Bayesian approach to greylisting of trolls

[Chan Chung Hang Christopher]

R Kimber wrote:
> On Thu, 07 May 2009 10:21:17 +0800
> Christopher Chan wrote:
>
>   
>> Are you volunteering to babysit the thing? A thought experiment is 
>> rather pointless even as theories are pointless until they become
>> proven laws. I personally feel, having had tried this sort of thing
>> against 419 scams, that this will not help as there is not set
>> pattern to filter and bayesian approaches to killing 419 scams, which
>> more or less have a set pattern or set patterns, will always need
>> constant eye balling.
>>     
>
> To which Bayesian approach are you referring?
>
>   
Perhaps I should not have said 'bayesian approaches'. In the end, there 
was only one tool used...bogofilter...it was the only one that was fast 
enough to cope with the load. So other approaches not used by bogofilter 
are not accounted for.