Sectioned discussion is in effect for all case talk pages in this case. Therefore, with the exception of arbitrators and clerks, all editors must create a section for their statement and comment only in their own section.

This case is now closed and pages relating to it may no longer be watched To request an amendment or clarification of an Arbitration decision, see Wikipedia:Arbitration/Requests/Clarification and Amendment. To report a violation of an Arbitration decision, see Wikipedia:Arbitration/Requests/Enforcement. To request the assistance of an arbitration clerk, see Wikipedia talk:Arbitration Committee/Clerks.

Wikipedia Arbitration
About arbitration Talk Arbitration policy (CU/OS) Guide to arbitration Noticeboard +  Talk Search archived proceedings Ban appeals Report
Open proceedings All open requests Case requests Clarification and Amendment requests Arbitrator motion requests Enforcement requests
Active sanctions Contentious topics ( Log) General sanctions Editor sanctions
Arbitration Committee Discuss + Contact Members Procedures Elections Talk History Clerks +  Talk Clerk procedures Conflict of interest reports
Audit Talk Statistics Talk Reports (pre-2016)
Track related changes
v t e

Behaviour on this page: Arbitration case pages exist to assist the Arbitration Committee in arriving at a fair, well-informed decision. You are required to act with appropriate decorum during this case. While grievances must often be aired during a case, you are expected to air them without being rude or hostile, and to respond calmly to allegations against you. Accusations of misbehaviour posted in this case must be proven with clear evidence (and otherwise not made at all). Editors who conduct themselves inappropriately during a case may be sanctioned by an arbitrator, clerk, or functionary, without further warning, by being banned from further participation in the case, or being blocked altogether. Personal attacks against other users, including arbitrators or the clerks, will be met with sanctions. Behavior during a case may also be considered by the committee in arriving at a final decision.

Lets get scope enforcement in early shall we to avoid the FUD.

Wbm's evidence is out of scope as it was defined when opening the case - as it is entirely about the validity of WP:COSMETICBOT it should be excluded. Only in death does duty end ( talk) 11:21, 4 January 2017 (UTC) reply

@ Only in death: This is currently being discussed by the Arbitrators and will be actioned once a decision has been made on wether to exclude this or not. Amortias ( T)( C) 08:43, 5 January 2017 (UTC) reply

Hows that discussion doing? Going well? Evidence has clearly stayed on scope in the meantime hasnt it... Only in death does duty end ( talk) 23:40, 11 January 2017 (UTC) reply

User:Only in death, all in due time. (As it happens, I prodded my fellow arbs tonight with this same question; I understand your impatience.) Drmies ( talk) 04:52, 12 January 2017 (UTC) reply

It's a committee. Although I'm a drafter, I thought it important to get the scope correct, so started a discussion. Doug Weller talk 17:33, 12 January 2017 (UTC) reply

Ramaksoud2000 extension request

I just put my evidence section into a word counter, and it came back to be about 1500 words and less than 50 diffs. Everything was included in the word count, and the diffs include all permanent links, including contribution records and logs. Since my evidence consists only of links, and minimal explanations of the links (no arguments), and considering the fact that this case spans a time period of 7 years and millions of edits, I would like to request an extension from 1000 words to 2000 words, to provide an opportunity to add evidence in case updates are needed. Also consider that there are only two parties to this case, resulting in less total evidence. Thank you, Ramaksoud2000 ^{( Talk to me)} 04:09, 5 January 2017 (UTC) reply

To note I have seen this request but in the interests of consistency and equality I've still placed the edit length template on the evidence page. Amortias ( T)( C) 08:41, 5 January 2017 (UTC) reply

Responses

@ HJ Mitchell:

1. What makes you say that it is Twinkle? There isn't the Twinkle edit summary like his other Twinkle deletions.

2. Even if it was Twinkle, my understanding is that Twinkle needs a list of pages to delete, like there would be at a large TfD. The pages deleted were obviously generated by a bot that scanned and found the pages meeting the specific criteria. What is the difference if a bot generated a list of pages, and he used Twinkle or a different program to delete them all at once?

@ Wbm1058: As I stated numerous times in the evidence section, it is nowhere near "every single instance". That would be impossible. In addition, all the subsections address different issues. Ramaksoud2000 ^{( Talk to me)} 21:30, 10 January 2017 (UTC) reply

Thanks wbm1058. That's embarrassing that I missed that. Pretty obvious. My second point still stands though. Ramaksoud2000 ^{( Talk to me)} 00:02, 11 January 2017 (UTC) reply

@ HJ Mitchell: I did take that care when actually gathering the evidence and putting it on the page. I misremembered the details on the talk page in response to your post, and relied on my memory instead of checking. 13:06, 11 January 2017 (UTC) reply

@ Drmies: Is the context you are looking for Wikipedia:Arbitration/Requests/Case/Magioladitis/Evidence#Magioladitis_unblocks_Yobot_without_the_consent_of_the_blocking_admin_despite_multiple_warnings? Or do you mean context behind each of the 24 blocks? 05:58, 12 January 2017 (UTC)

• Yes, that's the kind of thing that's helpful. We don't need to be exhaustive for the sake of exhaustiveness. If those blocks and unblocks are, in your opinion, evidence of this or that, and sufficient evidence of this or that, then that helps, yes. Thanks. Drmies ( talk) 17:21, 12 January 2017 (UTC) reply
• In response to Magioladitis and Opabinia regalis: After seven years of the same excuses and claims of ignorance, the line has to be drawn somewhere. If he can't keep it together just during this Arb case, then what will happen if this ends without a desysopping to remove AWB access? Ramaksoud2000 ^{( Talk to me)} 01:11, 20 January 2017 (UTC) reply

I'd like to support Ramaksoud2000's request for an extended length limit, since their existing presentation all looks relevant. Magioladitis can get an equal extension in the interest of fairness.

Re Only in Death's request, I'd rather that arbcom not limit the scope of evidence unless someone is trying to support actual interventions outside the case scope (e.g. against editors other than Magioladitis). The issues to be decided may be narrow, but the backdrop and context that should be considered in the decision is quite wide, as shown by the range of issues discussed in the preliminary statements. I'd say let people present whatever they want (within the length limits) in the evidence phase. Then in the post-workshop phase, arbcom can use its own judgment about what evidence warrants deep investigation, and what parts should just be looked at enough to take in the gist.

