Wikipedia:Bots/Requests for approval/Polbot 7


 * The following discussion is an archived debate. Please do not modify it. Subsequent comments should be made in a new section. The result of the discussion was Symbol neutral vote.svg Request Expired.

Polbot task 7
Operator: – Quadell (talk) (random)

Automatic or Manually Assisted: Automatic, but with some supervision

Programming Language(s): Perl, using Perlwikipedia

Function Summary: To add an image use rationale similar to Commercial logo rationale to some images of logos

Edit period(s) (e.g. Continuous, daily, one time run): one-time run, or in small batches until complete

Edit rate requested: 6 edits per minute

Already has a bot flag (Y/N): Y

Function Details: Bots are extremely limited in their rationale-writing capabilities. For most uses of most images on Wikipedia, a bot will never be able to write and accurate and useful rationale. However in a few cases, I believe that a bot could improve Wikipedia by adding a limited rationale for a very specific kind of image use. When an image is tagged as a non-free logo, and when that image is only used in a single article, and when that article appears to be about the organization represented by the logo, I believe that a rationale similar to Commercial logo rationale can be included by a bot, with a high degree of accuracy. A good example of what this bot will do is. Benefits and limitations are below.

Discussion

 * I am aware of the culture war between the "more non-free images" faction and the "fewer non-free images" faction, and I expect that disagreement (with its attendant drama) to show up here. I sincerely hope we are able to limit discussion to this proposed bot function, and not get sidetracked with arguments about BetacommandBot, changes to fair-use policy, or other side issues. Also, I recently ran this function without getting bot approval. This was a mistake. While I was out, the bot was blocked and her changes were reverted, which I understand. (That botrun also performed other functions, but I'm only addressing logo rationales here.) My error has been discussed here, here, and here. I've learned an important lesson, and I'd like for this discussion to focus on the merits or problems with this proposed bot, going forward. – Quadell (talk) (random) 15:43, 20 December 2007 (UTC)
 * Benefits of this bot function:
 * New images without rationales are routinely tagged (by BetacommandBot and others) as deletable, in line with policy. Some of these don't have valid rationales because the image's use is flat-out invalid, and no valid rationale is possible. But others are "rationaleable" -- that is to say, with a good rationale there's no reason to delete the image. Certainly it's better to add a rationale and save the image in these cases. – Quadell (talk) (random) 15:43, 20 December 2007 (UTC)
 * In addition, Betacommand has recently stated that, due to a previous agreement (which I can't find onWiki), BetacommandBot will begin tagging older images without rationale at the beginning of the year. (These images were uploaded back when rationales weren't required, or when that requirement wasn't enforced, and in many cases the original uploader is gone.) I understand the usefulness of this function, and I support it, but that makes it even more important -- perhaps even urgent -- that valid rationales be added to images. For the small number that can be added by a bot, I think it would be advantageous to do so. – Quadell (talk) (random) 15:43, 20 December 2007 (UTC)
 * The average quality of rationales added by humans is quite low. Most rationales have significant problems: they don't linkback to the article, they make incorrect assumptions about policy and copyright law, the "purpose" description is vague to the point of uselessness, etc. Even with Polbot's occasional errors, the average quality and accuracy of her rationales is far higher than the Wikipedia average. – Quadell (talk) (random) 15:43, 20 December 2007 (UTC)
 * Limitations of this bot function:
 * Polbot is obviously unable to determine whether a logo has a valid rationale already or not. According to her current code, if a logo does not use a non-free use rationale template (or its synonyms), it is assumed to not have a valid rationale. This can lead to Polbot adding a second rationale to an image that already had one. (Example). I don't consider this a major problem, since (a) the rationale I add is usually superior, in that it's deliberately-worded and templated for ease of maintenance, (b) sometimes multiple humans add different rationales for the same image use, and it hasn't been a problem in the past, and (c) the existing rationale is frequently invalid anyway. If, however, the community views this as a problem, I could limit the rationale-adding to those that don't [link to the article / use the phrase "rationale" / etc.]. – Quadell (talk) (random) 15:43, 20 December 2007 (UTC)
 * Polbot occasionally (very occasionally) adds a rationale for a use that does not apply. This occurs when a logo is only used in one article, and Polbot believes that the article is an article on the organization represented by the logo, but the article is actually about something else. (Example.) This is rare, but unavoidable. To deal with these cases, Polbot adds "other_information" to the rationale stating "This rationale only covers use in the article on this organization represented by this logo. Any other use requires a different, separate rationale. This rationale was generated by a bot, based on the fact that this is tagged as a logo and is used only in the [article name] article. If this is not correct, please remove." In these cases, the new rationale is incorrect and should be removed -- but I'll note that (a) it doesn't happen much (b) when it does happen, the image is almost always "unrationaleable", meaning that the image's use is against policy regardless of what rationale is used, and (c) in my opinion the benefit of adding correct rationales to 99% of images outweighs this small cost. – Quadell (talk) (random) 15:43, 20 December 2007 (UTC)

I find the task very useful. Can you estimate the error rate?  Snowolf How can I help? 15:56, 20 December 2007 (UTC)
 * Of the hundreds of logos Polbot added rationales to, only one has been found to have been incorrect (in that the article that the image was used in was not actually about the organization signified by the logo). Dbiel also pointed out a bug, which I fixed, and some users have noted Polbot adding a second rationale, which I don't consider an error. – Quadell (talk) (random) 16:02, 20 December 2007 (UTC)


