🚀 go-pugleaf

>>Things like how many articles were posted, how many crossposts,
>>how many multiposts, how many were cancelled, subject lines of
>>cancelled articles or whatever it takes to show the value of the
>>service.

>>If you aren't keeping statistics of that sort, then you should
>>be. How else do you measure if you are being effective in
>>achieving your objectives?

danny@cs.su.oz.au (Daniel Yee):
>That's exactly what they have been doing over the last few months
>-- posting lists of articles that would have been cancelled if the
>auto-cancellation were happening, with basic statistical analysis.

Greg posts statistics for the proposed aus.* system. Good.

What I was after was some stats for the ECP/EMP cancel system. 
Greg posts a notice as to what was cancelled and a copy of the 
article that was cancelled. Also good.

There is no summary of the ECP/EMP cancels.

I would like to see some stats along the lines of "there were 
nnnn messages available, and this service saved you from having 
to see nnnn messages that you didnt want to see".

The concept of "messages that you didnt want to see" complicates 
things just a little.

Just how much of a benefit the cancels provide depends on the 
individual readers circumstances.

For example, Joe Spammer posts a message about how you can become 
a millionaire simply by giving him $5.

He writes it under one message id, and crossposts it to 50
newsgroups.

How does this effect the reader?

If you are subscribed to all 50 newsgroups with software that 
doesnt know about crossposts, you see the same message 50 times.

If you are subscribed to all 50 newsgroups with software that 
knows about crossposts, you see the message once.

If you are subscribed to all 50 newsgroups with software that 
knows about crossposts and also knows that any article that is 
posted to more than x newsgroups is spam, you dont see the 
message at all.


How about if Joe Spammer posts 50 copies of his message under 50 
different message ids?

All of the above readers see 50 copies of the message.

In addition, the long haul transport costs are multiplied by 50, 
as are the disk storage requirements.

Bright ideas on how to solve this would be appreciated, but even 
I think this is getting close to "something should be done about 
it" territory.


How about if Joe Spammer posts just one copy of his message to
just one newsgroup?

All the readers see one copy of the message.

It uses no more resources than any other message.

Its still a con.

Bright ideas are also welcome on this. Perhaps you have to deal 
with it in the real world - bring it to the attention of Consumer 
Affairs or something like that.


Anyway, back to the stats. I'd like to see a separate breakdown 
for ECP, EMP, MMF etc so that readers in different situations can 
assess the value of the service to them.

Would it be possible to publish the raw data in a format that 
would make it easy for people to import it into a spreadsheet and 
do their own stats?

...Dave