I plan to add some further comments later about the actual evidence presentations seen so far, and also to post a little bit of evidence of my own. 50.0.136.56 ( talk) 08:36, 7 January 2017 (UTC) reply

I'm reluctant to get too deeply involved here. The case is already starting to get messy and is likely to be a massive timesink and I just don't have the time or the appetite for wikipolitics so my participation is likely to be peripheral.

For now, I just want to correct a mis-statement in Ramaksoud2000's evidence, Magioladitis ran an unapproved adminbot to delete more than one thousand pages. This was a use of Twinkle's mass delete function (available to all admins with Twinkle enabled). It's commonly used by admins when a need arises to delete more than a handful of pages. I don't know anything about the merits of those deletions, but this is not the same thing as running an unapproved admin bot (and requiring an admin to file a BRFA every time multiple pages need to be deleted would tie the project up in needless and interminable bureaucracy). HJ Mitchell | Penny for your thoughts? 13:24, 10 January 2017 (UTC) reply

@ Ramaksoud2000: Dude, check the log you're citing as "evidence"; those log summaries all end in "( TW)". Sure, that can be spoofed, but which is more likely: that Marios wrote his own untested, unapproved program and thought ahe'd give it a whirl for shits and giggles, or that he used a very well-established, tried and tested script that already existed? When giving evidence against another Wikipedian, please take the sort of care you would other to take if they were presenting evidence against you. And if you're not sure about something, ask instead of jumping to conclusions. Marios has plenty of questions to answer, but let's stick to actual misconduct instead of making up imaginary misconduct to make things appear worse than they are. (ETA: there's also no evidence that Marios used a bot to produce the list of targets. They could have come from a category or Special:PrefixIndex or a database report or any number of other places). Thanks, HJ Mitchell | Penny for your thoughts? 08:03, 11 January 2017 (UTC) reply

@ Ramaksoud2000: Per Wikipedia:Arbitration/Guide to arbitration § Brevity, try to keep your evidence concise, direct, and clear. Trying to show every single instance of a given user being a problem may be less useful than picking a few clear and obvious example [ sic] requiring little explanation and presented with minimal commentary. – wbm1058 ( talk) 17:47, 10 January 2017 (UTC) reply

@ Ramaksoud2000: the edit summary on all of those is (redirect from a talk page of dab page to a non-dab page ( TW)) – click on the TW and see where that leads you. wbm1058 ( talk) 23:48, 10 January 2017 (UTC) reply

I am not asking the committee to create new policy by fiat. Wikipedia:Arbitration/Policy#Policy and precedent states The Committee's decisions may interpret existing policy and guidelines. I am only asking the committee to interpret the policy. I've stated clearly how I think it should be interpreted. Interpretation of policy is clearly within scope. Interpreting policy includes resolving conflicts within the policy itself, and assurance that new policies have been implemented in conformance with procedures dictated by old policies. New policies must be "constitutional", so to speak. – wbm1058 ( talk) 23:28, 11 January 2017 (UTC) reply

The idea that ANI is a sufficient "broader venue" for determining support for the policy is laughable. How may who are happy or ambivalent with Magioladitis' editing are going to bother to randomly drive by there to discuss it. That is clearly a venue where editors who are displeased with Magioladitis' edits are going to disproportionately self-select to participate. I certainly haven't been following these discussions nearly as closely as those who are ready to "Farmbrough" him. There has not been a broad community discussion about the tradeoffs between the improvements of minor, incremental edits and the dis-improvements of undetected vandalism. Such a discussion might be valuable if an outcome of it is to redirect software development priorities away from unwanted rocket ships while keeping rather than discarding a productive contributor of incremental content improvements. – wbm1058 ( talk) 00:07, 12 January 2017 (UTC) reply

wbm1058, I'm sorry, but your section is unclear to me. I cannot easily tell what you are giving evidence of, or where that might lead. I'm glad you are suggesting here that you think M. is a "productive contributor", since that also was not clear to me. I'd say your thesis is missing, if you were in my class. What I would particularly like is for you to bring this evidence in clear relation to the case request and what you think the scope ought to be. Thank you. Drmies ( talk) 04:50, 12 January 2017 (UTC) reply

Perhaps it's difficult to include the "thesis" with sufficient evidence to support the thesis because of word-count restrictions on the evidence. I approached this case with an open mind, much as I would expect each member of the committee to. So I did not have a "thesis" going in. You'll note that I did not enter everything at once, but rather over the course of several days, as my research found things (facts) I deemed to be pertinent. I've provided you with facts that I feel are relevant. It's your job to interpret those facts.

Rich Farmbrough's last RfA might be a rough proxy for the issue here. This isn't entirely about one editor, it's about the legitimacy of the Check Wikipedia project. That project is all about making so-called "cosmetic" edits. Basically it seems like where this might lead is to shutting down Check Wikipedia and making it historical. I'll speculate that if there was an RfC on that project, that about two-thirds would support it and one-third would oppose it. Based on how the community split on Farmbrough's last RfA. Of course, it's the one-third who are filing all the complaints about M.'s editing.

Is the rest of the committee also having trouble understanding me?

Ramaksoud2000's section, with its evidence of some 24-odd blocks, is a bit overwhelming. It would take me considerably more time to research and analyze all of them in order to come to any sort of conclusion on the merits. If the committee could rule on the "constitutionality" of COSMETICBOT, and provide us with their interpretation of what it means, then I might wipe my evidence slate clean and start over with responding to each of Ramaksoud2000's charges. wbm1058 ( talk) 13:17, 12 January 2017 (UTC) reply

