Wikipedia talk:Bot policy

How to ask for permission:

If you want to run a bot on the English Wikipedia, please follow the policy at Wikipedia:Bots and WP:BRFA.

How to file a complaint:

If a bot has made a mistake, the best thing to do is to leave a message on its talk page and/or that of its owner. For bots on their initial one-week probation, please also leave a note on this page.

If a bot seems to be out of control, ask an administrator to block it temporarily, use WP:ANI for an immediate block, and make a post below.

Authorized bots do not show up in recent changes. If you feel a bot is controversial and that its edits need to be seen in recent changes, please ask a Bureaucrat for the bot flag to be removed.

Archives
Archive 1	Archive 2
Archive 3	Archive 4
Archive 5	Archive 6
Archive 7	Archive 8
Archive 9	Archive 10
Archive 11	Archive 12
Archive 13	Archive 14
Archive 15	Archive 16
Archive 17	Archive 18
Control proposals
Archive policy
Archive interwiki (also some approvals for interwiki bots)

ImageRemovalBot problem

I have been noticing some trouble with the way ImageRemovalBot handles the deletion of gallery images. Many of the images that ImageRemovalBot removes are music-related. Some of these images are album covers that have been placed in galleries on artist's pages. When ImageRemovalBot removes the image, it also removes the name, record label, release date, and chart information along with the image. In some cases, it has removed several or all albums released by a band. Here are some examples:

• Starsailor's entire discography was missing when I checked their page.
• So was The Velvet Teen's - with no albums listed, this article would then be a candidate for deletion!
• What good is a Replacements discography, if it doesn't even mention their first two albums?
• The Bob Schneider album listing was made into a complete mess.

I had a spirited discussion with the bot's owner, Carnildo, about fixing the problem. I suggested either modifying the bot to change images in galleries to some sort of default "no image" image, or simply stopping removing redlinked images from galleries (better to log them and remove by hand than have the bot removing content along with them).

Carnildo and I do not see eye to eye on the issue, so I am posting here because I think this bot is no longer harmless per WP:BOT. Chubbles 17:39, 10 August 2007 (UTC)

I don't believe there is a perfect solution -- the question is, which option is less bad? Option #1: the bot leaves missing images in galleries (or replaces with "no image"). Option #2, the bot removes them, as it does currently. Both options have pluses and minuses, but I think option #2 is the better option. I'll explain. The only reason to prefer option #1 is that the bot occasionally removes encyclopedic material that isn't anywhere else in the article. That's a legitimate problem, but the counter-arguments are strong: (1) Encyclopedic material shouldn't be only in captions. (2) This problem could apply equally well to thumbnail images, if their captions have material that isn't available elsewhere. It's more common for gallery images, but the if we leave empty images in galleries because the captions might have useful info, wouldn't we also need to leave thumbed images for the same reason? (3) Leaving a missing image (or a "no image" placeholder) practically begs newbies to put the covers back in, making the whole problem worse. (4) We should be discouraging the "format" of listing discography info in gallery captions, not working around this obviously incorrect practice. (5) Empty images (or filler images) in galleries look messy; note that the "complete mess" of the Bob Schneider article referenced above is that the empty images weren't yet removed -- that mess would be exacerbated by the change Chubbles recommends. (6) The info is still available in an HTML comment, if a person wants to manually put the information in the correct place. For all these reasons, I don't thing Chubbles' recommended change would be beneficial. – Quadell ^{(talk) (random)} 20:19, 10 August 2007 (UTC)

It is not clear to most editors why gallery format for discographies is discouraged by Wikipedia policy. (That's why thousands of band pages have discographies put in this format.) It is all well and good to nudge people in that direction - but not at the cost of removing important information; and the albums recorded by notable artists are arguably the most important information about them that can be provided. There is also the problem that, by removing the names of albums, some of these pages will become candidates for deletion. If you remove one album from a page of a band who has released two albums on an important label, they may appear to no longer meet the criteria for WP:MUSIC - and having spent enough time at AfD to know, it may well be that no one notices until it's too late.

