Wikipedia:Arbitration/Requests/Case/Magioladitis

I've had interactions with Magioladitis before, and I've often questioned their competency when it comes to the bot policy. The community has repeatedly failed to deal with the ongoing issues caused by Magioladitis' problematic automated editing. I believe it's time that the Arbitration Committee take a detailed look. The main questions we can't seem to answer (and why we can't answer them) are:

• Does Magioladitis' repeated violations of the bot policy and problematic semi-automated editing warrant restrictions? (The community often fails to hold the operator accountable for a bot's activity. They see the issue with an automated script as a purely bot policy issue, not an issue warranting a look at the operator. Further, the community at large is not knowledgeable about bots, making it difficult to reach consensus at ANI. Discussion just peters out because it's too niche for people to feel comfortable participating.)
• Does Magioladitis' continued violations of the bot policy constitute "sustained or serious disruption of Wikipedia" or "consistently or egregiously poor judgement" in the sense of WP:ADMINCOND? (Potential desysopping is purely a matter for the Arbitration Committee.)

Some other background:

• Wikipedia:Bot_owners'_noticeboard/Archive_10#User:Yobot_Issue
• User_talk:Yobot/Archive_9#Stopping

I found an additional disturbing instance of poor judgement. Magioladitis has, in multiple instances, replaced template redirects with the actual template name in a transclusion and then deleted the template redirect himself, without discussion. These deletions are wholly inappropriate. The fact that Magioladitis has taken similar situations to RfD makes me think he's aware they should be discussed. Others have shown evidence of Magioladitis unblocking his own bot, which is a clear violation of WP:INVOLVED. This was done repeatedly after requests to stop. This more directly brings his administrator status into question.

• @Drmies: I've added an explanation of why I believe the community cannot/will not fully address the issues related to automated and semi-automated editing here. See the parentheses after the first bulleted question. As for whether this is "pressing", I don't believe it is, but I also don't think that should preclude a case. If ArbCom only considers cases of "pressing" desysoppings, the effect is that the slow burn of abuse of administrator tools or poor judgement has no recourse. WP:ADMINCOND is a community policy that states desysopping should be considered for "consistently ... poor judgement". Since ArbCom is the only venue for desysopping, that's tacit community support for ArbCom accepting "slow burn" cases where serious judgement issues are raised. ~ Rob₁₃^Talk 08:06, 30 December 2016 (UTC)
  • @Drmies: The community seems to also be unable to deal with the semi-automated editing with the same issues on the Magioladitis main account whenever the bot is blocked, which is a pure behavioral issue. This is very similar to the case on Rich in terms of facts. I personally believe that should be handled here too, but perhaps the community could handle that if Magioladitis were not an administrator and AWB access could be revoked. Currently, it cannot be, since AWB is available to all admins. (Note that I sent an email to the Committee about this currently ongoing issue. Not sure if you've read that yet.) ~ Rob₁₃^Talk 20:05, 31 December 2016 (UTC)

Responding to kelapstick

@Kelapstick: I want to correct a few things in normal English to make sure the non-technical arbs aren't lost in the bot policy stuff. If I go over the word limit doing this, I'd appreciate an exception, since clarification is probably important to some arbs.

• The generally-accepted guideline is that edits which don't change the visual output of the page usually violate COSMETICBOT. There are some exceptions (i.e. accessibility-related changes), but the exceptions aren't at issue here. For instance, some of the edits resulting in the current block of Yobot were these: . They only perform wikicode cleanup, with no other changes.
• Whether Yobot is violating COSMETICBOT isn't at issue here. The community has found consensus that the specific edits violate COSMETICBOT many times, and it wouldn't be within ArbCom's remit to determine what edits do or do not violate COSMETICBOT when the community has been quite clear in this instance. Even Magioladitis agrees the above edits are cosmetic-only, although he's expressed more confusion about some other violations on his talk page recently.
• What the community hasn't been able to do is actually stop the cosmetic edits from occurring. Yobot is blocked and then unblocked (often by the operator) with claims that the issues will not recur. They always do, often very quickly after the unblock. Even when a block is in place, the edits wind up swapping over to the main account. The community appears to be ill-equipped to deal with this in a "final" way.