Thanks, but again, "evidence" always begs the question of "evidence of what?" This is kind of circular: if you need more evidence to support a thesis, then what was the evidence you did give meant to support? Does it not have a point? The legitimacy of the "Check Wikipedia" project wasn't part of the case request, as far as I can tell, though I may have missed something--I was looking here for stuff related to Magioladitis or to COSMETICBOT, but if you don't tell me what your evidence should make me think about either, then I don't really know how to read it. And even after this explanation I'm not quite sure where you stand on COSMETICBOT. You argue there is no broad community support for it as a policy, and you say "There is no guidance prohibiting minor edits", but you also state "Avoiding minor, insignificant, inconsequential edits is a longstanding AWB rule of use"--and I'm thinking that longstanding rules are guidance. But that latter section is a list of diffs/changes that come without explanation as to whether you think those changes are good, bad, community-supported, authoritative, helpful, muddying the waters, applicable to Magioladitis, etc. It is entirely possible that the rest of the committee is less dense than me, of course. Thank you, Drmies ( talk) 16:53, 12 January 2017 (UTC) reply

I didn't approach this with the idea of forming a thesis of my own. I suppose the thesis here is "His bot, Yobot, has been blocked 18 times, mostly for violations of WP:COSMETICBOT" and that this merits some form of sanctions. So, my first step is to look at what COSMETICBOT says, how that came to be, and is there a broad consensus for it? So this is evidence which either supports or doesn't support the validity of that part of the bot policy. I'm just putting the facts out, my opinion is that the basis for COSMETICBOT is weak, but others may have different opinions. I think there is a much stronger basis for the AWB-specific rule to "avoid making insignificant or inconsequential edits". That only applies to AWB and not to bots in general. So, if we assume good faith, Magioladitis tries to avoid making insignificant or inconsequential edits, and his bot is blocked either because his bot makes a "mistake" – not a mistake in the sense of introducing actual errors to articles but in the sense of making cosmetic changes without making any significant changes with the same edit – or because there is a difference of opinion as to whether a specific change is "cosmetic" or not. A bad-faith finding, which is probably the only thing that would merit sanctions, would be intentionally using a bot to introduce actual errors to articles. "Avoid making insignificant or inconsequential edits" is a bit of a grey statement in that it doesn't speak to how hard one should try to avoid making such edits. If one of every 10 edits a bot makes is insignificant, then maybe the operator hasn't tried hard enough to avoid making cosmetic edits. If one of every 200 edits the bot makes are cosmetic, has it tried sufficiently hard to avoid them? I share your desire for some discussion/context of/for the blocks and unblocks, but with the sheer volume of the discussions I've seen on various user talk pages and noticeboards, and it will be tedious to wade through all these and determine the merits of each and form an opinion as to whether he tried hard enough to avoid making cosmetic edits.

It seems that most of the action has moved on to the workshop page, and I'm perhaps as befuddled about what's going on there as you are with this page. wbm1058 ( talk) 18:04, 12 January 2017 (UTC) reply

Some thoughts on the subject, for whatever they're worth: I think it's unlikely that there would be a decision on the "constitutionality" of a policy unless it had been changed very recently, or conflicted with non-negotiable WMF or legal issues, or had been manipulated in a way that reflected foul play. However, it may turn out to be the case that a superficial consensus on a particular issue starts to fall apart when you unpack it, even if that apparent consensus is long-standing. It's fairly clear to me at least that when two people agree about "cosmetic edits", you better ask them what exactly they mean by that. A lot of people are working under subtly different assumptions about what "cosmetic" means, and others misunderstand the motivation behind edits that have no effect for them. ( WP:LISTGAP fixes are the obvious example of edits that are not cosmetic but appear to most editors as if they are - e.g. [1], [2].) Opabinia regalis ( talk) 04:33, 16 January 2017 (UTC) reply

Scope enforcement

We also need scope enforcement on SMcCandlish's section, which clearly veers off the rails into "This policy shouldn't exist" territory. That's well beyond what the arbitration policy even allows the committee to look at, let alone the scope of this case. The committee should be careful not to take the opposition of a few editors who disagree with current policy as a larger controversy over what the policy currently means. Past discussions at broader venues, such as ANI, have made clear that the community at large is supportive of COSMETICBOT. See, for instance, the final ANI on Magioladitis linked in the prior dispute resolution section of the original case request (also linked in my last section of evidence). Also, the fact that no-one has attempted to start a discussion to remove the section in the over half a decade it has been part of our bot policy. ~ Rob₁₃ ^Talk 18:43, 11 January 2017 (UTC) reply

BU Rob13 extension request

My evidence is around 800 words total. I've been careful to focus on things that aren't covered in other evidence sections and save the analysis for the workshopping phase, but I think all of the things in my section will be helpful to the committee. It's all clearly within scope and I've limited my predisposition for wordiness. While I'm not a party to this case, an extension to 1,000 words/100 diffs (the typical party evidence limit) would be appreciated to allow me to retain all the unique points I've made. ~ Rob₁₃ ^Talk 20:44, 11 January 2017 (UTC) reply

Restricting tools for certain types of edits

Some editors have stated that it makes little sense to allow edits but restrict how they can be made, so the AWB rule of use #4 and COSMETICBOT should be ignored for all edits that are in-line with guidelines and policies by default. There's actually a lot of precedent for saying that edits are okay, but implementing them with a certain high-speed/volume tool is not. If I were to nominate 1,000 templates in a day at WP:TFD, it would be disruptive bludgeoning of the process, even if each nomination was valid. If I spread them out over a longer period of time, there's no issue. Similarly, restricting cosmetic or minor edits to manual editing has a lot of merit to prevent spamming of watchlists, high use of server resources, and pollution of article histories with incredibly minor edits. Often, the hope with a minor MOS issue is that the "fix" will come as part of larger edits to the page. Again, this is not at all the venue to debate COSMETICBOT or AWB rule of use #4, but I wanted to point out that it isn't insane for an edit to be fine but the manner in which it is made to not be. Other examples of "good" edits that are disallowed due to the circumstances in which they're made include running an unapproved bot to make unproblematic edits or making "good" edits while blocked. ~ Rob₁₃ ^Talk 20:09, 13 January 2017 (UTC) reply

Just thinking out loud, as it's obviously not arbcom's job to decide these things, but one could just as easily make the opposite argument about "cosmetic edits" - deliberately bundling them with other, more substantial changes makes both the "substance" and the "cosmetic" aspects harder to review and audit, dirties the diff, and results in the unnecessary perpetuation of confusing code clutter. Wikipedia isn't a software project, but since this case started I've been trying to imagine how a large project would go if the formatting conventions were "everybody can use whatever combination of tabs and spaces they want, and you're only allowed to change the spacing if you're also making some other change at the same time". That would be crazy ;) Opabinia regalis ( talk) 22:20, 13 January 2017 (UTC) reply