I think a better way to encourage formatting of discographies as galleries could be to create a warning tag - say, {{baddiscography}} or something - which says "this discography should be reformatted according to Wikipedia policy" or something. After all, ImageRemovalBot's way of doing things isn't helping keep galleries of discography pages; people who notice the images getting deleted usually re-upload the images and add them back to the galleries (seen this happening already, as I clean up after IMB's mess). Chubbles 23:31, 10 August 2007 (UTC)

OrphanBot tagging issues

I uploaded Image:Second Ave Subway CGI station.jpg today with a rationale during its intial upload. Half an hour later, OrphanBot tagged the image as not having a rationale, despite it had one. An hour later, the bot tagged the same image a second time, leaving two identical tags on the page. I removed the tags and send Carnildo (bot's operator) a message about the issue. Then less than half an hour later, the bot tags the image for a third time, and I'm pretty sure it will do it again. If the bot has tagged my image incorrectly three times, I'm sure it has also done so to other images as well. –Dream out loud (talk) 00:29, 13 August 2007 (UTC)

In response to the message I left on Carnildo's talk page, Carnildo simply said, "I do wish people would stop coming up with new names for old templates." The statement is completely irrelevant, and it bothers me that the user does not seem to have any concern whatsoever about the issue. –Dream out loud (talk) 17:52, 13 August 2007 (UTC)

Actually, it's a concise summary of what caused the problem and what I did to fix it: someone came up with a new name for the {{fair use rationale}} tag, so OrphanBot didn't see the rationale, and someone else changed the name of the template that the {{nrd}} template inserts, so OrphanBot wasn't aware it had already tagged the image. --Carnildo 19:01, 13 August 2007 (UTC)

Real user working as bot

This page and other related talk about policies regarding bots, not about the technical details abouts bots, so excuse me if I ask a nonsense.

From what I have seen, despite working in other ways, a bot would be, from a strict technical point of view, a user like any other else. User page, user talk page, contributions, logs, etc; all the things a user has, a bot also has. Even more, there are some things of the kind of using a bot as a regular account (like answering to other users) that are not encouraged, but by not being so means they are possible to be done. But then, by logic, the opposite would also apply: if bots and users are technically the same, and a bot can be used as a main user account, then a main user account could be also used as a bot.

I guess that nobody would support an administrator employing bot techniques to tag all articles in a huge category for deletion, or to close a huge number of deletion discussions with a predefined result, or to delete all such articles in a row. But is there a way, other than the stadistic fact of more than 800 edits and 100 deletions in less than 10 hours, to prove that a regular account was being used as a bot?

Note: This topic is not for reporting the actions mentioned. That would be done somewhere else, at it's proper way. This topic is merely for the technical question, about if it's possible to recognize actions done by software or by human checking Perón 19:10, 22 August 2007 (UTC)

I don't know. Making 80 edits per hour, or 10 deletions an hour, is something I do manually all the time. I'm not sure how you'd prove it, since bots interact with Wikipedia the same way people do, usually. Have you tried asking the person? – Quadell ^{(talk) (random)} 19:35, 22 August 2007 (UTC)

it would just take some looking into it by someone who knows how bots work to check whether or not that the style of editing is human or bot related. dont forget using some simi-automatic scripts you can edit fairly rapidly. Ive clocked myself at over 120 edits an hour with some of my tools. β^_command 21:11, 22 August 2007 (UTC)

Regular accounts (without bot flag) can use scripts with manual approval of all edits, as long as the account doesn't edit "too fast", the edits have consensus, and the user stops and responds if another user questions the edits. "Too fast" in this context is debatable, but given that an "unflagged bot" is limited to 2 edits per minute, up to 120 edits per hour for a script-assisted task doesn't seem unreasonable to me. One sign of script-assisted editing is repetitive, identical or patterned edit summaries. Gimmetrow 22:20, 22 August 2007 (UTC)

Definition of Bot

I am aware that a bot it an automated program to read and edit wikipedia, but i was wondering if the following counts as a bot: Something that simply reads wikipedia and maintains its own log of issues to be manualy edited or if the bot is simply after statistics.

This wouldnt cause the problem of potentialy dangerous editing but it would download a lot of data from th server and i wondered whether wikipedia objects to such programs? —The preceding unsigned comment was added by 3.14 etc (talk • contribs) 23:29, August 22, 2007 (UTC).

Yes. — madman bum and angel 00:10, 23 August 2007 (UTC)

“ Bots that download substantial portions of the page database are prohibited. Instead, download the database dumps. See also Wikipedia:Mirrors and forks. ”

I have 2 queries :

1. is there some threshold for "substantial portions", for example would a script be okay if it only needed to read say 10 pages at a time and was used infrequently, and also was password protected to prevent thousands of users doing it?
2. could one of these bots run if approved by the bot approval guys, if i could justufy the usefulness of the bot, i'm sure that lots of bots must read a considerable about of data and that my idea wouldnt be unique in terms of bandwidth usage? (Pi 15:55, 23 August 2007 (UTC))

You'd have to give us more of an idea than that. 10 pages at a time, but for how long? What kind of throttling or maxlag value is being used? What do you consider a considerable amount of data? Maybe you should file a more detailed BRFA or post more information here. — madman bum and angel 17:35, 23 August 2007 (UTC)

Okay, here is a bit more info:

Bot 1 - I would like to make a PHP script. What this would do is allow me to enter the name of a user and then the script reads the Special:Contributions/User pages ( All pages) and the script calculates brief statistics on their usage including total number of edits, number of new articles started, top 5 most edited articles and so on. I don't think that this qualifies as a bot but may be prohibited under mirror laws.

