Extension talk:Antispam

Previous page history was archived for backup purposes at Extension talk:Antispam/LQT Archive 1 on 2015-06-25.

Not sure why this was censored from the article, but...

While MediaWiki is officially supported by the commercial CleanTalk.org service, the service was primarily designed for other purposes - such as blocking comment spam to Wordpress and similar blog software, blocking spam webforum posts and the like. A wiki differs from a forum or blog comment as the wiki user takes an entire existing section of an article (or an entire page), often modifies just a small portion, then resubmits the entire revised text as the "new" version of the page. If there's content that CleanTalk sees as questionable or spammy (such as external links or contact info) CleanTalk may add the user who edited the page to its semi-public IP blacklist or e-mail address blacklist, even if the offending content was in the original article before the user made the edit. This can cause false positives.

Cleantalk does provide a webpage where the administrator of a client site may see what's being accepted or blocked. This interface has a provision to manually report a result as a false-positive or false-negative. Unfortunately, reporting a result as a false positive will not get the unfortunate user removed from Cleantalk's "naughty list". Cleantalk does list more "bad" IPs than other comparable services - such as the (free) StopForumSpam - but it comes at a price of greater numbers of false positives.

The effects of a false-positive are also more severe on a wiki than on a blog-style comment form. If a blog comment is sidelined in error, there is usually some way to manually let it through. The same provision does not apply to failed attempts to edit a wiki page, and likely can't be easily provided if other users have been editing the same page in the meantime (leading to edit conflicts if the erroneously-blocked revision were manually resubmitted later).

I would like to know if we can run this just for new page creation and edits and not for account creation?

Are there other configuration options available, i.e. not to run this for certain user groups, etc?

We use AWS servers to host our wiki and our ip address of our servers change constantly as we scale, will this cause problems recognizing the traffic as coming from a single wiki?

It is possible (and advisable) to exempt all but new or unregistered users from Cleantalk, due to the risk of false-positive results from their server.

    // Skip antispam test if user is member of special group

    if ( $wgUser->isAllowed('cleantalk-bypass') ) {

        return;

    }

There are also some settings in extension.json which may be useful. I'd suggest at a minimum setting:

                "CTNewEditsOnly": {

                        "value": true

                },

so that only user registration and page creation are submitted to Cleantalk. There's otherwise too much risk that Cleantalk will see something in the original article body which looks spammy or suspect, blame the last user who edited the page (as an edit resubmits the entire article or article section to Cleantalk's scrutiny) and trigger a false-positive which gets the user added to Cleantalk's naughty list. That can cause problems for the user later, as well as causing the edit attempt to fail with an error.

If you have groups like autoconfirmed, sysop and the like you may want to grant those groups the "cleantalk-bypass" attribute so that they're not having everything they write submitted to Cleantalk for review.