If you have any technical questions, please do ask. ~ Rob₁₃^Talk 16:44, 30 December 2016 (UTC)

• Ideally, the scope should also include semi-automated editing from the main account. This is mostly in relation to the problematic edits being swapped over to Magioladitis' main whenever Yobot is blocked - something currently happening as we speak. This is probably within the "Yobot" scope, but it should be made explicit to prevent scope shenanigans later on. ~ Rob₁₃^Talk 21:29, 30 December 2016 (UTC)

I have been involved in some of the discussions on this in the past, having brought issues to the attention of Magioladitis and his bot. I too think there is good reason to consider a case, but perhaps not for the reasons of the proposer. The issues mentioned, individually, do not need arbitrator attention. Taken together they would normally be resolved some other way, such as a discussion and community sanction on another notice board.

There are issues here though that cannot be easily resolved at another venue, to do with Magioladitis's actions and behaviour as an administrator. First his block log although not long shows repeated instances of poor judgement as an editor, never mind as an administrator. Second the similar log for Yobot has a number of unblocks performed by Magioladitis in contravention of WP:INVOLVED, that administrators should not use the tools in a matter they are at all involved in. And under the bot policy a bot’s contribution are considered the responsibility of the bot’s owner, making for a very lengthy series of blocks which Magioladitis’s edits have been subject to. Given all this, this seems well within the scope of the committee.--JohnBlackburne^words_deeds 16:56, 27 December 2016 (UTC)

(Full disclosure: I blocked Yobot twice, once in 2010 and once in 2011 and I was involved in ANI discussions about Magioladitis and Yobot at that time.)

I've met Magioladitis a couple of times. We even chatted about this issue a few years ago. He's a nice guy. I have no doubt that he means well, nor that he is very good at writing bots that—for the most part—perform useful functions.

Nonetheless, there have been issues with Yobot, Magioladitis' approved bot account, going back several years. Specifically, the bot makes trivial or non-rendering changes on a mass scale—edits such as bypassing template redirects or formatting whitespace—contrary to the Bot Policy and the AWB rules of use. Dozens of editors have attempted to discuss this with Magioladitis over the course of six or so years and there have been numerous ANI threads, as linked by the filer, but in all cases the discussion either peters out with no resolution or is closed with assurances from Magioladitis that the problems will be addressed, only for them to resume a few weeks later. Frankly, Magioladitis does not take this issue seriously and by implication dos not take the community's concerns seriously, and does not take his responsibilities as a bot operator or an administrator seriously. Several of the principles from WP:ARBRF apply here.

These issues alone, and their longevity, make this ripe for ArbCom—it needn't be a long and complex case, but arbitration can break the deadlock and set a precedent. But there are associated issues of admin misconduct, which can only be handled by ArbCom—Magioladitis has repeatedly unblocked his own bot, which has long been frowned upon, all the more so given that he has been warned against doing so in the past; and he has repeatedly continued the task for which Yobot was blocked under his own account (for which he has been blocked in his own right five times). Cf. Wikipedia:Arbitration/Requests/Case/Rich Farmbrough#Misuse of administrative tools.

I'm sorry that this has got to this stage, but this is a long-term problem that the community has failed to resolve. Only Arbcom can adequately address these issues. HJ Mitchell | Penny for your thoughts? 18:45, 27 December 2016 (UTC)

@Rich Farmbrough: I'm sorry, Rich. I very much value Marios' work, just as I value yours, and while I disagree with the implementation of the remedies from your arbitration case, the root issue was very much the same. Your many virtues are not a licence to ignore policy, just as a certain other editor's many featured articles (for which I respect him greatly) do not excuse calling other editors cunts. Moving whitespace around or changing {{foo}} to Foo or Foobar is not a helpful edit. Persisting with it after you've been asked not to is problematic. Persisting with it after your bot was blocked for it more so. Unblocking your own bot so it can continue making the changes people are complaining about, or evading the block by running the task on your own account is disruptive. The only people who dispute that are the people who run bots that make these sorts of edits. You're welcome to start an RfC if you think that these sort of edits should allowed (either in general or on a case-by-case basis), but if consensus is against you, you can't just subvert it by sheer brute force. HJ Mitchell | Penny for your thoughts? 19:59, 27 December 2016 (UTC)

Perhaps it is easy to forget the sheer amount of work that Magioladitis has done, both through his coding work on AWB, his bot edits, through Yobot, and through manually editing himself.