I've always interpreted COSMETICBOT to mean that the community doesn't care too much about cleaning up wikicode vs. making the actual pages look proper. I'm not even saying the edits shouldn't be made by themselves, just not at high speeds with automated or semi-automated tools, based on our current policies and rules. Another similar situation is WP:CANVASS, which states that even allowable notices cannot be made via bots. ~ Rob₁₃ ^Talk 23:11, 13 January 2017 (UTC) reply

Well, these are comparable in the sense that they're a bit of social engineering based on the assumption that people's natural intolerance for tedium will keep edit volumes reasonable. I'm not sure they're too similar otherwise, though - what makes a good notification is entirely determined by what the community wants to be notified about, and what makes a good mainspace edit is determined by what benefits readers and facilitates future editing. Opabinia regalis ( talk) 20:38, 15 January 2017 (UTC) reply

Reply to Izno

@ Izno: There's an unprecedented alternative to desysopping to remove AWB access of an admin. See my workshop section on enforcement. Also, can confirm, had I come across a non-admin with the behavioral pattern of Magioladitis, AWB access would have been removed many times over. ~ Rob₁₃ ^Talk

Magioladitis continuing to violate AWB rule of use #4

@ Opabinia regalis: Sadly, I need to report the following 45 diffs out of the past 50 from the Magioladitis account as of this time. I've included the URLs in a collapsed section below. All of these diffs solely alter wikicode without producing any rendered changes (clearly cosmetic-only/minor in the sense of AWB rule of use #4) or move around only whitespace (something that is both minor in the sense of AWB rule of use #4 and specifically listed as an example of a cosmetic-only task at WP:COSMETICBOT, in the context of cosmetic_changes.py's "removeUselessSpaces" task). Magioladitis has stated on his talk page that "This is CHECKWIKI APPROVED TASK.Jonesey95 I thought I had only to stop during the evidence phase. OK. I'll keep it till the end of the ArbCom compltetelly." This demonstrates clearly that he intends to continue these edits beyond the end of this case in the absence of restrictions. I'd like permission to include this in evidence, as it's a substantial new development. This is a staggering 90% error rate, and evidently, it will continue in the absence of restrictions. ~ Rob₁₃ ^Talk 17:58, 19 January 2017 (UTC) reply

Pinging Kirill Lokshin and Doug Weller as the drafters - I think historically events after the close of the evidence phase aren't added to evidence, but just noted somewhere on the talk page.

Magioladitis, can you explain what's going on? Two things: 1) why are you editing at high speed on your main account? 2) why are you still making edits that are "cosmetic"? Are you not reviewing them, or are you unclear on what kinds of edits people are objecting to? Opabinia regalis ( talk) 23:52, 19 January 2017 (UTC) reply

@ Opabinia regalis: I believe this is Magioladitis' response to my posting, which I think says more than I possibly could. ~ Rob₁₃ ^Talk 00:15, 20 January 2017 (UTC) reply

@ Opabinia regalis: I also pointed out this issue in my section below. There is some information on Magioladitis' talk page, as well. — Carl ( CBM ·  talk) 00:05, 20 January 2017 (UTC) reply

@ Opabinia regalis: Please read the discussion with CBM. In my understanding I ceased the CHECKWIKI fixes to help the Evidence phase. I fixed one particular CHECKWIKI error with a very specific edit summary I stopped since it was asked. Yes, for me it's not clear what edits are people objecting to. It's also not clear to me that there is consensus not ti fix minor potential errors. Rob claims 90% error rate. I claim 100% success rate because I fixed all pages listed in CHECKWIKI 001. The list of diffs could be shown by a single link. (RU Rob13 is not familiar with that?) 8 edits per minute is not a high editing ratio for me. so, I repeat since Jonesey05 made it clear the editing should be limited for the entire ArbCom procedure I stop. I also underlie the fact that no evidence why this series of edits caused any problem was brought. -- Magioladitis ( talk) 01:00, 20 January 2017 (UTC) reply

Admittedly, when I began compiling a URL list, I expected to be showing 10ish edits among 50, not the unusually high number of 45, so perhaps I should have just linked to your contributions. This cosmetic-only edit issue is not difficult. AWB rule of use #4 states that you must not make edits which don't change the rendered output of the page without specific consensus that you can use AWB to make those edits. Even if you have bot approval to do this, and I would say you do not, because you just slapped the label CHECKWIKI on this particular fix at a later date and said it was approved without seeking approval for the expanded task, that wouldn't impact your main account. There may be consensus that these changes are appropriate, but there's no consensus that the fixes should be automated and made systematically without being accompanied by non-cosmetic changes. Further, the onus is on you to clearly demonstrate such consensus, as AWB rules of use #4 clearly states you must stop when an editor disagrees with your editing and demonstrate consensus to make the edits with AWB. Failing to make any rendered change to the page is an error, and that error occurred at a 90% rate. ( edit conflict) The problem is that you still won't abide by community consensus that these edits are not acceptable. The problem is that you treat the AWB rules of use with flippant disdain. ~ Rob₁₃ ^Talk 01:08, 20 January 2017 (UTC) reply

The point which Magioladitis has misunderstood is that CHECKWIKI is simply a small group of a few editors, which cannot overrule the general bot policy or the AWB rules of use just by listing something as a "CHECKWIKI fix". Separately, Yobot did not have permission to fix this particular error (possible because it was too clearly cosmetic for bot approval). Actually, this is one of a small number of the CHECKWIKI items which is purely cosmetic; this is why there would not be too many complaints about the CHECKWIKI items, because most of them are objectively worth fixing and non-cosmetic. — Carl ( CBM ·  talk) 01:15, 20 January 2017 (UTC) reply

@ Magioladitis: Yes, you fixed all the pages that had that specific error. However, that specific error in isolation meets the most-common definition of "cosmetic" and therefore shouldn't be fixed in isolation.

