 10. So, you'd think that someone would call
every 60 seconds.
But that isn't what happens. If you model this process, you'll find
there are periods of several minutes where no one calls, and then a
minute where three guys call at once.
The actual process by which callers call you is probably a lot more
complex than the Poisson process, but it will show "clumping" much
the same.
Bill Coleman, AA4LR, PP-ASEL Mail: aa4lr@arrl.net
Quote: "Not within a thousand years will man ever fly!"
-- Wilbur Wright, 1901
_______________________________________________
CQ-Contest mailing list
CQ-Contest@contesting.com
http://lists.contesting.com/mailman/listinfo/cq-contest
_______________________________________________
CQ-Contest mailing list
CQ-Contest@contesting.com
http://lists.contesting.com/mailman/listinfo/cq-contest
|