It would be strange indeed if, over the decade plus, and the 5 million or so edits, Magioladitis had not ruffled a few feathers, especially given how many Wikipedia editors suffer from various manifestations of WP:OWN.

It is perhaps unfortunate that Marios responds to messages with explanations of how events transpired, and the steps he is taking to avoid repetition, in the case of "errors" (which are generally nothing more than failing to correct a tricky issue, rather than actually causing a problem). Were he to apologise fulsomely, he might find some of his interlocutors more receptive. On the other hand he might not.

I refer to the case of User:Lightmouse who patiently waited for years to achieve consensus that he was allowed to implement consensus, only to be stonewalled at every turn.

Magioladitis has done sterling service, for little or no reward, and always been civil with those who question what he does.

It is a crying shame that anyone should consider the renaming of "NeuroethologyNavbox" to "Neuroethology" anything other than improvement, even more so that they should be so incensed that they dig into 7 year old discussions with someone who has effectively long left the project, but who supported consistency in template naming.

Using ArbCom as a means of "winning arguments" is to fatally misunderstand what Wikipedia is about, which is the collegial production of good quality, well presented, maintainable, encyclopaedic content.

All the best: Rich Farmbrough, 19:09, 27 December 2016 (UTC).

To conflate "NeuroethologyNavbox" => "Neuroethology" with "foo" => "Foo" is clearly ridiculous.

The issue here is a MediaWiki bug. Worth notifying Mag about, but not worth an ArbCom case - not even fodder for one, except to show how patiently Marios explained the situation.

And perhaps it is also worth mentioning that Magioladitis has often asked people to take over these tasks. So far I have refused, it is simply not worth the aggravation.

All the best: Rich Farmbrough, 20:12, 27 December 2016 (UTC).

Incidentally ArbCom does not set precedent, so nothing form ARBRF can be used here. Thank God, it was one of the worst arb cases ever, even including items as evidence which Arbs had agreed on the talk page were incorrect. All the best: Rich Farmbrough, 20:17, 27 December 2016 (UTC).

I have observed many edits by Yobot that were cosmetic or unsuccessful. My observation of this bot over the years makes me feel the need to check each Yobot edit that appears on my watchlist. Jc3s5h (talk) 18:37, 27 December 2016 (UTC)

As an editor who has experience in both Bot Editing/AWB and with administrative overreach, I feel compelled to weigh in on the issue.

Cosmetic edits by themselves have not been approved since before I started doing bot edits. If in the context of persuing your bot edit, other editors question the action you're not supposed to continue slugging on. Keeping in mind Betacommand 2 and other prolific bot programmers who have fallen afoul of the consensus of the community, the actions of Magioladitis are significantly lacking.

On the second point of Administrative overreach and refusal to accept community consensus I point at the numerous ArbCom cases in which "Old Guard" administrators having been informed repeatedly of actions that diverge significantly with community consensus were dealt with in various ways from warnings, admonishments, desysopping, topic bans, and community bans.

For these reasons I feel that a case is ripe regarding this dispute as lower forms of intervention have not effected sustaining changes. We don't ask Administrators or Bot Operators to be infallible, we simply ask them to not repeat the same mistakes or faults after correcting the problem. Hasteur (talk) 20:25, 27 December 2016 (UTC)

With respect to the introduction of additional "grey areas" as enumerated by Mag in this revision are crystal clear with respect to policy. Bots are allowed to make any edit not prohibited by policy/standards/best practices (i.e. WP:COSMETICBOT) and for which they have recieved appropriate authorization (The BAG authorization). Editors are allowed to do make any edit not prohibited by policy/standards/best practices for which they have not been restricted by community consensus (i.e. outside a topic ban). As to non-context related edits I would suggest falling back to the COSMETICBOT standard (Cosmetic changes should be applied only when there is a substantive change to make at the same time.). If a editor is challanged on what others have percieved as cosmetic changes, the interim is to not continue making changes similar to the disputed one, but to seek consensus as to if the edit is cosmetic. Baring a few minor exceptions (such as adding sort orders to categories and similar non-printing changes) if a change does not change the visual outcome, then it is very likely a cosmetic edit. It has been suggested that BAG should take a more active role in reviewing the requests for approval both in the initial authorization and subsequent operation of bots, but at this time the group is just the technical side of the review (i.e. expert witnesses) and has little enforcement tools to get bots back to what they were authorized to do. Hasteur (talk) 13:06, 29 December 2016 (UTC)