Taking a step back a bit, I've said this on the mailing list discussion about this case and I'll say it here too: I, personally, not speaking for the committee and not in a position to make changes to the policy in any case, think COSMETICBOT is a really dumb way to do things. The main reason for it is as a workaround to an old, equally dumb bug. (Why aren't there more comments on T11790?) Most large change-managed environments would not encourage the contamination of substantive edits with a bunch of secondary unrelated changes, or prohibit systematic fixing of syntax errors and non-conforming code styles. If one of my coworkers started making commits like that, I'd start sending him emails like "Hey Joe, just wanted to make sure you're up to date on next week's schedule! The regular meeting of the Office Ergonomics Committee will take place at 10am on Tuesday, the campus gym will be closed from 5pm Tuesday to 12pm Wednesday for maintenance, your big presentation to the site visitors from our funding agency has been moved to 8am on Monday, there will be a Q&A with the head of the Parking Department on Thursday at 5pm on the subject of the resurfacing project in the east lot, and participants in Saturday's Student Doctor Association 5K Fun Run are reminded to pick up their packets from the Student Union building by Friday at 1pm."

That being said, thinking a policy is really dumb doesn't mean we get to ignore it, and the current status quo is that you can't make those edits if you aren't making some other, substantive change at the same time. Probably the Checkwiki project should, at minimum, start sorting its error types into non-cosmetic and cosmetic changes so that it's obvious which ones can be done by themselves. (As for the edit summary, I think a link to the definition of the error type would be more useful than a link to the dump, but that's a side point.) As I said somewhere else on the case pages, it seems to me that a lot of underlying issue here is that there's no effective change management structure for Checkwiki fixes (or AWB's general fixes) such that the community could spot potential problems. I'm not convinced a full-on positive consensus is needed for every change, but a "post a new change for a week and if no objections come in, do a test run before letting it loose in the wild" process seems like a smart idea. Opabinia regalis ( talk) 23:32, 20 January 2017 (UTC) reply

Opabinia regalis I like your last idea. In fact if the community was more active n CHECKWIKI maybe there were not these problems now. We have people complaining about some CHECKWIKI tasks but unfortunatelly not helping to improve them. We changed a lot of the tasks thanks to community's feedback. would be more than happy with a CHECKWIKI re-evaluation. Recall that we spend years (really!) to adjust the lists because in fact the job is done by only 4 people when it comes to the implementation part. I also wrote that WPCleaner will soon become more appropriate to use than AWB. This is the reason I ask other willing editors to take over. These facts were ignored and there is an attempt that I take responsibility from CHECKWIKI list to AWB's code and Yobot's operation. I am a single person. -- Magioladitis ( talk) 23:43, 20 January 2017 (UTC) reply

The community oversight you'd like could be provided if you acknowledged that your initial Task 16 approval was for a specific set of tasks and file a new BRFA when you want to expand that set of tasks. I'd be happy to participate and examine whether a given fix is cosmetic or not if you would submit to that type of community scrutiny. So far, you've been incredibly against that, claiming that Task 16, which listed out 40 or so specific tasks, is carte blanche to perform any task that a handful of editors slap the "CHECKWIKI" tag onto. ~ Rob₁₃ ^Talk 00:03, 21 January 2017 (UTC) reply

It's obvious that we disagree. I think you should re-read the Function overview and the Function details. The discussion also says that this is a list "possible fixes" this means the list is not limited. You are from Canada if I am not mistaken so you are native English speaker. -- Magioladitis ( talk) 12:56, 21 January 2017 (UTC) reply

Am I right that the sequence of events is, roughly: 1) Yobot is approved for general "checkwiki fixes"; 2) new checkwiki items are added that entail "cosmetic" edits; 3) Yobot makes edits based on these new "cosmetic" fixes, creating a situation in which Magioladitis believes he has approval to make edits (because they're checkwiki fixes) that others believe he is prohibited from making (because they're cosmetic)?

Opabinia regalis ( talk) 05:48, 23 January 2017 (UTC) reply

user:Opabinia regalis, no. Yobot was never approved for that. Read through Wikipedia:Bots/Requests for approval/Yobot 16. Only specific CHECKWIKI errors were approved. Magioladitis tried to get a general approval for all CHECKWIKI fixes, but was explicitly denied due to cosmetic editing concerns, and he acknowledged that. Ramaksoud2000 ^{( Talk to me)} 06:08, 23 January 2017 (UTC) reply

user:Opabinia regalis, mostly yes. Yobot was approved to perform CHECKWIKI fixes. Read through Wikipedia:Bots/Requests for approval/Yobot 16. I provided a sample list which I later sumbitted to the community via the CHECKWIKI page and the list was updated forming the list that we see today. The cosmetic editing concerns refer to edits outside the CHECKIKI part. It refers to the "general fixes part". Ramaksoud2000 got it wrong. -- Magioladitis ( talk) 08:40, 23 January 2017 (UTC) reply

Can you link the post you made on the checkwiki page? I guess my earlier post should read "1) Yobot is approved for specific checkwiki tasks; 2) there are changes to the nature of those tasks; 3) same result". It sounds like a lot of people are not aware of, or don't want to follow, the checkwiki page, and maybe the project needs to find another way to communicate with the community about updates and get feedback before deploying significant changes.

On the other question, isn't the list Rob gave in the first post of this section a list of edits made for checkwiki purposes that resulted in cosmetic edits? Opabinia regalis ( talk) 20:34, 23 January 2017 (UTC) reply

User:Opabinia regalis, even if you amend your post to that, you must understand that Yobot was never approved for any kind of redirect bypassing, and under no circumstances may anyone run bots from a non-bot account. This case is not at all about fine distinctions and clarifications of BAG approvals. If it was, then WT:BRFA would be a much more appropriate place to get those clarifications. Ramaksoud2000 ^{( Talk to me)} 21:21, 23 January 2017 (UTC) reply

Yobot never did redirect bypassing on purpose. Evidence prove that. They also prove this. I removed many of these so they are not done by bots nor humans. -- Magioladitis ( talk) 21:28, 23 January 2017 (UTC) reply