 * I'm afraid that two aforementioned 'limitations' are unacceptable:
 * Having two rationales on one image description page is misleading, how should we decide which one should be used? Also, people may correct only one of the rationales, increasing discrepancies beween them even more.
 * Wrong rationale means copyright violation. Imagine, someone creates a vanity page about himself and uploads a logo of his favorite football team to be used in the article. Your bot will simply tag the image with a 'rationale' that will the problem from human eyees. Max S em(Han shot first!) 16:06, 20 December 2007 (UTC)
 * A correct rationale does not protect from copyright violation. The rationales are to satisfy Wikipedia policies. Fair use does not require an existing rationale (you only have to provide one when challenged). I think your first point is valid, and I'd suggest limiting this bot's operation to image pages that do not have a "Rationale" section already present. Pages that fail to use the word rationale usually do not have good rationales. They should be flagged for immediate attention by humans (that is something a bot can do). Carcharoth (talk) 16:21, 20 December 2007 (UTC)
 * Concerning point one, having two rationales is not uncommon, and is not contrary to any policy, but I'm open to skipping logos that have existing rationales (even if those existing rationales are almost certainly inferior). Concerning point two, the violation exists both before and after the change. The bot doesn't create the problem, and it doesn't hide it either. The problem with the use is that it fails NFCC#8, and bringing the image in line with NFCC#10 doesn't change that. – Quadell (talk) (random) 16:27, 20 December 2007 (UTC)
 * If an image has no rationale, it will be automatically detected by a bot and tagged for delayed deletion. If it has invalid rationale, bots will be unable to detect it automatically and it may avoid deletion, that's what I mean. Max S em(Han shot first!) 17:38, 20 December 2007 (UTC)
 * I see what you mean. But I think that's a good thing. If an image is otherwise acceptable, but has no rationale, then my bot will add a valid rationale and prevent it from being deleted. If the image is used improperly, then no rationale will save it. But it should be deleted for being used inappropriately, not for missing a rationale, if the use is the real problem. – Quadell (talk) (random) 17:50, 20 December 2007 (UTC)
 * Actually, a wrong rationale does not equate to a copyright violation. It merely equates to our losing, in that particular case, a double-check we have to make sure images are used correctly.  Wikidemo (talk) 02:31, 21 December 2007 (UTC)

Hi Quadell, could I ask what if the foundation resolution and the proposed policy created in response, are going to change the bot function? Do you envisage having to substantially rewrite the bot program? Addhoc (talk) 16:39, 20 December 2007 (UTC)
 * No, I think it'll be fine. The 2007 Wikimedia licensing resolution simply says "By March 23, 2008, all existing files under an unacceptable license as per the above must either be accepted under an EDP, or shall be deleted." That's really the motivation behind creating this bot. I don't expect it's function to change in March. (I suspect it will be completed long before, actually.) – Quadell (talk) (random) 17:18, 20 December 2007 (UTC)


 * Issues related to this bot and this specific set or functions:
 * Images that have multiple FUR's for a single article, where the second FUR is missing the article name, the bot will repeatedly edit the first FUR adding the article multiple times. This has been posted on the Bot's talk page with an example. This has been noted as having been fixed some time after this post was first started, so this is probably a non issue at this time, except for the need to test the fix.
 * The bot generates a generic source statement in an attempt to bypass the need for a valid source statement. This maybe something that needs to be discussed elsewhere as the issue is not very clear in the guidelines.
 * There is a template currently in use di-no source which would imply that a source statement is required.
 * non-free use rationale provides a field for source info but provides no instructions or link to any information as to how the source should be identified.
 * Non-free content criteria make no reference to the need for any source information but include the following:
 * "For the full non-free content use guideline (including this policy and its criteria) see Non-free content."
 * Non-free content does state the following:
 * "(a) Attribution of the source of the material and, if different from the source, of the copyright holder. See: Citing sources."
 * Citing sources does provide an example of a source for a image downloaded from the web. It does not make any comment regarding generic source info.
 * It would appear that work needs to be done on the guidelines regarding what is required as the source of uploaded images.
 * User:Quadell has removed the references to the other functions done by Polbot, does that mean that function #7 is replacing all the other functions?
 * User:Quadell has provide a link to the source for Polbot, but unfortunately that links to a page that has nothing to do with this function. Dbiel (Talk) 16:58, 20 December 2007 (UTC)
 * Thanks for your comments. Your first error you note is fixed, and I'll need to prove that during a trial run. Your observation about the source information is a valid concern, but consider: the reason we require a source is to ensure that the copyright-holder is credited. For a GE logo, the ultimate source is the GE company -- the proximate source might be the GE website, or a third-party site, or a scan from a magazine ad, but that's really not relevant. (This is also the case for album covers -- frequently the image is sourced to Amazon.com, but Amazon.com is just a middleman.) What's important is that the ultimate source be credited, not the proximate source. And the ultimate source is the organization itself, not any particular website. I personally feel that "intellectual property owned by the organization represented" is a valid source. If consensus is that this is not valid, I could change it to say "intellectual property owned by General Electric", where "General Electric" is automatically inserted as the article name using the image, but that adds uncertainty and I'd rather not. The bot is not really able to determine the source link -- there's too much room for error there. I hope the sourcing will be deemed adequate as it is.
 * As for the previous Polbot functions, I didn't delete them, I just moved them to a subpage. That's linked as "Older functions are at older tasks" in the "other_information" section. Polbot hasn't really been working on any of those older functions for many months now. You're right that I haven't yet put up the source code for this function. I'll go do that now. All the best, – Quadell (talk) (random) 17:15, 20 December 2007 (UTC)
 * Thanks for the reply. It points out a major problem with terminology regarding the FUR. Just what is the definition of "source"? The guidelines are not much help in answering this question. The only thing I could find was:
 * Images must include source details and a copyright tag on the image description page. It is important that you list the author of the image if known (especially if different from the source), which is important both for copyright and for informational purposes. Some copyright licenses require that the original author receive credit for their work. If you download an image from the web, you should give the URL:


 * Source: Downloaded from
 *  http://news.bbc.co.uk/1/hi/world/europe/4280841.stm 


 * If you got the image from an offline source, you should specify:
 * Source: Scanned from public record #5253 on file with Anytown, Somestate public surveyor
 * You seem to be using the term "source" to be the copyright holder. The guidelines need to be cleaned up to address this issue Dbiel (Talk) 18:00, 20 December 2007 (UTC)
 * Source has always meant all stages from creation to arrival at Wikipedia, but only where relevant. The original source is the most important, but sometimes intermediate sources need to be credited as well. Ideally the proximate source is quoted as well, to allow complete verification of the provenance of the image. There may also be a few cases where the copyright holder is different from the source. Sometimes the creative source (which may still need to be credited) sells the copyright to another entity. Carcharoth (talk) 20:26, 20 December 2007 (UTC)
 * and when that article appears to be about the organization represented by the logo cannot be evaluated programmatically. This should be manually attended. —Random832 17:20, 20 December 2007 (UTC)
 * I disagree. The bot has only made one error in this, out of hundreds done, and that error only changed an image from not having a valid rationale, to still not having a valid rationale. – Quadell (talk) (random) 17:35, 20 December 2007 (UTC)
 * No, it changed from having an invalid rationale that BetacommandBot would tag, to having one that it would not. —Random832 19:05, 20 December 2007 (UTC)
 * So should Betacommandbot be the only entity that is allowed to tag invalid rationales? If the rationale is invalid, a human can still tag it as invalid. Unlike Betacommandbot, the human can give a specific reason, rather than a generic "10c, might be incomplete rationale or a missing one, but I can't tell because I'm a bot". Ironic, really. Carcharoth (talk) 20:22, 20 December 2007 (UTC)