Post facto "update" I should have clarified my last statement (about what BAG does) as "it is my impression that BAG only does ..." due to the fact that several discussions of rogue bots have gone stale for lack of progress or have been "We're not responsible for it, go to AN*" in my impression. Hasteur (talk) 19:34, 29 December 2016 (UTC)

I urge the Committee to take this case, but suggest they expand the scope to include all bots and bot operators who regularly make cosmetic changes. Beyond My Ken (talk) 01:52, 28 December 2016 (UTC)

@Opabinia regalis: Do you mean "curious", as in "casually interested in how this came about" or "curious", as in "suspicious that there's more to the story, such as ulterior motives"? Beyond My Ken (talk) 06:28, 28 December 2016 (UTC)

@Callanecc: I know little about BAG, and wonder how effective are they at policing the application of those tasks by bot operators once they are approved, or if this is even part of their remit? Beyond My Ken (talk) 06:35, 28 December 2016 (UTC)

Thanks for the clarification. If BAG does not control or monitor the editing of the bot operators and bots they approve, and the community has not been able to control these "supereditors", then ArbCom needs to take up this case in order to provide a check on cosmetic edits by bots and the sometimes battleground-y behavior of some operators. Beyond My Ken (talk) 13:41, 28 December 2016 (UTC)

@GorillaWarfare: - For instance Bg19bot, run by admin @Bgwhite:. Parts of the following edits contravene WP:COSMETICBOT: , , , , , , , , , , , and so on. (I stopped arbitrarily, but any page of the bot's contributions will show examples such as this.)

Every one of these edits has in it one or more changes which make no difference in rendering the page which bots are not supposed to be doing; they're certainly not supposed change "disambig" or "dab" to "disambiguation" or "cn" to "citation needed" or "commonscat" to "commons category" These changes all seem to fall under the general "Fix problems if they exist" rubric, but the fact is that these are not problems, and there is no need to fix them. BG19bot does this kind of thing regularly. Beyond My Ken (talk) 22:27, 28 December 2016 (UTC)

@Anomie: Regardless of the minutia of COSMETICBOT, bots should not be making cosmetic edits. Beyond My Ken (talk) 23:40, 28 December 2016 (UTC)

• Regarding Bgwhite, I would point Arbs to this discussion in which Bgwhite's response to a legitimate complaint about the bot changing the wording of a direct quote was to (1) tell me that I should have formatted the quote in a way that his bot would understand, as if I was editing for his bot and not for the Wikipedia reader, and then (2) to accuse me of "shouting and bullying" when I politely insisted that he needed to take responsibility for his bot's edits. This is unfortunately typical of Bgwhite's response to criticism, which is certainly not behavior becoming of an admin.
  It seems to me that these "supereditors" who make edits in the millions regard themselves as above the community, answerable only to BAG-approval and their own views about what should be and shouldn't be acceptable. There's an arrogance there that's palpable. Something certainly needs to be done about policing some bot operators. Let them fix the actual problems, something they do well, and let the other stuff alone. Beyond My Ken (talk) 01:43, 29 December 2016 (UTC)
• @Amanda: I understand your POV, and you may well be right. It may also be possible that examining the behavior of one bot operator would have a salutory effect on the behavior of others, but given, for instance Bgwhite's combative and defensive attitude as expressed here (and elsewhere), I'm somewhat doubtful. Certainly, a case focused on a single editor would be easier to keep in focus. (And BTW, I'm not anti-bot in any way. Most bots that I've run across do excellent work and I applaud them and their operators.) Beyond My Ken (talk) 10:17, 29 December 2016 (UTC)

I agree with Beyond My Ken that the raised issues are not specific to Magioladitis, but this case might set a precedent (I don’t follow arbitration cases, and hence have a feeling that some of it had been mentioned earlier, e.g., with Rich Farmbrough).