@ Opabinia regalis: Regarding the "really dumb way to do things", I think that it is better to view COSMETICBOT as a different kind of compromise. The issue is not really with whether an edit that needs to be done is better done separately or with other edits. Edits such as [3] really do not need to be made at all. However, a relatively small number of bot operators have a strong desire to make such edits. There was a compromise via COSMETICBOT: if there is some other more significant edit happening anyway, then a bot can go ahead and make the insignificant edit at the same time. But this should not suggest that the insignificant edit is equally important with the other one. It's like the way that some airline employees can fly for free if there is an open seat: this does not mean the airline is choosing whether to fly them on an existing flight or to fly them on their own flight, it only means they are tolerated if there is an existing flight that they can take. The reason that insignificant edits are not allowed to be done separately is that, in the end, there is no real need to do them at all. — Carl ( CBM ·  talk) 16:38, 21 January 2017 (UTC) reply

Lots of edits don't "need" to be made. These edits primarily benefit future consumers of the wikitext. Others primarily benefit other small groups, like screen reader users, or downstream reusers of the citation template metadata. I don't think dismissing the first motivation is particularly useful, since obviously some people think these edits are beneficial. Opabinia regalis ( talk) 05:48, 23 January 2017 (UTC) reply

I concur entirely with BU Rob13 immediately above me. SMcCandlish's section is a call for Arbcom to grant itself the power to impose direct rule on Wikipedia—this is not only totally outside the scope of this case, it's grossly against the fundamental principles of Wikipedia (to the extent that I'd lobby for, and feel confident of getting, the removal from office of any arb who supported it). ‑  Iridescent 23:37, 11 January 2017 (UTC) reply

Not a drafter, so this is plain old kibitzing, but: SMC himself acknowledges that that section doesn't exactly meet the definition of "evidence", and would likely acknowledge that the position he's advocating is unlikely to attract support. Speaking for myself, if someone objects to material submitted as evidence, I have to go read it to see why. Then, once it's become "something the arbs read in connection with the case", it seems more transparent to leave it there for the historical record than to remove it, unless it goes beyond "irrelevant" or "implausible" into "actively harmful". Pinging SMcCandlish to be sure he's seen this - and yes, that last section is really more commentary and advocacy than "evidence", and would better be placed on the talk page. Opabinia regalis ( talk) 21:58, 13 January 2017 (UTC) reply

Notes to peeps above:

• @ Iridescent: I'm not sure if you mean my entire section or the meta-commentary subsection. As for "a call for Arbcom to grant itself the power to [yadda yadda] ... removal from office of [etc.]": That's just melodramatic fiction on every level ("WikiOpera"?) and has even less pertinence to the case than anything I've said. I've certainly made no such call. My comment to Rob below clarifies this.
• @ BU Rob13: I have not argued that any policy should not exist, only that as presently formulated, its wording does not actually represent practice, conflicts with policies that have a much greater WP:CONLEVEL, can trigger legitimate IAR any time it gets in the way of improving the encyclopedia, and has a dubious consensus record. It's definitely within ArbCom's remit to observe when one policy conflicts with another. In particular to this case, I'm objecting to punitive enforcement action being taken against Magioladitis on the basis of that bot policy section, in which the key term does not even have a definition; where it even approaches giving examples of what it's supposed to mean, it fails by weasel-wording with constructions like "some AWB fixes" without specifying which. It's like "Commandment 11: Thou shalt not be naughty, which sometimes means being annoying on Tuesdays and sometimes doesn't." I was not the first to point these problems out, wbm1058 was. He just took a historical approach to it, while I'm doing some analysis of the import of it.
• @ Opabinia regalis: I have no objection to the meta section of my comments, or the entire thing, being refactored to the talk page if Arbs think it is too side-band. Having already injected into the debate some distinctions between "minor", "trivial", "cosmetic", and "reader-invisible", that is probably sufficient for my principal purpose, which was to pry part their fallacious conflation which has had a negative effect on the reasoning I've seen presented in several places in this case.

 —  SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  04:49, 14 January 2017 (UTC) reply

@ SMcCandlish: In what you describe as "expert testimony", you've listed what you consider to be the definitions of "trivial edit", "cosmetic edit", etc. But one of the examples under "cosmetic" is repeatedly fixing a typo. Many others commenting on this case have been using a working definition of "cosmetic" as "change to the wikitext that doesn't change the rendered page", which of course would exclude edits like typo fixes. (Of course this is the usual Wikipedian word abuse - "cosmetic" here means "something that doesn't affect how it looks" ;) Opabinia regalis ( talk) 20:21, 15 January 2017 (UTC) reply

@ Opabinia regalis: I think this clearly illustrates the problem with undefined prohibitions using ambiguous language and an "it means whatever I want it to mean" approach. Even if we moved typo fixes out of that segment, I think my analysis would still hold.  —  SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  08:15, 16 January 2017 (UTC) reply

PS: I replaced the example with something reader invisible but useful to do, fixing invalid </br> markup to the correct <br />; another would be normalizing citation template parameters to replace deprecated ones. And so on. We do this kind of stuff all the time that doesn't directly affect the article text in the reader's eyeballs. —  SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  08:23, 16 January 2017 (UTC) reply

@ CBM: "[I]n some cases a small group decides on their own (with no broad consensus) that some parameter needs to be changed, without seeking broader consensus. Unfortunately, AWB and other tools can make this very easy to do." That doesn't seem relevant, since it's not a minor edit, a trivial one, a cosmetic one, or good one that happens to be invisible to readers, but simply making likely-to-be-controversial changes without consensus, and if pursued on a mass scale is a WP:FAITACCOMPLI problem. Just because it's "edits I question" + "automation was used" doesn't make it parallel. As for the br code fix, there's a general consensus that it's better to fix such errors, because 99+% of WP editors are not HTML experts, and they learn technical markup here by observation. Errors like that, if left in place, inspire more errors of a similar sort that the parser does not auto-catch. That said, I agree it's not the best example, and so will replace it. I don't have a pro or con position one whether YoBot's scope was exceeded, only with what sort of scope limitations, for what rationales, are being asserted as a general matter, and whether they make sense, are properly understood, or even have consensus.  —  SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  17:13, 16 January 2017 (UTC) reply

@ CBM: Agree http → https is a good example. As for the distinction you're drawing between the two bots, I'll just take your word for it.  —  SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  17:35, 16 January 2017 (UTC) reply