Bot 2 - This would be run approx once per month i think. It would collect a list of dissambiguation pages from Catagory:disambiguation, at each run it would collect a set number then stop. Then the program would cycle through it's previous list ( stored in a text file) and for each one read the Special:Whatlinkshere page for that page and read the list of articles and store them in a seperate text file.

After it's run i can manualy look at the output file and for each one investigate whether it would be better for the link to be to the correct article ( if there is one) rather than the disambiguation page.

For example there are currently numerous pages with links to the LA disambiguation page and i am sure that a great deal of them should realy be pointed at Los Angeles

I think that this second bot would almost certainly need approval as it would look at enough pages, perhaps 1000 per run, however i don't know if it is a bot as it doesnt edit manualy and it would be close to impossible for it to do so. (Pi 17:54, 23 August 2007 (UTC))

Also, is it prohibited to run a bot simply as a means of testing the bot code, on the condition that the only page edited is the user's own sandbox? —The preceding unsigned comment was added by 3.14 etc (talk • contribs) 18:16, August 23, 2007 (UTC).

Bot 1 is pretty much covered by WP:EDITCOUNT/WP:COUNT. There are many, many tools available that do what you want, and most of them are on the Wikimedia toolserver, which provides access to Wikimedia database cluster slaves, which means they put almost no load on the servers.

Bot 2 is pretty much covered by Wikipedia:Disambiguation pages with links. Those pages are generated by querying the database dump, and since that's already done for you, you can get started right away.

As for approval, BAG members have stated in the past that bots that edit only in their own userspace or in their operator's userspace need not be approved. Anyone more senior than I can feel free to correct me if I'm wrong about that; I know at least Betacommand has indicated such in the past. Keep in mind, however, that all unapproved bots must run at a maximum of two edits per minute and may not edit fully automatically.

Thanks! — madman bum and angel 02:07, 24 August 2007 (UTC)

Edit-warring bots

Following this thread, I have decided to start here a more general discussion on whether bots should be allowed to keep doing controversial or even against-consensus edits. In my opinion, it's quite obvious that bots should not engage in edit warring with humans (apart from the cases of vandalism). I therefore propose to add something to this effect to the policy. At this moment, I don't see any exception other than reversion of vandalism, but I am sure there must be other ones. My suggestion is that bot should not revert humans unless explicitely authorized to do so. Tizio 01:06, 24 August 2007 (UTC)

the thing is the bot doesnt know its reverting, interwiki bots do their job they make interwiki links. they are not human, all they see is page foo should have X,Y, and Z interwiki links. if the page doesnt have them it adds it. bot cannot think dont know they are reverting users. β^_command 01:11, 24 August 2007 (UTC)

I gave it for granted that bots were reverting unintentionally (otherwise, they are presumed to be approved for that). The problem is: are we letting bots enter or create edit wars just because this way they are technically simpler to code? I again assume that bots work for the convenience of humans, not the other way around. Tizio 01:21, 24 August 2007 (UTC)

since this is not just one bot (its all of our pywikipedia interwiki bots) have you talked with the pywikipedia development team? they can be found on irc://freenode/pywikipediabot or did you file a pywiki bug request to have this addressed? β^_command 01:48, 24 August 2007 (UTC)

Yes, Fut.Perf. did this a few months ago, see Wikipedia:Bot owners' noticeboard#Interwiki bots forcing unwanted links, again, but did not get an answer. -- Jitse Niesen (talk) 07:33, 24 August 2007 (UTC)

Small hint: The (wiki) world is larger than enwiki, and pywikipedia developers really are not going to check all 200+ local bot pages for feature requests or bugs. If you've got a bug or a feature request, file it at the bug tracker of pywikipediabot, not at a local page. See the pywikipediabot website. ValHallASW 10:07, 24 August 2007 (UTC)

Please do some research before being sarcastic. Wikipedia:Bot owners' noticeboard#Interwiki bots forcing unwanted links, again is where Fut.Perf. said that he made the request. I hoped that it wouldn't be necessary to chase down the links, but apparently it is. The feature request was posed on meta (m:Talk:Interwiki.py) and the pywikipedia mailing list (post). -- Jitse Niesen (talk) 10:47, 24 August 2007 (UTC)