Magioladitis is one of those few highly skilled, dedicated, experienced, respectable, etc., etc., editors who mass-apply minute "fixes" to a vast number of articles at a high rate. In my observation they all miss one important point: those edits inevitably (because of their high rate) interfere with reverting "vandalism" (a general category of rogue edits that also includes spam, WP:NPOV violations, WP:OR, whitewashing, etc.). To simplify my explanation, instead of seeing suspicious user:Vandal in my watchlist I see the trusted user:Magioladitis and relax; when user:Vandal eventually pops up and I check their contributions, I ignore those that can't be reverted (because user:Magioladitis edited after user:Vandal). We end up with one article permanently vandalized and dozens cosmetically improved. I argue that the former grossly outweighs the latter.

The key point here is this: "semiautomatic editors" do not apply their intelligence to have a brief look through the article that they edit, and to find that it was obviously vandalized.
I see three major solutions to this problem: 1) eradicate "vandalism", 2) implement some sort of flagged revisions to all vulnerable pages, or 3) restrict uncontrolled (semiautomatic) editing of mainspace articles and pages (templates) that are transcluded into mainspace articles.
1) appears impossible; 2) has been rejected on this wiki; 3) is feasible and can be implemented with this case. Materialscientist (talk) 05:26, 28 December 2016 (UTC)

@GorillaWarfare: - I've returned here after Beyond My Ken. Incidentally BMK mentioned Bg19bot, and indeed, Bg19bot is one of those that cover up edits to examine. You can find a fresh example at Concrete plant. The bot made some cosmetic fixes . As a result I've missed an edit, and reverted it several days after, only because the contributor popped up on Asphalt plant. Incidentally that edit was also covered up by a typo-fixing wikignome , that is, I've got to both by sheer chance. Both additions are not vandalism, they are just unsourced, unencyclopedic and add spamlinks. I mention them only because they are very fresh and illustrative of the cover-up by "semiautomatic AWB editors" (AWB is very limited in evaluating an article, its references, history and compiled appearance). Materialscientist (talk) 22:52, 28 December 2016 (UTC)

@Anomie: 1) this case is not about bots, it is about editors (who happen to operate bots) and their attitude to editing, 2) my statement is clear: restrict cosmetic, semiautomated editing - it hides previous changes and spams watchlists. Materialscientist (talk) 00:05, 29 December 2016 (UTC)

Bgwhite, I don't recall interacting with you directly, but I had much respect to your work; if you care about this respect, choose your words more carefully and don't twist mine. Back to my point, it is not about the right of bots and editors to edit, it is about editors not checking what they are trying to fix. Of course we can't request people to use their brain, and not only fingers when editing Wikipedia, but we can reduce the number of cosmetic edits. There are many ways and venues to achieve that (we might need to rethink our attitude to BAG approvals, for example), and I'm curious to see what can be done within this case. Materialscientist (talk) 03:30, 29 December 2016 (UTC)

@Anomie, Jo-Jo Eumerus and others: Bots are tagged by wikisoftware and are predictable (say, we know that if a bot rescued refs then someone deleted them). They can be filtered out by wikisoftware. They can be addressed via channels like phab:T11790 and don't need a dedicated Arbcom case, bot-like editors do. Materialscientist (talk) 00:12, 30 December 2016 (UTC)

@Callanecc: the bot approvals group has nowhere near the resources that would be needed to police all running bot activity. Additionally, I do not see the bot policy limiting this function to, or requiring BAG to perform such policing. Bots running out of policy (off task) that are causing problems can and should be dealt with by contacting the operator, and baring a resolution or in an urgent situation by administrative blocks - followed up by a discussion with the operator. BAG will deal with de-authorization review requests from the community. A core principal of bots is that they are an extension of their operator, the human editor who must maintain responsibility and accountability for the contributions that they make via their bot. — xaosflux ^Talk 12:57, 28 December 2016 (UTC)

@Ramaksoud2000: regarding "I thought that BAG members remained on BAG for life" - BAG memberships are not for fixed terms, but members have been retired for inactivity. My interpretation of the membership requirement is that BAG memberships are dependent on community support. When Magioladitis' continued membership was brought to question I asked Magioladitis if he would stand for reconfirmation and he agreed to the reconfirmation. The reconfirmation was just closed as unsuccessful. This closure only specifically reviewed that community support for continuing BAG membership is not currently present. — xaosflux ^Talk 17:08, 29 December 2016 (UTC)

