CLP stats: last 500 posts

[Terry Reedy]

On 12/10/2016 9:43 AM, Steve D'Aprano wrote:
> On Sat, 10 Dec 2016 09:28 pm, Terry Reedy wrote:
>
>>>> The spammer will still be counted,
>>
>> Why reward someone who actively evades defenses?  If you want to count
>> spam, it is mostly missing, at least as far as python-list is concerned.
>
> Its not a reward. Spammers are not like trolls, they don't hang around to
> see the result of their posts.

To me, the relevant difference is between posts related to python and 
those not.  It is usually clear which is which.

 > There no evidence at all that this Italian
> spammer is looking for replies or responses to his(?) posts. He apparently
> just fires them out.
>
> I think that it is relevant that comp.lang.python receives X spam messages
> from a certain person. It gives a picture of the health of the newsgroup:
> how much of it is spam? Hopefully only a small amount.

Python-list gets unrelated-to-python spam from lots of people.  They are 
not outliers (unlike jmf's now blocked trolls), but contaminents from a 
different universe.

I agree that the fraction of messages that are clearly spam has some 
interest in itself, and definitely should be as small as possible. But I 
contend that they should be excluded from a study of the universe of 
python-related messages.

My other point is that this small sliver that used to get passed through 
is extremely biased and statistically worthless as a study of 
python-list spamming.  If one wanted to study the rate and nature of 
contamination, or the effectiveness of filtering, one would need access 
to the raw stream of submissions.

-- 
Terry Jan Reedy