jm wrote:
> On 09/05/2005, at 4:04 PM, Robert Klemme wrote:
>
>>>>
>>>> When this was first published, I thought, wow this is cool.
>>>> But, since then I have been reading a book called Freakonomics.
>>>> And now, well, I wonder if this email provides an incentive
>>>> (that did not previously exist) for those seeking public
>>>> recognition, to post 'junk' so their name appears near the
>>>> top of the stats.
>>>
>>> Yes, agreed.
>>>
>>>> Anyway, just wondering...
>>>
>>> I OTOH think optimistically, these stats will also provide us an
>>> informations to locate and avoid such trolls.
>>
>> Hm, honestly, I don't think so.  First, your stats come after the
>> fact but more important, how do you tanslate quantity into a troll
>> indicator?  I'd say we need at least more sophisticated figures for
>> that (text pattern analysis or whatever)... :-)  Apart from that,
>> usually you can detect them well by just looking at their articles.
>
> Anyone care to give the new classifier a go at this? See if it can
> evaluate the fitness of the threads for "useful content" or some other
> classification system. I'll leave it to someone else's imagination as
> to what constitutes fit. You could also then come up with a thread or
> post rank/score which could be added up for each post the author
> makes. This would mean that people with a good post score would be
> more likely to be read than those with a low post score who would be
> ranked lower in such a weighted stats systems.

A funny idea to play with but for real world application I prefer to use
CS (common sense, not computer science).  Note: not everything technically
feasible has to be done or even makes sense. :-)

Kind regards

    robert