I didn't want to have to comment here since I have nothing to add to the core issue here regarding Magioladitis's behavior, but I see some other things that need to be noted:

• Beyond My Ken seems to misunderstand WP:COSMETICBOT. It's not possible for an edit to be partially against that policy. Either the whole edit consists of cosmetic changes to the wikitext, in which case it's against the policy, or the edit contains at least one change that makes an actual difference to a reader, in which case the edit is not against policy no matter how many cosmetic changes are riding along.
• Materialscientist seems to be asking the committee to somehow change the bot policy to restrict "semiautomated" edits (and presumably also fully-automated edits, since such edits result in the same situation Materialscientist is complaining about) in some poorly-specified manner. Changing policy is, of course, not within ArbCom's remit. Is there a dispute about Wikipedia:Bot policy#Bot-like editing or the like here, or just a wish to change policy so patrollers don't have to look at more than the most recent edit to a page while patrolling?
• Since I'm posting here anyway, I'll confirm Xaosflux's description of the scope of BAG's "responsibility": approval and revocation of approval, and general review or "expert testimony" when requested, but not proactive monitoring of every bots' behavior.

Anomie⚔ 23:29, 28 December 2016 (UTC)

Replies:

• @Beyond My Ken: If you disagree with what the policy actually says, ArbCom isn't the place to try to change it. Misrepresenting your own opinion as policy is also not helpful.
• @Materialscientist: You're clearly making a distinction between semiautomated and fully-automated edits, yes, but why you're making that distinction isn't at all clear since the problems you cite are identical to both types of edits. Perhaps it's just that you know to check behind bots but not behind bot-like editors, so you only consider the latter a problem? At any rate, new policy is still not within the scope of ArbCom.

Anomie⚔ 03:02, 29 December 2016 (UTC)

• Hasteur is inaccurate in stating that BAG is only responsible for technical aspects when reviewing bots, details here. He is correct in saying that post-approval enforcement is mainly left to the community in general. Also, an "invisible" change may be non-cosmetic if it affects screen readers, microformats, or the like.

Anomie⚔ 16:38, 29 December 2016 (UTC)

I also didn't want to add anything here as Magioladitis is the person I'm closest with on Wikipedia, thus biased. But, BMK and Materialscientist decided to ping and attack me.

• Beyond My Ken accuses me of doing WP:COSMETICBOT. On every article he mentions, the bot did what the edit summary states. Every one of the edits he mentioned has BRFA approval and are based upon MOS. Checkwiki error #61, punctuation after a reference, is MOS:REFPUNCT. Checkwiki error #48, wikilink redirecting back to itself, is WP:OVERLINKING. The bot was there on a BRFA task and made a BRFA approved edit.
  BMK has disapproved of my editing in the past and what a substantsive means. BMK has disagreed with removing blank lines between list items (BRFA approval, MOS:LISTGAP). Removing a blank line in itself would be cosmetic. However, when it interferes with screen readers, removing blank lines becomes substantive to the disabled. BMK also argued his inconvenience is greater than all the disabled. Moving TOC to the "correct" spot (WP:LEADELEMENTS) for accessibility reasons (screen readers won't "read" some of the article) has also caused disagreements (#1, #2) and several trips to ANI. What is cosmetic to one is substantive to others.
• Materialscientist does make a good point about vandalism and the point also applies to any bot making an edit. AnomieBOT tries to fix refs or adds dates to some templates, such as {{disambiguation}}. On some BRFA tasks, AnomieBot usually edits within 1/2 hour of the page being edited. Other bots also add dates to other templates such as {{citation needed}}. EmausBot runs daily to migrate interwikilinks and FrescoBot is another example of running daily. Ser Amantio di Nicolao does thousands of bot like edits every day (personally, I think he is an AI that passes the Turing test). Doing manual editing of CheckWiki task everyday, I do see Yobot, my bot and all the above bots covering vandalism. I've had people complain that my bot covered up vandalism from articles I got from monthly dumps... which means the flagged article had the issue the bot fixed for at least a week. Materialscientist also accused my bot and a "gnome" of making a cosmetic edit (the bot did what the edit summary states) that covered up vandalism. Essentially, Materialscientist wants no bots or "gnomes" making any edit as it may cover up vandalism.
• Bots have approval for and do changes that readers don't see. Cosmetic doesn't mean what the readers can't see. Examples are merging templates as result of TfD, changes to categories, changes to Defaultsort, removing duplicate parameters in templates, ...