Seeing as BU Rob13 is saying that changing template redirects to direct transclusions is "obviously cosmetic-only": Not sure how well it translates to other pages, but I notice that https://en.wikipedia.org/?title=Wikipedia:Sandbox&oldid=760027571 and https://en.wikipedia.org/?title=Wikipedia:Sandbox&oldid=760027419 (same page content, except for an additional underscore and the former using {{ Collapsetop}} redirect while the latter uses {{ Collapse_top}} directly) takes longer to parse and load in the NewPP report indicated in the HTML sourcecode, by as much as 7 milliseconds. It's a small difference (on longer pages with more templates, it may be a much bigger one) but perhaps worth noting as to potential effects this edit has that aren't merely cosmetic. Jo-Jo Eumerus ( talk, contributions) 15:11, 14 January 2017 (UTC) reply

Izno's comments are only about my editing and not about bot's edits. Moreover, it is an assumption. -- Magioladitis ( talk) 08:21, 15 January 2017 (UTC) reply

It seems that everybody agrees that there is a gray zone of edits that can't be defined as cosmetic or not cosmetic by default. -- Magioladitis ( talk) 18:08, 15 January 2017 (UTC) reply

One of the evidence Carl brought is that I deleted templates that were decided to be deleted via TfD.

One of the evidence about the section header naming: If someone looks closer they 'll see that from main account I also fixed name like "Weblinks" which were not fixed the bot exactly to avoid problems. Moreover, I fail to see what the problem is exactly since my editing was flawless. -- Magioladitis ( talk) 22:54, 16 January 2017 (UTC) reply

I claim that every time my bot was blocked or even stopped I did some fix in the bot/AWB code that fixed the problem in the given series of pages. -- Magioladitis ( talk) 22:56, 16 January 2017 (UTC) reply

I don't read any evidence that Yobot actually hid vandalism. -- Magioladitis ( talk) 13:14, 17 January 2017 (UTC) reply

The evidence phase is completed. -- Magioladitis ( talk) 18:19, 19 January 2017 (UTC) reply

And as everyone can see more editors agreed that edits that move pages in/out of tracking categories are not cosmetic. Thanks Martin for bringing this up. I've been saying since 2010 and now there is a discussion in 2016 that proves it. It's clear that some people have lack of understanding of the term "cosmetic edit" which does not mean only that the rendered output is the same. Thanks, Magioladitis ( talk) 13:37, 23 January 2017 (UTC) reply

Btw, this BRFA is a clear example where the bot trial revealed a minor error in the script which was instantly fixed. Perfect. -- Magioladitis ( talk) 13:39, 23 January 2017 (UTC) reply

Magioladitis extension request

I would like to add to the Evidence the Wikipedia:Bots/Requests for approval/Yobot 25 which Martin just brought out. It's a clear example where: It is shown that edits that add/remove tracking categories are not considered cosmetic, that I clearly explained by position on the matter, the BRFA was approved after a minor error that was causing minor changes was fixed. -- Magioladitis ( talk) 13:41, 23 January 2017 (UTC) reply

New evidence come that RU Rob13 has not exactly understood the COSMETICBOT policy even in the way written today. Maybe we should alow more evidence to be added? -- Magioladitis ( talk) 13:50, 23 January 2017 (UTC) reply

New evidence show that community was aware of the CHECKWIKI tasks and the task are done for years from multiple bots. -- Magioladitis ( talk)

Meanwhile, other bots in trouble

• Two people complained that FrescoBot is doing "cosmetic changes". User_talk:Basilicofresco#Cosmetic_only_changes. I was the one to reply to both.
• A complain for Cydebot running unapproved task: Wikipedia:Bot_owners'_noticeboard#Cydebot_running_unapproved_task
  • In 4/19/14, Cydebot created a thousand of duplicated categories while trying to merge categories
• BG19bot's tasks in question: User_talk:Bgwhite#BG19bot_BRFA
• Dexbot accused of violating COSMETICBOT User_talk:Ladsgroup#Cosmetic_edits.

So, three active bots all received complains the last 2 months. Complains are regularity for bot owners. All complains are similar to Yobot's complains. -- Magioladitis ( talk) 21:55, 30 January 2017 (UTC) reply

Yobot unblocked

Yobot was unblocked. -- Magioladitis ( talk) 21:25, 1 February 2017 (UTC) reply

So everyone is aware, a question has been posed on the phab ticket about the best solution to the watchlist issue. People who are interested in this problem might want to weigh in: phab: T11790. Opabinia regalis ( talk) 20:12, 15 January 2017 (UTC) reply

As I remarked elsewhere there is a gadget that will do this, though I have forgot the name entire. All the best: Rich  Farmbrough, 00:39, 16 January 2017 (UTC). reply

I believe there are several options for workarounds - for example, using the enhanced watchlist eliminates it - but most don't quite fit the use case of people with very large watchlists who are using them primarily for countervandalism. Opabinia regalis ( talk) 03:34, 16 January 2017 (UTC) reply

Apologies - I missed the "sectioned discussion" notice. — Carl ( CBM ·  talk) 00:47, 16 January 2017 (UTC) reply

Comment on Opabinia's section

Thanks. That would resolve the concern about bot edits may hiding vandalism. There is another concern, though. Bot edits, even if correct, still get reviewed by editors watching the pages. This is particularly true then the bot's edit summary implicitly says there was an error - people naturally want to see what the error was. So every edit that a bot makes has a cost in human time. If there is a significant improvement to the page, that cost can be justified, but if the edit just pushes around whitespace or changes a template redirect, the cost isn't worth it. Add in the fact that most of the MOS is only general guidance, not mandatory, and the minor bot edits related to things like CHECKWIKI may also have no policy basis. — Carl ( CBM ·  talk) 20:26, 15 January 2017 (UTC) reply

An example of this was reference footnote rearrangement. Based on the preferences of a few AWB users, AWB began to rearrange footnotes to put them in numerical order, without any support from the MOS or other policy. It took years for this to finally be disabled in AWB bots, although it was never approved as a bot task [4]. — Carl ( CBM ·  talk) 20:26, 15 January 2017 (UTC) reply

Comment on section of Jo-Jo Eumerus

A long established aspect of Wikipedia is Wikipedia:Don't_worry_about_performance. The Mediawiki developers and site operators take responsibility for performance. Ordinary editors do not need to worry about it, and bot operators in particular do not need to make edits based on some perception of performance. This is particularly because the heavy caching on the site, which is not always visible to logged-in editors, reduces the number of times each page has to be rendered for non-logged-in readers, who make up the majority of hits. — Carl ( CBM ·  talk) 18:12, 15 January 2017 (UTC) reply

Reply to section of SMcCandlish

BR tags Looking at the wiki code and generated HTML of [5] shows that Mediawiki silently replaces both "</br>" and "<br/>" with "<br />". So there is no actual benefit in fixing it in the wikicode, particularly as a solo edit. Many editors and bot operators are not aware that even when pseudo-HTML is put into wikicode, it is still cleaned up by Mediawiki before actual HTML is generated. So there is no immediate benefit in "cleaning" these sorts of things.

