Stories
Slash Boxes
Comments

SoylentNews is people

Artificial Intelligence Aims to Make Wikipedia Nicer to Newcomers

posted by cmn32480 on Monday December 07 2015, @04:47PM   Printer-friendly
from the don't-be-a-meanie dept.

HughPickens.com writes:

Tom Simonite writes at MIT Technology Review that the Wikimedia Foundation is rolling out new software trained to know the difference between an honest mistake and intentional vandalism in an effort to make editing Wikipedia less psychologically bruising. One motivation for the project is a significant decline in the number of people considered active contributors to the flagship English-language Wikipedia: it has fallen by 40 percent over the past eight years, to about 30,000.

Research indicates that the problem is rooted in Wikipedians' complex bureaucracy and their often hard-line responses to newcomers' mistakes, enabled by semi-automated tools that make deleting new changes easy. The new ORES system, for "Objective Revision Evaluation Service," can be trained to score the quality of new changes to Wikipedia and judge whether an edit was made in good faith or not. ORES can allow editing tools to direct people to review the most damaging changes. The software can also help editors treat rookie or innocent mistakes more appropriately, says Aaron Halfaker who helped diagnose that problem and is now leading a project trying to fight it. "I suspect the aggressive behavior of Wikipedians doing quality control is because they're making judgments really fast and they're not encouraged to have a human interaction with the person," says Halfaker. "This enables a tool to say, 'If you're going to revert this, maybe you should be careful and send the person who made the edit a message.'"

Original Submission

 
This discussion has been archived. No new comments can be posted.
Artificial Intelligence Aims to Make Wikipedia Nicer to Newcomers | Log In/Create an Account | Top | 50 comments | Search Discussion
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.

  • (Score: 3, Interesting) by NoMaster on Tuesday December 08 2015, @05:53AM

    by NoMaster (3543) on Tuesday December 08 2015, @05:53AM (#273218)

    How do you make a project like Wikipedia, an online encyclopedia anyone* can improve (*assholes and retards not welcome) and keep it, well, working? How do you motivate and encourage contributors while keeping out both the ill-intentioned and the well-intentioned that end up doing more harm than good

    Do things to address the "assholes & retards"?

    1. Get rid of the "everything has a rule, and there's a rule for everything" structure - it has created a Kafka-esque byzantine system which repels newcomers, encourages edit-warring, and rewards editors who thrive on article-camping and rules-lawyering.
    2. Keep the remaining rules clear and unambiguous, continually updated and clarified as edge-cases become evident, and make them easy to find and cite. At the moment the existing rules cross-reference and contradict each other, interpretations are spread across different places with different versions and clarifications depending on where you look, and they're hidden behind / referred to by cryptic acronyms more appropriate to the CIA or a cult than an "encyclopedia anyone can edit".
    3. Make it easy for casual editors to escalate issues. Currently, you have to know which of the byzantine & contradictory rules applies, how to find the appropriate dispute/escalation page, and how to correctly format and fill out the complaint. There is a barrier to entry there that rewards editors with the time and inclination to play the system, while at the same time discouraging the casual editor from escalating problems before they become major issues.
    4. Make a clear distinction between subject matter expertise and conflict of interest - one is good, the other is bad.
    5. Ditch anonymous / IP edits. Make it mandatory for all editors to log in and provide a working email address for contact by Wikipedia staff / automailers (but keep it hidden from other editors). Lock/delete editors whose email bounces or who don't respond to system-generated keep-alive verification emails after x months.
    6. Drop the silly belief in "behaviour outside Wikipedia counts for nothing" when it comes to editor behaviour / conflicts of interest. Currently it's nearly impossible to show a conflict of interest if it isn't evident from other on-Wikipedia behaviour - all the person accused has to do is say "nope, nothing to do with me, just someone with the same handle, either stop harassing me or take any on-Wikipedia proof you have to WP:COIN".
    7. Absolutely fsckin' nail editors who are operating in poor, if not outright bad, faith. The first step in moderating every dispute that is escalated should be to check the previous discussions for signs of editor bad faith. An editor arguing their point, discussing differences, and attempting to find a workable middle ground is operating in good faith - an editor who continually quotes TLAs without detailing the relevance, reverts other's changes, and tries to make the talk/discussion/arbitration process fail rather than 'lose' isn't, & should be suspended if not outright banned.

    Yes, there are some fairly fundamental changes involved involved in those suggestions. That's because Wikipedia is fundamentally broken...

    (On that last 2 or 3 of points: I'm currently closely watching - but not involving myself in - a case where a person had a [quite reasonable, imho] gripe with the owner & participants of an outside forum, and ended up letting the door hit their arse on the way out. They've since gone full-on stalker-nutjob for the last few months, harassing the forum with spam/offensive posts, using the DMCA to try to take down the forum & owner's YouTube channel, & re-activating an old Wikipedia account to involve themselves in an AfD discussion & general edit-warring of the forum owner's Wikipedia page.