Bgwhite (talk) 01:17, 29 December 2016 (UTC)

In my mind, Opabinia's suggestion of prodding developers about phab:T11790 until they fix it - or having people here submit the requisite code patches - is a better solution than an Arbcom case. I think that WP:COSMETICBOT being a policy is creating more issues than we would have if WP:COSMETICBOT was not a policy; bot edits obscuring real ones on watchlists is a real issue but it can be fixed in ways other than by a make-dispute policy. Not all policies are of equal value and sometimes ditching them is better than having them. Also, I'd like to note that at least in some tests I have made, replacing template redirect transclusions with direct transclusions speeds up the loading time of a page, so I would dispute that such edits fall under the COSMETICBOT banner. Jo-Jo Eumerus (talk, contributions) 11:33, 29 December 2016 (UTC)

Postscript for the clerks: That I posted here does not mean that I want my talk page filled by notifications about whatever proceedings happen in this case, thanks! Jo-Jo Eumerus (talk, contributions) 11:41, 29 December 2016 (UTC)

Last edit: With regards to unblocking one's own bot, this search link has a few prior conversations on the issue. Jo-Jo Eumerus (talk, contributions) 23:34, 30 December 2016 (UTC)

At the heart of this issue is the fact that WP:COSMETICBOT's only actual policy statement, Cosmetic changes (such as some of the AWB general fixes) should be applied only when there is a substantive change to make at the same time, is uselessly vague in defining boundaries. While several interventions above and in other parts seem to assume an underlying collective understanding of what a cosmetic change is, it is also obvious that these definitions vary from person to person, and from page rendering method to page rendering method. The absence of a clear definition of cosmetic change is at the heart of continuous frictions between various bot operators and editors who act as stewards for articles on their watch list. Whatever else happens, Arbcom should request, by motion, that BAG drive a project-wide RFC to create a usable COSMETICBOT policy which defines what constitutes a cosmetic only edit in 2017, now that rendering methods go far beyond 3 different browsers on a PC.

"A change I cannot see on my browser, and I don't care if your preferred rendering method shows one" falls short. MLauba ^(Talk) 15:14, 30 December 2016 (UTC)

• OID, I haven't made any claims about the civility of the frictions caused by a vague policy in my statement above. This isn't the first bot-related issue or even case tied to cosmetic edits, and the fact that these don't necessarily explode into ugly messes doesn't mean the extremely vague COSMETICBOT isn't overdue getting a precise definition. MLauba ^(Talk) 17:03, 30 December 2016 (UTC)

@Clerks, please don't include me on the notification list for the case, thanks. MLauba ^(Talk) 09:45, 31 December 2016 (UTC)

Contrary to what Mlauba says above, when cosmetic bots are brought up, the discussion is generally civil. Editors complain to bot operator, bot operator says 'ah right it shouldnt do that' and stops it doing it. Or where there has been disputes over if something is cosmetic or not, a discussion is had which comes down either one way or the other. The situation here is that Mag has been told repeatedly over a number of years by multiple people, including experienced bot operators, admins and users, that his bot has been making cosmetic edits, he says 'oh it shouldnt be', fails to fix the issue, bot keeps going, gets told again, fails to fix the issues, block, etc. WP:BOTPOL is explicit enough in that cosmetic edits are not allowed. Complaining that it doesnt define cosmetic edits well enough is really not a defence here as Mag knows full well the difference between a purely cosmetic change and the edge cases where a discussion is needed. RE Drmies: if (as an example) you had a disclosed alternative account which had been blocked for violating a wikipedia *policy*, and you unblocked it with your admin account and continued over an extended period to make the same edits that caused the block in the first place, where would you expect to end up? It has been raised with Mag directly. Its been raised at admin noticeboards. Its been raised at BAG (who do not police misbehaving bots or operators). At this point there is no other venue to see this resolved. Only in death does duty end (talk) 15:43, 30 December 2016 (UTC)