The function summary for Polbot function #7 does not seem to be accurate at least based on its past edits. It appears to have multiple functions dependant upon the state of the image file. I would like to see a more complete description of just what this bot will do especially since some of its edits have already been challenged. Dbiel (Talk) 17:23, 20 December 2007 (UTC)
 * To add the article link entry to the FUR if it is missing.
 * To add an entire FUR which apparently varries based on a number of different factors.
 * As I said above, "That botrun also performed other functions, but I'm only addressing logo rationales here." The function summary for function #7 is the only thing I'm requesting permission to run. – Quadell (talk) (random) 17:33, 20 December 2007 (UTC)
 * I am not talking about functions 1-6 only function #7 which itself seems to be multi-functioned as state above. Or are you saying that the addition of only the article title has been removed from the bot function? and that it is now limited to only adding entire rationales?Dbiel (Talk) 17:47, 20 December 2007 (UTC)
 * I'm saying that my botrun yesterday, which was terminated and reverted, did several things. It added backlinks to rationales. It added rationales to logos. It added rationales to album covers and book covers too. But that's not what I'm asking permission to run. I'm only asking for permission to run the function described above -- adding rationales to logos that lack them, under certain limited conditions. My botrun yesterday was multifunctioned, but the task under consideration here is not. – Quadell (talk) (random) 17:54, 20 December 2007 (UTC)
 * Thanks for the clarification, sorry if I missed the note that those functions had been deleted from the bot. I got confussed when you stated that you "fix" the problem of adding the article name multiple times. Did not realize that your fix was to remove that functionality. Thank you again for the clarification Dbiel (Talk) 18:06, 20 December 2007 (UTC)


 * Id like to make a note here, I reviewed over 100 Polbot tags, 98% of what I looked over had problems, it was adding invalid logo rationales to images that were not used properly. and it was adding rationales to images that should clearly be deleted. (images of BLP used on the article about the person). rationales serve a purpose they are not just some stamp you add to an image that makes it ok. they are for saying that this image should be on wikipedia and here is the reason. any bot wrtten rationales are crap, its an attempt to bypass policy and just rubber stamp OK on all single use non-free images. I hate to say this but bots cannot write rationales. βcommand 17:27, 20 December 2007 (UTC)
 * You see, at the moment, Betacommand (and others) can use bots or scripts to tag images for deletion under NFCC#10c because they can programmatically evaluate the absence of a link or name on the image pages as an "invalid rationale" (which is quite correct). This allows vast numbers of images to be quickly tagged for deletion even if there is an incomplete rationale as opposed to no rationale at all. Now, if this method of detecting 10c-invalid rationales were removed, then Betacommand and others would have to find a different way of detecting invalid images. The absence of a copyright tag (NFCC#10a) can also be detected programmatically, but the correctness of the tags can only be evaluated by humans. Similarly for NFCC#8 (relevance to article) and NFCC#10b (sources). These can only be judged by humans. So the end result, if Polbot gets approval to do this, is that a large chunk of the images that could potentially be tagged for deletion, will now have to be checked manually to see if they comply with NFCC#8 and NFCC#10b. But this is not news at all. All images used under the NFCC will have to be manually checked for NFCC#8 and NFCC#10b even if a bot-generated or templated rationale is used. My feeling is that the rationale can be checked at the same time as the sources and relevance are assessed, but they shouldn't be deleted for an invalid rationale simply because that can be detected programmatically. Ultimately, non-free images should have a checking system whereby trusted users can tick the boxes for the NFCC, but that is best left until after March 2008. Carcharoth (talk) 17:39, 20 December 2007 (UTC)
 * Oh, and Betacommand, can you provide a list of those 100 images you looked at? Thanks. Carcharoth (talk) 17:43, 20 December 2007 (UTC)
 * i Dont have a list of the images I looked at, there were a lot of album covers that were missed labeled along with a lot other images that were improperly tagged. My point is simple, its a hell of a lot simpler to Identify images with with no rationale then a messed up rationale. When writing rationales you need to consider the usage, bots cannot do that. when images are tagged currently you have seven days to address the issues. when an image is tagged as no rationale who ever writes the rationale needs to consider the reason for its usage. if you slap on some invalid rationale it makes finding images that are used improperly hard. Now setting aside that, if we can create a valid system of categories that include only corporations/organizations we might be able to do this function by crosschecking the image use page with the list of pages that are in that category. doing such a massive re-organization of the cat structure will take a lot of time.  without the crosscheck I cannot support this bot. βcommand 18:00, 20 December 2007 (UTC)
 * Those categories are generated from the copyright tags, aren't they? non-free logo and the like? If that tag is being used on album labels, that is an NFCC#10a problem, not an NFCC#10c problem. I suggest you write a bot to find the mistagged album covers and retag them with non-free cover. Carcharoth (talk) 18:07, 20 December 2007 (UTC)
 * Carcharoth, you miss-understand what I am saying. the category check would not be with the image but to see if the page where the image is used is in the Corporation (or what ever we name it) category. βcommand 18:13, 20 December 2007 (UTC)
 * Oh, right. OK. Carcharoth (talk) 18:17, 20 December 2007 (UTC)
 * (ec, to Betacommand) I take great issue with your statistics. Can you really point to around a hundred logos where Polbot added a misleading rationale? I have only found one, and I've been through them as well. Could you please provide links? I'm also confused about your statement about BLP: can you link to any logos that were being used to illustrate a living person? I certainly understand about the reasons and importance of rationales, and if you read my "Function details" above, I agree with you most of the time. But saying "any bot wrtten rationales are crap" is clearly false; this is a bot-written rationale, for instance, and it is not crap. It's perfectly valid, and it has taken an out-of-policy image and brought it in line with policy. That's what this bot will do 99% of the time. Your statement that "its an attempt to bypass policy and just rubber stamp OK on all single use non-free images" is a terrible assumption of bad faith, and it's just not true. I have no interest in bypassing policy, and besides, we're not discussing all images, just logos, and just logos that are used correctly but lack a rationale. – Quadell (talk) (random) 17:45, 20 December 2007 (UTC)
 * I can cite 15 images where Polbot added correct rationales. Those rationales added to a few radio (and one TV) logos yesterday. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 -  NeutralHomer  T:C 18:59, 20 December 2007 (UTC)
 * Comment The above cited added rationales are not always correct: Image:044acdb0.jpg, Image:2007wqmz.JPG, and Image:2007wwwv.JPG are all not of low resolution, despite what the FUR states. Additionally, the generic fill-in for the "source" section is most likely unacceptable, as something like that could be written for EVERY image, and it doesn't mean much at all. This is the problem with bot-added FURs - they are too generic to apply to all cases. JPG-GR (talk) 21:24, 20 December 2007 (UTC)
 * I see what you mean about lo-res. I could try to put in code than restricts that portion of code to ones where the images is below a certain size. Regarding the other concern, though, "something like that" could not be written for every image. In the case of logos, the logo represents the entity that holds the intellectual property -- the whole purpose of the image is state who the source is. – Quadell (talk) (random) 22:04, 20 December 2007 (UTC)
 * I would like to add that Polbot is writting some excellent rationales, my main question is regarding the source entry and if the entry being used by the bot is acceptable, noting that the guidelines are as clear as mud on this issue. Dbiel (Talk) 17:58, 20 December 2007 (UTC)


