9
edits
Wiki spam: Difference between revisions
m
Reverted edit of 217.70.108.106, changed back to last version by 66.250.68.54
(explaining from NTPOV) |
m (Reverted edit of 217.70.108.106, changed back to last version by 66.250.68.54) |
||
| (11 intermediate revisions by 8 users not shown) | |||
| Line 1: | Line 1: | ||
All [[large public wiki]]s get [[spam]] - unsolicited commercial content which, in [[wiki]]s, is typically inserted randomly in articles by [[anonymous edit]]s. | |||
[[Consumerium Services]] must be especially watchful of commercial activity since the [[Consumerium buying signal]] directly influences buying decisions! Any removal of accurate negative data, or insertion of inaccurate positive data, or any [[systemic bias]] in demoting or promoting some commercial service or product over another, would seriously degrade trust in that buying signal and the whole [[healthy buying infrastructure]]. So it is important to really understand '''wiki spam''' and all the ways [[funded troll]]s might provide some advantage by manipulating or altering the overall [[Publish Wiki]] and [[Research Wiki]] mechanisms that, end to end, affect the [[price premium]]s and [[green light]]s. | |||
Here are some references on the generic problem of public wiki spam: | |||
* CommunityWiki http://www.emacswiki.org/cgi-bin/community/WikiSpam | |||
* c2 http://c2.com/cgi/wiki?WikiSpam | |||
* OpenWiki http://openwiki.com/ow.asp?WikiSpam | |||
* Meatball http://www.usemod.com/cgi-bin/mb.pl?WikiSpam | |||
* s23 http://wiki.s23.org/wiki.pl?WikiSpam | |||
* GründerWiki http://www.wikiservice.at/gruender/wiki.cgi?WikiSpam (German) | |||
* blog chongqed.org http://chongqed.org/ | |||
Most differentiate between '''wiki spam''' which is externally motivated, and genuine "[[new troll]]s", those '''honestly attracted by the [[wiki mission]].''' Not the old trolls that started the wiki, because they '''neurotically thought they were the best people to control''' the wiki mission. By definition the new trolls are sincere, while the old trolls simply seek to retain power. In this view, a '''spammer''' is an old troll from the existing ''real world''' [[power structure]] of money and interests, who ''masquerades'' as a new troll. This makes new trolls look bad, and helps justify [[sysop vandalism]]; | |||
True trolls are made by [[troll-formative injustice]] and maintained by [[anti-troll bias]]. In general they do not make bald pro-corporate comments or link to commercial web sites unless they are truly the best reference on some issue. This is rare, and the rarity of it can actually be detected and used to determine who is a [[funded troll]] from the old power structure, and who is a genuinely concerned activist from the [[New Troll point of view]]. | |||
From [[NTPOV|that view]], here is how [[trolls]] view [[spam]] in the context of their struggle: | |||
---------------- | ---------------- | ||
| Line 13: | Line 25: | ||
::::So-called "[[spam]]" is a symptom of having no clear process to deal with the [[funded troll]]. Most [[wiki spam]] is actually subtle and consists of inappropriate references to commercial products or services including raising questions or issues about them that put one type of service in a competitive advantage to another. | ::::So-called "[[spam]]" is a symptom of having no clear process to deal with the [[funded troll]]. Most [[wiki spam]] is actually subtle and consists of inappropriate references to commercial products or services including raising questions or issues about them that put one type of service in a competitive advantage to another. | ||
:::::''side note'' - Obviously [[Consumerium buying signal]] is doing this honestly and openly. But much [[Wikimedia corruption]] consists of an over-tolerance for specific corporate interests, e.g. | :::::''side note'' - Obviously [[Consumerium buying signal]] is doing this honestly and openly. But much [[Wikimedia corruption]] consists of an over-tolerance for specific corporate interests, e.g. Bomis, that advance their own interests over the [[GFDL corpus]] as a whole by sponsoring [[sysop vandalism]]. | ||
::::Tolerating unlimited [funded troll]]s is to permit the [[systemic bias]] of "whoever has money to pay them" into the [[community point of view]]. But to react by censoring them has of course the opposite effect to that intended: if I wish to promote [[Coca-Cola]] then I simply insert spam for [[Pepsi]] and the reactive stupidity of the [[sysop power structure]] will end up favouring my actual sponsor. So the right reaction is one process that doesn't care who is funded and who is not, and simply determines that: | ::::Tolerating unlimited [funded troll]]s is to permit the [[systemic bias]] of "whoever has money to pay them" into the [[community point of view]]. But to react by censoring them has of course the opposite effect to that intended: if I wish to promote [[Coca-Cola]] then I simply insert spam for [[Pepsi]] and the reactive stupidity of the [[sysop power structure]] will end up favouring my actual sponsor. So the right reaction is one process that doesn't care who is funded and who is not, and simply determines that: | ||
| Line 19: | Line 31: | ||
:::::"Hey, you are not representative of the [[New Troll point of view]], the statistical evidence and sequence of events is such that the [[balance of probabilities]] favours the theory of you being paid to be here, not having shown up due to some natural or normal expansion of our user base to include such as yourself." Is this easy to figure out? No. But given a [[faction]] system is possible. If in a given month you get 20 new "Reds" and 10 new "Blues", you might observe that one blue defends commercial products very specifically, two reds attack them. So if in one month four of the ten Blues are strongly promoting commercial products but only two of the Reds are strongly defaming them, this is a big shift, and there are tests you can apply to determine if this skew is the result of some deliberate attempt to influence or alter the [[neutral point of view]]. | :::::"Hey, you are not representative of the [[New Troll point of view]], the statistical evidence and sequence of events is such that the [[balance of probabilities]] favours the theory of you being paid to be here, not having shown up due to some natural or normal expansion of our user base to include such as yourself." Is this easy to figure out? No. But given a [[faction]] system is possible. If in a given month you get 20 new "Reds" and 10 new "Blues", you might observe that one blue defends commercial products very specifically, two reds attack them. So if in one month four of the ten Blues are strongly promoting commercial products but only two of the Reds are strongly defaming them, this is a big shift, and there are tests you can apply to determine if this skew is the result of some deliberate attempt to influence or alter the [[neutral point of view]]. | ||
:::::But to get to that, you must acknowledge that there is a [[New Troll point of view]] and that changes in the flow of attention into the wiki | :::::But to get to that, you must acknowledge that there is a [[New Troll point of view]] and that changes in the flow of attention into the wiki do matter - that there are good reasons to study them and be concerned with [[systemic bias]] problems: | ||
:::::In this model, "spam" is simply statistically infrequent linkage or approval of some commercial product, perhaps as quantified by a [[URI]]. | :::::In this model, "spam" is simply statistically infrequent linkage or approval of some commercial product, perhaps as quantified by a [[URI]]. | ||
:::::There are other useful tests like "is the URI linked to, presently for sale?" If so then it's likely someone trying to boost up the [[page view]]s. | :::::There are other useful tests like "is the URI linked to, presently for sale?" If so then it's likely someone trying to boost up the [[page view]]s. | ||
==Solutions...how to fight spam on wikis== | |||
I have a mediawiki site set up that just got spam links (ment to increase page rank in google), how do you fight this? | |||
*The only solution I can think of is having what many sites use to stop automated form submission, which is have an image that contains a word that the user has to retype. This would be a slight burdon to the user, but it would make the user and the spammer on the same level, since one could delete spam just as fast as a spammer could produce it. [[wikibooks:User:RobKohr]] | |||