See this link. Pywikipedia now supports {{nobots}}, i just tested it with my bot on Ingria and the page was skipped. So if you dont want bots on a page, just but the nobots template on it and the pywikipedia interwiki bots will skip the page. multichill 09:47, 24 August 2007 (UTC)

That's only the case with the latest version of the framework, if I recall correctly. — madman bum and angel 11:39, 24 August 2007 (UTC)

Yes. It doesnt work in old versions of the framework. Bot operators are supposed to keep their bot up to date. It's fixed in r4096, we are now at r4100. multichill 12:31, 24 August 2007 (UTC)

And what if they don't? A call to block all pywiki bots that are not on the latest release to fix a single page problem seems like overkill, no? — xaosflux ^Talk 01:17, 25 August 2007 (UTC)

No, I don't think it's overkill to block those bots that add the ru-sib interwiki link to that one page. It's not that much work to upgrade, and the policy says it should be done on a daily basis. I think the latter may be a bit strict, but if you cause problems by not upgrading, that's a valid reason for blocking the bot. -- Jitse Niesen (talk) 02:47, 25 August 2007 (UTC)

How about not being a WP:DICK and just ask the bot op to update? β^_command 03:09, 25 August 2007 (UTC)

Good point. I fully agree that if it's not urgent (which I think to be the case here), then we should ask first, instead of blocking without warning. -- Jitse Niesen (talk) 05:17, 25 August 2007 (UTC)

A point that is perhaps missing in the discussion above is that it is the operator's responsibility to ensure that the bot is working correctly, not of the people pointing out bugs. The discussion so far: A says "this bot shouldn't add these links", B,C, and D reply "go to the xxx framework developers (who have no specific responsibility regarding this particular wiki) and convince them to do some changes". This is wrong. It's againts the main principle of the bot policy: operators are responsible for their bots, and a bot working against consensus is buggy. Tizio 10:56, 25 August 2007 (UTC)

Please dont be a WP:DICK, as a bot operator myself I have observed some things. When a bot is functioning properly it and its operator are ignored. If there is any problems, real or imaginary, instead of being nice about it and addressing the operator in a nice, calm, civil manner, people come screaming at us saying that "your d*m F*#k&*g bot screwed up and messed my page up" and that is about the sum of most complaints. instead of trying to address the root of the issue bot operators are abused. As for your It's against the main principle of the bot policy lets address the issue at hand. about the only tool that does interwiki linking Pywikipedia has a content dispute on a single page about a single link. The obvious solution that comes to mind, address the root of the problem find and talk to the development team of pywiki and attempt to find a solution. if AWB has an error you dont go to the user who operates AWB and expect them to fix AWB, no you address the AWB developers. To a point I agree with your statement though if this were not such a large class of bots I would say bring that to the operators attention in a civilized manner. but in either method as with anything follow up and clear communication are critical. β^_command 00:25, 26 August 2007 (UTC)

Wrong example ;-) I indeed once asked the author of AWB about a problem about it, and the answer was that the editors using AWB are responsible for every edit they make (the problem was that several editors where incorrectly "cleaning up" a point of ASCII where the non-clean version was indeed correct; I suggested a solution similar to that proposed by Fut.Perf. on this issue).

The issue here is: what is more important, the consensus of editors or the easiness of building bots? When you say "don't make changes only because of one page", you are saying that, at least in some cases, coding easiness has precedence. This is against the principle that bots exist to help human editors. Perhaps we should change our policy to "human editors are allowed to edit if they do not interfere with bots", then.

As for the civility issue, I just don't understand. Am I being a dick for saying that human editors are more important than bots? I too maintain a bot, and I too get complaints. The ones about this issue has been civil, at least the ones I have read. If a bot is operating incorrectly, what do you expect the complainers to do? Go away when you say "it's not my fault, in spite of the bot owner being me"? That's not the way of maintaining a piece of software. Tizio 13:11, 26 August 2007 (UTC)

As for the AWB example I was talking about errors that AWB itself makes not choices the user makes. (these are few and rare but happen) One point that I should point out that most interwiki bot ops are not programmers themselves, instead like I said address the root of the issue. I never said "don't make changes only because of one page". your method of handling the issue was a little off. how you handled the AWB problem was ok, but instead of doing the same thing with the interwiki bot that is maintained by the pywiki team you try to block the bots, even though those operators cannot do anything about it themselves because the code is maintained by someone else. as referring to WP:DICK you missed it. your idea of straight blocking is a little too harsh. β^_command 00:35, 27 August 2007 (UTC)