 * The Atran rationale is a good example - I'm reasonably sure that most Wikipedians would consider adding the rationale to be a significant improvement. Addhoc (talk) 17:54, 20 December 2007 (UTC)


 * Betacommand, I think you may be confused about what this bot will do. You mention BLP and album covers and the like, but that is expressly not relevant to this bot function. I really don't think there is a problem with logos being given incorrect rationales, but, as I said, I invite you to provide difs so we can discuss the matter intelligently. I think I would paraphrase your argument as this: "You can't be 100% sure that every rationale will be accurate. There is always the possibility that a tiny fraction of logos will be given rationales that don't actually apply. And it's better to leave the 99% untagged, and have them needlessly deleted, than tag the 1% incorrectly, and have them needlessly kept." I just don't agree with this position, and I doubt very many others do either. – Quadell (talk) (random) 19:03, 20 December 2007 (UTC)


 * I had to bring this out from the bit above. Quadell said: "If the image is used improperly, then no rationale will save it. But it should be deleted for being used inappropriately, not for missing a rationale, if the use is the real problem." - Absolutlely! This is what I'm getting at with my bit about about deletions under 10c for incomplete rationales. Carcharoth (talk) 18:03, 20 December 2007 (UTC)

From my side this bot correctly patched up over four dozen images that I was looking to have to update. It served a necessary administrative purpose and left behind a happy user. How handy is that? The only thing I had to do was go back and re-do some of the bot patches that were un-done - despite their being correct - by over-zealous copyright copz. Some folks could take a lesson, get a broader perspective, stop being so mean-spirited, and engage in a friendly and helpful act or two. Good work Q. Wiggy! (talk) 20:45, 20 December 2007 (UTC)

As I see it, the bot, though not perfect, does provide several useful functions dependent on which side of the fence you stand on.
 * It creates a FUR that makes the image fairly well protected from other bots that have as their primary goal the deletion of images with invalid FUR's.
 * It provides a standardize FUR that is well worded and can be easily edited to fix any miss representations.
 * For logo's it creates a generalized source statement that is accurate to the extent of ownership, as the logo is generally owned by the organization it represents. I still question what Wikipedia expects as a source statement due to the fact the guideline are very lacking in this area.

I would say that based on the current limited function of this bot, the chances of it make a serious error appear very slim and it is better to error on the side of keeping a few images (which can easily be tagged for deletion if necessary) then to open a vast number of images to automatic deletion due to the lack of some technical requirement that can generally be easily fixed. BUT I would like to get the commitment of the bot operator that he will not change the bot's function again without first going though the approval process, even if that change is simply moving to a different license type and doing basically the same function as with the logo license as the results may not be as acceptable and could be prone to a much greater error factor.

I still believe that the Guidelines need to be rewritten to clearly state what the FUR source entry should look like and what information it should contain, but until that happens, I see no problem using the generic one use by this bot. Dbiel (Talk) 22:29, 20 December 2007 (UTC)


 * Angel-smiley.svg You have my word. If this bot performs without too many problems, I may later request approval to run something similar for CD covers, etc. But either way, I will not run this bot function for anything other than logos without going through the approval process again. – Quadell (talk) (random) 22:59, 20 December 2007 (UTC)


 * I would have no problem with this specific task if and only if you can prove the page the image is actually a organization/corporation. (it would need to crosscheck the use with a proven list of such pages IE limited categories). βcommand 00:08, 21 December 2007 (UTC)


 * Isn't a product or brand logo the same?


 * Rather than (simply) adding a comment that the use rationale was added by a bot, would it help to also add an extra hidden field that gives it a revision / run number? That way, if anyone ever decides they need to examine, delete, change, or categorize all these new rationales there will be an easy tag to grab onto.  You may also want to consider using a template like Template:logo fur.


