On Thu, Jul 26, 2012 at 5:00 AM, Derric Atzrott datzrott@alizeepathology.com wrote:
This way if people feel motivated at cheating at captcha they will end up
helping Wikipedia It is up to us to try to balance things out.
I'm pretty sure users will be less annoyed at solving captchas that
actually contribute some value.
Obligatory XKCD: https://xkcd.com/810/
;-)
The best CAPTCHAs are the kind that do this. Look at how hard it is to beat reCAPTCHA because they have taken this approach. One must be careful though that the CAPTCHA is constructed such that it won't be as simple as a lookup though, and will actually require some thought (so that probably eliminates the noun, verb, adjective idea).
This idea has my support.
We should use less CAPTCHAs.
If the problem is spam, we should build better "new URL" review systems. There are externally managed spam lists that we could use to identify spammers.
'new URL' s could be defined as domain names that were not in the external links table for more than 24 hrs.
Addition of these new URLs could be smartly throttled.
un-autoconfirmed edits which include 'new URLs' could be throttled so that they can only be added to a single article for the first 24 hours. That allows a new user to make use of a new domain name unimpeded, however they can only use it on one page for the first 24 hrs. If the new URL was spam, it will hopefully be removed within 24 hrs, which resets the clock for the spammer. i.e. they can only add the spam to one page each 24 hrs.
Another idea is for the wiki to ask the user that adds new URLs to review three recent edits that included new URLs and ask the user to indicate whether or not the new URL was SPAM and should be removed. This may be unworkable because the spam-bot could use the linksearch tool to check whether a link is good or not.
-- John Vandenberg