Magioladitis says above As written in many other places what appears to be as do-nothing edits is only the result of some other fail and appears to believe this is a justification for the problematic edits; as has been pointed out ad nauseam in the assorted discussions linked above, he's been saying variants of "the problems were down to a bug which I've now fixed" for at least seven years now, and promptly goes back to doing whatever he was doing before as soon as the heat dies down. (It bears repeating that Wikipedia has dozens of bots working in article space, none of whom since the departure of Betacommand have ever caused this many disputes over so protracted a period, and even Betacommand immediately stopped his bot and discussed the issue whenever anyone raised a legitimate concern about its edits.) The WP:IDIDNTHEARTHAT mentality and general contempt for the concept of "community consensus" (Magiolatidis doesn't appear to grasp that it doesn't matter if he considers edits helpful, if significant numbers of people consider them unhelpful), mean that at the very least the discussion of sanctions of some kind is inevitable. Since the alternative is a free-for-all at ANI—which will be nasty and protracted, probably turn into a flame-war between self-appointed "content creation" and "essential maintenance" editors, and the result of which Magioladitis is likely to disregard if it doesn't go his way in any case—and since desysopping at least needs to be on the table as a potential option given that a key allegation is that he's repeatedly unblocked his own alternative accounts, this is one of those rare occasions where an arbcom case will actually be the least problematic means of resolving a dispute. By all means punt it across to the incoming committee, but please accept this. ‑ Iridescent 18:01, 30 December 2016 (UTC)

At the bare minimum Magioladitis needs to be investigated for WP:ADMINCOND: namely breaking WP:INVOLVED six times by unblocking Yobot . To me this is clearly breaking policy, and at the least requiring of an admonishment of some sort, as well as warning that futher unblocks will lead to a ban of some sort. To quote In general, editors should not act as administrators in disputed cases in which they have been involved. This is because involved administrators may have, or may be seen as having, a conflict of interest in disputes they have been a party to or have strong feelings about. Involvement is generally construed very broadly by the community, to include current or past conflicts with an editor (or editors), and disputes on topics, regardless of the nature, age, or outcome of the dispute. Clearly unblocking your own bot is not allowed by policy. --Jules (Mrjulesd) 21:28, 30 December 2016 (UTC)

This isn't the first time that an admin has got into trouble for unblocking their own bots. Personally I would be happy to see a system where we encourage blocking of bots that appear buggy but with consent for the bot operator to debug and unblock. I'd welcome Arbcom giving guidance that bot operators may only self unblock their bots where specifically authorised by the blocking admin. I'm sure bot operators that like Magio already encourage admins to block their bots with userpage instructions such as "Administrators: if this bot continues causing harm after receiving a message, please block it or" would rephrase their userpages to encourage admins to use block message such as "OK to unblock yourself when you've debugged". I suggest a reminder from Arbcom that they can and will desysop for self unblocking except for accidental self blocks or where explicitly authorised by the blocking admin.

The issue of minor edits hiding vandalism is real and troubling. I'm probably part guilty, part solution myself. Undoubtedly some of my edits will fix a typo without spotting a vandalism, other times my antennae twitch and as well as fixing the typo I revert longstanding vandalism. Sometimes from many edits back, earlier today from 2013. A compromise here would be for bots doing minor edits other than vandalism reversion to exclude articles edited in the last 48 hours. The real solution however is for us to introduce some form of flagged revisions, not doing so means accepting a higher level of vandalism. I have been considering a compromise proposal on this that I may launch at RFC - as far as I'm aware the relevant debates are far from recent and I think my forthcoming compromise proposal should resolve the main objections. But excluding articles edited in the last 48 hours from this sort of bot would be a useful stopgap. Either would greatly reduce tension between watchlisters and gnomes.

We have too much jargon on this project, it is offputting to newbies. I wouldn't want to stop someone from doing useful edits such as expanding cn to citation needed

@Beyond My Ken: one of your examples corrected 06 to 6. Can I suggest you review your evidence and strike examples of non cosmetic edits?

I endorse the comments made by Anomie, Rich Farmbrough, and BGWhite. Statements like "By 'cosmetic' I mean, as a working definition, 'edits that don't affect what the reader sees'" are troubling, given the points made about accessibility for people using non-visual devices to read Wikipedia. If Arbcom takes this case, I urge them to appoint a neutral expert advisor (or team thereof) to explain such technical matters to them.

And I once again make the point that case proposals such as this one should each have their own pages, so that they can be watchlisted separately. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:29, 1 January 2017 (UTC)