 * Finally, if the experiment goes well for logos you may consider film posters, book covers, CD covers, etc. but the bot could add an unenabled rationale, meaning it still technically has no rationale until a person looks at and approves it. Just as we require someone to examine an image by hand before deleting they can and probably should look at it by hand before approving the new rationale. You can add the rationale in a template newly created for the purpose, that clearly explains that the rationale is as yet unapproved pending human review, and puts the image into an "images to review" category.  Image reviewers can easily follow behind the bot to see if the rationale is correct for that particular image and use.  If yes they can change the template name and it's approved.  If no they can delete the rationale or flip some other switch to have a bot delete it.  I hope that's clear.Wikidemo (talk) 02:24, 21 December 2007 (UTC)


 * Just for my own clarification, was this edit needed? The page had a link to the article in the "Purpose" field. Had Polbot not made this edit, was this image in any danger of being tagged by any of the FUR-checking bots? Gimmetrow 07:25, 21 December 2007 (UTC)


 * Without that edit, it is possible that the image would have been deleted for having an incomplete rationale. I think the rationale was already sufficient, but Polbot's edit at least made it more clear, more obvious, and more machine-readable. But this particular edit is not what I'm requesting permission for at this RFBA. – Quadell (talk) (random) 12:09, 21 December 2007 (UTC)


 * The edit was needed as the "Article" field is a required field in the template. As far as being in danger of being deleted, any admin who would have deleted it on those grounds should be shot. It was in danger of being tagged by a bot for the missing field, but anyone with any sense would have simply deleted the tag and made the same edit that Polbot did by inserting the missing "Article" field and entry rather than deleting the image. To restate, the original entry DID meet the requirements of Wikipedia for a fair use rationale, which currently does not require the use of the template, but it did not meet the current requirements of the template that was being used. Note: the article field was added on September 27, 2007, therefor any image using that template prior to that date needs to be edited, which is probably a good function for Polbot, but that would need to be covered under a separate request. Dbiel (Talk) 17:40, 21 December 2007 (UTC)


 * Where in FURG, NFCC or the Non-free use rationale docs does it *require* the use of the article field if the template is used? I don't see it. (I realize this is tangential since the bot won't be doing exactly this, yet.) Gimmetrow 02:21, 22 December 2007 (UTC)


 * That is actually one of the problems, it is not flagged as a required field, but the way the template has been edited, it has become a functional requirement. The article field is used to fill in the article name at the top of the rationale. If the field is not used the the following is added to the header "Non-free / fair use media rationale - NEEDS ARTICLE NAME and the image is added to Category:Non-free images lacking article backlink If the article field is used then the article name replaces NEEDS ARTICLE NAME and the article is moved from Category:Non-free images lacking article backlink to Category:Non-free images with valid backlink. This is why I call it a required field. Dbiel (Talk) 04:01, 22 December 2007 (UTC)
 * Just to be even clearer here, it is not required to have a link (only recommended). It's only required to have the name of the article the image is used in, and that name can be anywhere, not just a rationale template. Thus those images in the "lacking article backlink" category really only refer to the lack of a link in the template. It is possible the article is named or linked elsewhere on the page. If that's the case, then Betacommandbot won't tag it. If Betacommandbot does tag images in the "lacking article backlink" category, it is because most of those images are a subset of those lacking any name or link to an article. That wasn't clearer, was it? The other complication is that the article name field in that rationale was a late addition (or at least was not there for a substantial period of time). Some rationales were written using a template that didn't include the "article" parameter, or were written when there was no article parameter. Thus we get incomplete rationales due to that, and incomplete rationales due to people not filling it in correctly. The latter is the fault of those filling it in. The former is the fault of the system. And yes, some admins (when clearing huge backlogs) do delete images with rationales that are easy to fix by adding an article name (and give NFCC#10c as the reason, even if they really mean they are deleting it for another reason). I don't think such admins should be shot, but they should be told not to do it again, and be asked to undelete and fix the images. Carcharoth (talk) 09:04, 22 December 2007 (UTC)

No that I have checked the source, and not that I'm taking any standpoint on if this bot should be allowed to do this or no, but I've though about some implementation limitation about article/image matching: The longest word in the filename must match any word in the article title, that same name must also be found in the description of the "Non-free use rationale" template. Only images that are using this template should be updated, others shouldn't be touched, and only images used on one single main space page should be edited. → Aza Toth 17:28, 21 December 2007 (UTC)


 * If the bot is only adding rationales to logos, and is using a template, why not just use a template transclusion? Gimmetrow 02:21, 22 December 2007 (UTC)
 * It's not using a template, other than the rationale template, which is transcluded. – Quadell (talk) (random) 02:44, 22 December 2007 (UTC)
 * But it is using a template, by which I mean all the text it adds is identical except for the name of the article. This could be reduced to  and  Gimmetrow 03:54, 22 December 2007 (UTC)
 * OK, seeing no response: I'm basically asking in a roundabout way, why not use Logo fur ? Gimmetrow 00:57, 24 December 2007 (UTC)
 * To be honest, I like my way of putting it better. :) – Quadell (talk) (random) 04:12, 24 December 2007 (UTC)


 * (repost comment) I would have no problem with this specific task if and only if you can prove the page the image is actually a organization/corporation. (it would need to crosscheck the use with a proven list of such pages IE limited categories). please implement this cross check. βcommand 01:15, 25 December 2007 (UTC)

(I'm on Wikibreak until Jan 2, but I'm able to briefly log on.) The bot can't prove that the page is actually an organization. Very few things about images can be proven. Besides, even if it could, the one error that has so far been found has been a case where the article was on an organization -- it just wasn't on the organization represented by the image. So even if I could prove this, it wouldn't reduce the error rate from 0.5% down to 0.0%.

It's about to be moot anyway. At the start of 2008, the process will begin to tag all non-free images for deletion if they don't have valid rationales. I appreciate the desire to make this bot as good as possible, but we've come down to the wire now. There isn't time for a major re-coding. The community just needs to decide -- is this bot, with its <1% error rate, good enough? Is the fact that it will save thousands of otherwise-valid images from deletion worth the hypothesis that it may add a few incorrect tags? If this bot doesn't get approved soon, there'll be no point. – Quadell (talk) (random) 13:04, 30 December 2007 (UTC)


 * I will agree with Quadell, the error rate is well within the acceptable range and as such the bot should be approved for the requested purpose. Dbiel (Talk) 19:24, 30 December 2007 (UTC)
 * Agree, bot should be approved and error rate monitored. Addhoc (talk) 19:35, 30 December 2007 (UTC)


 * This bot should not be rushed just so it can work against another bot. That opinion gives me a bad feeling. Now, Betacommand has a valid concern, logos must be associated with a recognized organization or they cannot be used on that organization's article and this bot needs to ensure that it is not being exploited or reading erroneous information or it should not be permitted to run. --uǝʌǝsʎʇɹnoɟʇs (st47) 23:37, 30 December 2007 (UTC)