Deprecated template parameters are a better example. If there is consensus a particular parameter needs to be changed, it is easy for a bot to go through and change them all at once. However, in some cases a small group decides on their own (with no broad consensus) that some parameter needs to be changed, without seeking broader consensus. Unfortunately, AWB and other tools can make this very easy to do. It would even be possible for an editor to make changes that are directly opposite popular consensus and/or the MOS. This is one reason why it is key for bots to only perform tasks for which they are specifically authorized, and for BAG to ensure there is broad consensus for each task.

Getting back to this case: the problem with Yobot is that it has routinely exceeded its authorizations, or has relied on vague bot requests such as #16 (the CHECKWIKI one) which could be used to justify almost any edit, regardless of consensus. — Carl ( CBM ·  talk) 12:50, 16 January 2017 (UTC) reply

@ SMcCandlish: I think a good example of positive bot editing with AWB is Bender the Bot, which has been replacing "http" with "https" for a while now. I have seen dozens or hundreds of such edits on my watchlist, and I have no complaints. The bot edits are simple to review [6] and the edit summary is clear. Unfortunately, Yobot is more or less the opposite in terms of its edits. This is entirely because a different approach from the bot operator. — Carl ( CBM ·  talk) 17:25, 16 January 2017 (UTC) reply

On the purported vagueness of COSMETICBOT

Some editors have claimed that the COSMETICBOT rule is vague. However, there is a simple solution: an explicitly approved bot request (BRFA) will override COSMETICBOT, by everyone's understanding of how bot requests already work. This is how the system was designed to work: BAG looks for evidence of consensus for some particular kind of edit, and once they believe there is sufficient consensus they approve the bot job.

In this case, the issue is that Yobot does not have bot approval to make various kinds of cosmetic edits, but has continued to make them for a number of years. It would be ideal, actually, for Magioladitis to put them in as a specific bot request, so that it could be denied on the record. — Carl ( CBM ·  talk) 13:02, 16 January 2017 (UTC) reply

January 19 request to add additional evidence

I would like to request the ability to add the following paragraph to my evidence section. It is about a sequence of bot-like edits made on Magioladitis' main account today, after the evidence section had closed. — Carl ( CBM ·  talk) 17:02, 19 January 2017 (UTC) reply

Bot-like jobs on main account; insignificant edits On January 19, Magioladitis performed a sequence of 98 edits between 16:21:53 and 16:33:45 [7]. The edits appear to perform the CHECKWIKI error fix job of Yobot, which was blocked at the time. The edit rate was over 8 edits per minute on average. The content of the edits was to remove "Template" from template transculsions, as in [8]. This change has no effect on the page rendering, and appears to violate the AWB rules of use #4.

@ L235: pinging a clerk.

@ CBM: We're still polling the drafters on this, but it does not look likely that the evidence will be formally entered. It is stressed, though, that the arbitrators base their decisions on the totality of the information available to them, not only what happens to be formally stated on the evidence page. Thanks, Kevin (aka L235 · t · c) 03:38, 20 January 2017 (UTC) reply

I understand. Thanks for the response. — Carl ( CBM ·  talk) 11:56, 20 January 2017 (UTC) reply

@ CBM: Read and noted, no need to place it elsewhere. Doug Weller talk 13:16, 20 January 2017 (UTC) reply

It is far from clear that these violate rule 4 which is relatively subtle as written, and even if they do, rule 4 has no standing, even as policy or guideline. All the best: Rich  Farmbrough, 17:50, 4 February 2017 (UTC). reply

@ Doug Weller:, @ Magioladitis: FYI. All the best: Rich  Farmbrough, 23:21, 9 February 2017 (UTC). reply

@ Od Mishehu, Magioladitis, and BU Rob13: My point is not so much the evidence I've presented as that: other administrators have been forced to deal with Magioladitis's use of AWB as if he were an administrator--because he is. Especially, it is damaging to the trust of other users on the project when administrative actions (regardless of policy) are used in an WP:INVOLVED fashion--in this case, WP:WHEEL is relevant. While none of the administrators in this case made administrative actions that were overturned by the user in question, and subsequently re-instated by those administrators, there is a significant damping effect that WHEEL places on administrators who are seeking to deter further (mis)use of a tool such as AWB when interacting with another administrator who is willing to unblock his, or one of his, accounts, or restore permissions by e.g. re-adding his or his account's name to the CheckPage. -- Izno ( talk) 12:47, 17 January 2017 (UTC) reply

Hello parties and others, the evidence phase for this case is closing at 23:59 (UTC) today. Please submit evidence before then. After the evidence phase is closed, your contributions at the workshop are still valued. Thanks, Kevin (aka L235 · t · c) 21:56, 17 January 2017 (UTC) reply

I did not provide any evidence earlier, mainly because other editors had already written very effectively what I could add. I will just link to Wikipedia:Bots/Requests for approval/Yobot 25 because I think this is instructive as to the general issues presented here. Firstly the failure to disclose what kind of general fixes the bot would make, the apparent or wilful lack of understanding of the term "cosmetic edit", and his reaction to User:intgr when errors were identified in the trial. Most of the problems with Magioladitis/Yobot can be seen in that single BRFA. I believe that BAG are/were a little too willing to wave through Magioladitis's tasks in the past, perhaps because he was a member of BAG. — Martin ( MSGJ ·  talk) 13:18, 23 January 2017 (UTC) reply