    And yet - despite the fact that 30 seconds of Googling and http requests is enough to connect the disgruntled forum user with their real name, location, employer, their behaviour, and everything else to their Wikipedia account and at least at least two other sock/meatpuppet IPs used in the AfD discussion / edit war / attempts to "win, or derail the whole discussion - Wikipedia won't do anything because all the original dispute happened off-site, the editor concerned denies any connection, and Wikipedia assumes good faith...)

    --
    Live free or fuck off and take your naïve Libertarian fantasies with you...
    Starting Score:    1  point
    Moderation   +1  
       Interesting=1, Total=1
    Extra 'Interesting' Modifier   0  
    Karma-Bonus Modifier   +1  
    Total Score:   3  

  • (Score: 0) by Anonymous Coward on Wednesday December 09 2015, @03:25AM

    by Anonymous Coward on Wednesday December 09 2015, @03:25AM (#273775)

    The last I heard, decisions in the Articles for Deletion process were supposed to be based upon the arguments presented in the deletion discussion, not on the mere number of accounts favouring each side.

    If someone appears to be misusing multiple accounts, one can request a so-called sockpuppet investigation [wikipedia.org] and sometimes the matter will get looked into. In doing those investigations, only a few of Wikipedia's administrators (those with "CheckUser" privilege) are able to see the IP address from which someone is logging in, and they are supposed to only look at that information only when there's "clear, behavioural evidence" of abuse.

    • (Score: 2) by NoMaster on Wednesday December 09 2015, @07:38AM

      by NoMaster (3543) on Wednesday December 09 2015, @07:38AM (#273845)

      On consideration, I probably shouldn't have mentioned that specific case - it was an example only of how easily Wikipedia's too-rigid processes can be manipulated and, in order to keep it non-identifiable except to those already involved in / watching it, I kept detail to a minimum. But since I did bring it up and you've suggested an option for dealing with it, I'll expand on relevant parts a bit:

      • The known sock/meatpuppet IP accounts I referred to did not vote in the AfD process (although others did - all on the side of the editor with the CoI), nor did they (as far as I know) edit the page. They only commented in the discussion to back up the editor involved and muddy the waters further. Other IP accounts did vote (all on that editors , but I haven't bothered to look at them).
      • While I don't know the IP the editor involved uses to log in to WP, those sock/meatpuppet IP addresses are somewhat geographically separated and not 'owned' by him. They can, however, be easily connected to him through his off-wiki activity - amongst other things, his name, job, employer, contributions to his field and hobbies, and other interests are well known in certain circles, and investigating those avenues shows he has access to the IPs in question.

      WP:SPI won't work, because:

      • WP:SPI requires behavioural evidence from within Wikipedia; the page you linked clearly states "The evidence will need to include diffs of edits that suggest the accounts are connected". Since the known sock/meatpuppet IPs have not edited, only commented on the talk page of this one article, there are no diffs to show.
      • Even if you could push past that to initiate a WP:CHK technical investigation, it will be rejected because "CheckUsers will only conduct a technical investigation if clear, behavioural evidence of sock puppetry is also submitted; if you ask for technical evidence to be looked at but do not provide behavioural evidence, the investigation will not be allowed to proceed".

      There's a lot more to it than that - the article has been to WP:AfD twice (the second time nominated by the editor in question just after his falling out with the article subject); it's just been through a (failed) WP:DRN, and 'discussion' is still ongoing.

      The matter of IP accounts has even been raised several times by the editor in question to hint darkly that there's ongoing off-wiki collaboration organised by the article subject to influence the vote/content - despite the fact that all the IP comments have agreed with the editor in question, and all the IP !votes were for deletion. One result of that has been to make it impossible for anyone other than the editors who have been involved up to this point to work on the page - anyone, from a casual IP right up to a respected but previously-uninvolved editor, will be (& have been) accused of bias or collusion and have (and had) their edits reverted &/or their contribution to the ongoing discussion tainted by those accusations.

      Basically, the editor in question is playing a long game - not to improve wikipedia content, but to succeed in his little on and off-wiki vendetta against the subject (who has repeatedly stated he does not care about his Wikipedia page) by either getting it deleted or failing all the Wikipedia processes until everyone else gives up.

      The truly stupid thing is I actually agreed with his reasons for giving up on the article subject's forum - what he was saying was right, he stood up for it, and the article subject (who I agree is a bit of a cock) and sycophantic forum members harassed and trolled him until he left. But his behaviour since then - literally trying to destroy the article subject's website, credibility, and livelihood - is seriously unhinged. Though it's not obvious to the casual observer or Wikipedia hierarchy because he's compartmentalised it all, he's keeping calm about it, denies it & makes vague claims of harassment if it's ever mentioned, and is concentrating on playing the game of rules to the bitter end.

      --
      Live free or fuck off and take your naïve Libertarian fantasies with you...