 * The argument that this bot is needed to save images from being deleted sounds like a muddy way of saying "let's tag all these logos with these tags so those that should be deleted are harder to find." Auto-generated tagging defeats the whole purpose of the FUR in the first place, IMO. JPG-GR (talk) 00:12, 31 December 2007 (UTC)

I'm back from my break. Having read the above (especially ST47's comments), I'd like to try to clear up a misconception. Polbot has nothing to do with βcommandbot. Polbot will not "work against" any other bot. For the record, I approve of βcommandbot's function to tag for deletion those images without valid rationales, as required by the Foundation Resolution. My bot is designed to add proper rationales to some images that don't have them, which is a complimentary function to βcommandbot's function of tagging-for-deletion those that still don't. I really think there's some assumptions of bad faith here, as if I were trying to undermine the Foundation's requirement. I support the that requirement -- I'm trying to satisfy it by adding proper rationales in those few cases where a bot can do that. The urgency is not "that βcommandbot is about to go on a rampage" or anything like that. The urgency is that the Foundation requires rationales, and soon.

I'd like to respond more directly to βcommand's concern about making sure the article is about a "recognized organization". This is a bad idea, and I'll tell you why. I think this concern was raised in good faith, but I don't think it can be used to improve Polbot's function. I do think it can be used as a delaying tactic, but I hope it's not used for this purpose. – Quadell (talk) (random) 16:23, 2 January 2008 (UTC)
 * 1) It's clearly impossible. Nothing in the category structure, infoboxes, text parsing, etc., can ever "prove" that an article is on a "recognized organization", as any bot-operators can tell you.
 * 2) It's unnecessary. Even if it were possible, Polbot's perceived false-positive rate is under 1%, which is well under the rate of some of her other functions (which were approved without hesitation).
 * 3) It would not prevent false-positives. The only false-positive found thus far was a logo for an organization that was used on the article about a different organization. This requested (impossible) check would not have prevented this.
 * 4) It will create many false-negatives. Logos don't have to be for "recognized organizations". An individual (e.g. Trent Reznor) can have a logo, a concept can have a logo, etc. Logos are used to illustrate the thing they represent, but that thing can be almost anything. Limiting its scope to "recognized organizations", even if possible, would prevent valid images from being given valid rationales.


 * it can be done, but you will need to do some work on the categorization side. and a sanity check is always a good thing. If I had more free time I could get this operating in 2-3 months. βcommand 16:32, 2 January 2008 (UTC)
 * That would be after 23 March 2008, right? Carcharoth (talk) 03:36, 3 January 2008 (UTC)


 * It seems likely that certain narrow classes of images, such as logos, have a fairly specific use which can be justified in a mechanical manner. What if this task  involved manual approval of each proposed edit? Gimmetrow 03:17, 3 January 2008 (UTC)


 * Why not let Polbot run, and add the Polbotted images to a category that a human can check? If no-one has objected after 7 days, the images are removed from the category and kept. This is almost exactly the reverse of how Betacommandbot operates. I would even go so far as to say that Betacommand might like to spend time trawling through categories with thousands of images to see which ones can be saved. Oops, I meant "deleted". Carcharoth (talk) 03:36, 3 January 2008 (UTC)


 * Maybe the following simple change would help, as already mentioned above. Create a new Category: Category:Images with auto created FUR's or something similar. Polbot could simply add any image it added a FUR to be part of this new category, The images could be removed from this category after being reviewed. The limitations already put on this request will result in very few invalid edits. It would greatly reduce the work of creating proper FURs while creating a standarized form for this type of image. It turns the job into a two step process, but should resolve all concerns about errors.Dbiel (Talk) 02:28, 4 January 2008 (UTC)

Why is the minor possible error factor related to this bot (which by the way would still add a useful shell FUR that could be easily edited) such an issue when BetacommandBot is permittted to make stupid errors, (tagging files that have vadid FUR's as being invalid forcing other users to clean up his mess. see http://en.wikipedia.org/w/index.php?title=Image%3ALooney_Tunes_Back_in_Action_Game_Screenshot.jpg&diff=181611335&oldid=180882086 Dbiel (Talk) 16:54, 4 January 2008 (UTC)
 * One BCBot's edit rate is very very low. .000833333% is an error rate that is hard to beat. As for the requirement of a cross-check, its the only method of ensuring that the rationales are appropriate. βcommand 17:00, 4 January 2008 (UTC)

Break
I'd like it if we discussed BCBot elsewhere. I just want this bot approved. Can we please get this moved forward? It's been a long wait. – Quadell (talk) (random) 19:02, 4 January 2008 (UTC)
 * I Agree. I see no reason why this should not be approved. I added the reference to BCBot as a case for approving this one, not to get into a discussion over BCBot behavior here, it is going on several other places. Dbiel (Talk) 19:54, 4 January 2008 (UTC)
 * This is not going to be approved until checks are put in place so that the bot is tagging the right images. Make sure it's a logo, and it's a logo for the page you're making a rationale about. Otherwise, we can close this discussion right now. --uǝʌǝsʎʇɹnoɟʇs (st47) 19:56, 4 January 2008 (UTC)
 * Making sure it's a logo is easy. Making sure it's a logo of the subject of the page linked to? Well, as discussed above, this can be done with over 99% accuracy. And, as discussed above, it's impossible for a bot to know with 100% certainty. . . but it can create a list to be checked manually after the fact. You seem to be acting in an authoritarian manner, and ignoring the consensus here. That's not what BAG authority is for. – Quadell (talk) (random) 20:03, 4 January 2008 (UTC)
 * BAG is intended to make sure the bot works, and I'm not convinced that it does. Are you checking that the image is used on the page of a corporation and is at least somewhat related to that page, either by comparing titles or checking for the page name in the image description? --uǝʌǝsʎʇɹnoɟʇs (st47) 20:08, 4 January 2008 (UTC)
 * It might not be the logo of a corporation. It could be for an individual, a city, a concept, or anything else. This was all discussed above. If you're concerned about whether the bot works, why not authorize a trial run? Authorize, say, 50, and I'll go through afterwards and see if there were any mistakes. – Quadell (talk) (random) 20:26, 4 January 2008 (UTC)
 * Regardless, you need to be sure that it won't tag a screenshot or any other image with a logo rationale, or a logo that is on an article unrelated to the subject of the logo, or a logo on no pages or on more than one, or one used improperly. As far as I see, you aren't checking the content of the article it's used on, and you aren't checking whether the use is relevant. By adding the rationale, you're guaranteeing that the rationale is relevant to the use, the use is valid, the image is tagged correctly. --uǝʌǝsʎʇɹnoɟʇs (st47) 20:32, 4 January 2008 (UTC)
 * The proof would be in the trial run. So why not approve it on a trial basis only to see just what the reality actually is? Dbiel (Talk) 20:37, 4 January 2008 (UTC)