Thanks for the clarification: I wasn't the one who blocked the bot, although I generally support blocking a bot when there is a problem with it and its operator flatly refuses to acknowledge the problem (which means they are not going to ever fix it). IMO, it is the bot operators' responsibility to contact the developers about possible problems with their bots. Otherwise, we may have the situation when nobody is responsible for bots because operators aren't developers and developers aren't Wikipedia editors. Tizio 13:16, 27 August 2007 (UTC)

To a point I agree, if this wasnt such a large class of bots, (interwiki) and this as a single bot then I agree if the operator doesnt listen block the bot. But interwiki bots are kinda the exception to that. There are several pywiki people on en.wikipedia. Cyde, Misza13, and ValHallASW. I am not a pywiki dev myself but I do work with the framework, and can attempt to poke the devs to fix it. Personally I wouldnt mind if I was considered a dev on en.wiki (I use pywiki for all my bots) and I would be happy to try and help. β^_command 15:20, 27 August 2007 (UTC)

It appears that there is a consensus of the bot developers, and supported by the operators (who could always just NOT operate the bot) that all foundation language wikipedia articles should link to all the other language versions of the same article. People who don't speak that other language aren't going to follow the link there, and if they are multi-lingual odds are they are not limited to just en: and ru-sib:, so they will still end up there when following the links on all of the other languages. Not speaking ru-sub, can I assume that this dispute is summed up with not wanting to link to them because

1. The ru-sib: article is not written from NPOV, and though you've tried to correct it it remains that way and?
2. The ru-sib: wikipedia does not subscribe to NPOV, and dispute resolution there has been exhausted?

If so that makes the case that en: shouldn't link to ru-sub: ever. — xaosflux ^Talk 15:25, 25 August 2007 (UTC)

'who could always just NOT operate the bot': i.e. you would like to update all interwiki links on enwiki by hand? It's fairly easy to create a switch for the bot to never write to enwiki... but I assume we can agree that is not an acceptable alternative?

Anyway; if ru-sib does not subscribe to NPOV, it should be *closed*. IMO it is not up to editors of a certain page on enwiki to oppose to a decision to keep ru-sib open in a way that could very well be described as Civil disobedience. Just my $0.02 ValHallASW 19:25, 25 August 2007 (UTC)

I personally would certainly support closing ru-sib, and I quite agree that, in the absence of a meta decision to that effect, a project-level decision at enwiki would still be preferable to fighting this out at the individual article level. For the record, there is no "decision to keep ru-sib open" either. There's a proposal for closure at meta, which has been running for more than half a year and has resulted in a very strong supermajority in favour of closing. As appears from the talk page, the only reason no binding decision either way has been taken is that none of the stewards wants to assume the responsibility. In which case, I'm certainly of the opinion enwiki would be entitled to go ahead and make its own decisions.

However, on the individual level, if editors do oppose a link on an individual article level, for whatever reasons, I as an admin still have no justification for just telling them they are wrong and they should just shut up. There may be a "consensus of the bot developers, and supported by the operators", but bot developers and operators don't get to dictate policy, so the local objectors still have the right to expect their position to be taken seriously in a normal dispute resolution process. That's why my initial reaction was: we first need to get the bots out of the way, on this page, and then let the humans work out the underlying content issue. Fut.Perf. ☼ 07:37, 27 August 2007 (UTC)

Bot for adding links

Is there a bot available that can take a newly created template with many links and systematically add the template to the bottom of each article the template links to? The only problem might be getting it to add the template in the right location. Richard001 09:36, 24 August 2007 (UTC)

You may want to add this question to Wikipedia:Bot requests, with more details about what template in particular and where it should be added on the given pages. — madman bum and angel 11:40, 24 August 2007 (UTC)

Ganeshbot articles

I find that several hundred or maybe a few thousand stub articles were built by Ganeshbot without a reference section and the only source being Template:GR is listed wrong (see WP:ASR} it should have been a proper footnote to the Indian Census (but some like Ghorawal did get a reference section). Many of them like Ghorabandha have not matured past stubs. Some like Kangeyam have matured into large unreferenced articles. Some like Mandi Gobindgarh have grown and captured a few references that could use {{citation style}}. I got in a fairly long discussion with the bot builder at User_talk:Jeepday#Re:_Kangeyam after posting a {{uw-unsor1}} to his talk page, suggesting he figure out how to add a reference section and list the reference instead of the wikilink on all those unreferenced articles. He responded that it was beyond his ability and that I make a request at WP:BOTREQ . Before posting here I did a reality check at User_talk:DGG#Bot_articles and did a review of Wikipedia:Bot policy. Thoughts? Jeepday (talk) 12:59, 29 August 2007 (UTC)