 * Question: would it be possible or reasonable for the script to run with manual approval? Gimmetrow 21:13, 4 January 2008 (UTC)
 * That would be fine if he wanted to run it under his own account, reviewing each edit, making sure the use is valid. That would in fact be better than a bot using the restrictions I stated above, as a user looking at the image, at the uploader's history if necessary, at the page the image is used on, can be sure of the accuracy of the rationale. --uǝʌǝsʎʇɹnoɟʇs (st47) 21:17, 4 January 2008 (UTC)
 * I'm really only good with Perl. I don't know enough JavaScript to make that work. My request is still for a perl app that adds rationales, as specified above, to logos only. Can I have a trial run please? – Quadell (talk) (random) 22:05, 4 January 2008 (UTC)
 * Not until you address the issues. --uǝʌǝsʎʇɹnoɟʇs (st47) 22:33, 4 January 2008 (UTC)
 * What issues haven't been addressed by Quadell? Gimmetrow 01:02, 5 January 2008 (UTC)


 * "This is not going to be approved until checks are put in place so that the bot is tagging the right images." - what was wrong with the suggestion to have a two-step process? Bot adds rationale - human checks it and approves it. You will find more people to do such checking than you will to write rationales from scratch. This directly addresses the concerns raised, and if Quadell can agree to put Polbotted images in a "auto-generated FUR" category, then I see no reason for not approving this bot (speaking as someone who works on images, not on bots). Carcharoth (talk) 11:31, 5 January 2008 (UTC)
 * I'm willing to check the bot's edits. As Carcharoth says, it would be far quicker to check and approve the bot's edits -- and fix the very few incorrect rationales -- than manually add tons of logo rationales. I'm sure that editors from the new Task of the day project would be willing to help too.  Bláthnaid  14:16, 5 January 2008 (UTC)
 * My problem with that is that the bot is proposing removal of the deletion template before the rationale is checked. --uǝʌǝsʎʇɹnoɟʇs (st47) 23:53, 5 January 2008 (UTC)
 * You seem to be misunderstanding the proposal. This bot is not designed to work solely on images tagged for deletion by Betacommandbot (ie. fixing images tagged for deletion without removing the deletion tag), but to fix both untagged images (so that Betacommandbot doesn't have to bother tagging them) and to fix tagged images (so that the admins or other editors reviewing the image deletion categories can remove the deletion tags if they agree Polbot has fixed the images). If the reviewing admins or editors find anything wrong, then they can delete anyway (or try and fix it properly). Or is your statement above based on reviewing the bot code? Carcharoth (talk) 01:57, 6 January 2008 (UTC)
 * There's no guarantee that each image will be fully reviewed to the extent that they should, or even at all. It's best that either the bot make sure of as much as is feasible. I realize that this bot isn't specifically following betacommandbot or any of the image bots, but it will inevitably run across a page tagged for deletion, and it should not undo that tag on the basis of a bot-generated rationale. Before this moves on, I'd like to see the code, so I can be sure of what checks are in place. --uǝʌǝsʎʇɹnoɟʇs (st47) 02:06, 6 January 2008 (UTC)
 * How about if the bot skips any image with a dispute template? Then check how it's doing, and if it's doing OK, expand the scope. (Quadell's proposal has received a lot more discussion than certain other much less well-thought-out proposals that were speedily approved.) Gimmetrow 02:23, 6 January 2008 (UTC)

It seems to me that some of us have a very different view of the fair use project than other do. Betacommand and some others seem to think the best thing to do is to delete any image that does not have a fair use tag. This appears to me to go contrary to the goal. As I understand the goal, it is to be sure that if a non free image is posted on Wikipedia that it is being used within the requirements of fair use. So far I have not seen any images tagged/edited by Polbot that have not been used under acceptable fair use rules. Yes there are problems with missing rationales. But the way to fix that is NOT to delete the images, but to and and fix the rationales. I have personally seen betacommandbot tag images with valid FUR's with an invalid tag. Polbot will help to make Wikipedia better by helping to add FUR's as needed. Betacommand seem more intent on making Wikipedia worse by deleting images that are use properly but lack the required FUR's. Now which approach is actually better in the long run? Dbiel (Talk) 03:01, 6 January 2008 (UTC)
 * Betacommand isn't trying to delete all the images, he's trying to make sure we comply with United States copyright laws.  Mønobi 03:03, 6 January 2008 (UTC)
 * Monobi - I've highlighted this because this is a common misunderstanding. Betacommand's tagging is intended to get images in compliance with the Wikimedia Foundation's Licensing Policy, not US copyright laws. The Licensing Policy is stricter than US copyright laws. Carcharoth (talk) 14:19, 6 January 2008 (UTC)
 * Granted that is his primary goal. But his method is to use the deletion of the image rather than the fixing of the missing rationale when the image IS being using properly. Polbot's approach is a much better way to accomplish the purpose. Dbiel (Talk) 03:09, 6 January 2008 (UTC)
 * The problem is whether or not an automated process can write a fair use rationale.  Mønobi 03:10, 6 January 2008 (UTC)
 * Like I have said, I would like to see this bot operational, if and only if there is a cross check. blindly adding rationales hoping that they are appropriate is not a good idea. if the bot does a category cross-check and confirms that the rationales are appropriate I have no problem with that. But blindly adding rationales and hoping that they are correct is not OK. βcommand 03:13, 6 January 2008 (UTC)
 * Making it a two step process should meet this concern. Simply add the images to a new category indicating that they need to be check. Is this any different in function from what Betacommandbot has been doing, tagging images for deletion that have valid fair use information and then hoping someone will fix the rationale format before someone else just deletes the image based on the invalid tag being added by Betacommandbot? Which method is better for Wikipedia in the long run? Also, keep in mind that Betacommandbot has already been approved for phase 4 auto deletion of images; a bad idea as far as I am concerned, but that is a separate issue not be be discussed here. Dbiel (Talk) 03:32, 6 January 2008 (UTC)
 * BCBot does not tag valid rationales. and BCBot does not have permission for image removal. BCBot does do a cross check prior to tagging. either implement a crosscheck or this task will not be approved. βcommand 03:36, 6 January 2008 (UTC)
 * That depends on how you define a valid rationale. BCBot does tag many images that are properly being used under fair use rules but only have technical issues with the fair use rationale that can be easily fixed. These are then far to easily deleted based on the tag rather than taking the time necessary to fix the rationale and remove the tag. Polbot's rationales are over 95% accurate greatly reducing the work of creating valid rationales a much better approach in my book. Dbiel (Talk) 05:18, 6 January 2008 (UTC)


 * What exactly is the cross-check you want? If it's not something fairly trivial (such as: the image is in Category:Company logos and the article using it is in one of the subcats of Category:Companies), it's not clear to me what it might be. Gimmetrow 05:13, 6 January 2008 (UTC)
 * that is what Im asking for. βcommand 05:15, 6 January 2008 (UTC)
 * OK, that doesn't seem unreasonable, though Quadell might find it too much work retooling for each logo category. Gimmetrow 05:39, 6 January 2008 (UTC)
 * Actually, Perlwikipedia has a get_all_pages_in_category function, which gets, as the name states, all pages in a category or subcategories. It would be easy to make that into a lookup hash and compare each link to the image to that list. It would also be trivial to check for the name of that article on the logo's description page, to make sure that the logo is related to the article it is used on. --uǝʌǝsʎʇɹnoɟʇs (st47) 12:58, 6 January 2008 (UTC)

Regarding the above: one problem is, as I've said twice above, the logo tag is used (correctly) on company logos, city seals, personal logos, idea logos, etc. Basically, a logo can be a logo for a thing, not just a company or whatever. And there is no check to make sure a wikipedia article is about a thing. It's not "too much work retooling for each logo category", but there simply isn't a logo category for each type of logo, and the vast majority of logo images don't use those categories. This would prevent this function from fixing most of the images that need it.

Also, ST47 stated "It would also be trivial to check for the name of that article on the logo's description page, to make sure that the logo is related to the article it is used on." But Image:ADC.gif is used in Apple Developer Connection, but outside of the rationale the name of the company is never mentioned. Image:IDC logo.png never mentions International Data Corporation. Etc, etc. Besides this, such a check would not fix any problems: the Image:Abc seal.gif image description page does mention National Space Society, even though that seal does not represent the NSS.

It was also suggested above that the bot skip images tagged as having incomplete rationales. This seems counter-intuitive to me. Should this bot only fix images that haven't been tagged as needing to be fixed? The impetus behind this suggestion seems to be that Polbot should never "save" an image from deletion if there's a chance, however slight, that the rationale would be incorrect. But this seems to misunderstand the Di-no fair use rationale tag. This tag should not be seen as promoting image deletion unless someone steps in and interferes. Instead, it should be seen as promoting that the image be fixed, and conceding that the image must be deleted if no one does.

With all three of these suggestions, I notice a common thread: they would severely limit the functionality of this bot, making it effective on as few images as possible. ST47 has taken it upon himself to delay and limit this bot as much as possible, despite its laudably low error rate and the willingness expressed by many people to manually check its edits. Most bots sail through the approval process with far less. I've never seen a bot with such a low error rate, so many safeguards, and so much community support, be rejected. But, to be honest, I'm tired of fighting this. It's obvious to me that the community wants this function run, that the bot will perform a positive function, and that this bot is being stymied for ulterior reasons by the BAG. I've listed my reasons, carefully and fully, above. If BAG will not approve this bot, despite community consensus, then BAG is broken. – Quadell (talk) (random) 15:53, 6 January 2008 (UTC)
 * I made some of those suggestions. I do, in fact, think this bot is a good idea, and I was trying to find a way to get it operable in a narrow-enough scope to overcome the objections, with the confidence that, once it's operating, the wider community wouild find the bot beneficial and desire its scope expanded. I was just trying to help. Gimmetrow 16:08, 6 January 2008 (UTC)
 * (No offense meant; I think your suggestions are good-faith attempts at compromise, to try to get things moving. I just don't think these are good reasons to reject the bot's operation. – Quadell (talk) (random) 13:49, 7 January 2008 (UTC))
 * And what was wrong with the "Category:Images with auto created FUR's" suggestion? Betcommandbot tags images for admins and other editors to check and fix (and remove the tag) or delete. Why can't Polbot do the same thing? Add rationales and tag (or replace an existing 10c tag) the images for admins and other editors to check and improve (and remove the tag) or note a false-positive and delete while informing Polbot's programmer of the false-positive? Carcharoth (talk) 19:26, 6 January 2008 (UTC)
 * I intend to do this (or something equivalent, such as including a blank template so "what links here" can be used), if approved. – Quadell (talk) (random) 13:47, 7 January 2008 (UTC)

It seems to me that as long as Polbot adds the new maintenance category there should be no reason to refuse a trial run. Remember an unauthorized run covering a much broader range of images was previously done with very few false edits. This just seems a much better way of fixing the problem, than Betacommandbot's tag and delete method. Polbot actually adds a useful FUR than can be easily edited where Betacommandbot only tags the images for deletion. So just what is the real goal here? To tag and delete images or to add missing rationales to images that are being used properly? Polbot has already shown that it works effectivly in adding FURs that make sense. Dbiel (Talk) 20:34, 7 January 2008 (UTC)

It is hard to understand why a bot that provides useful and accurate edits to assist in the major problem of dealing with images that do not have fair use rationales but are actually being used under fair use rules continues to be stonewalled. As a related note Betacommand reverted over 80 edits in 36 minutes, which I am slowly working my why through and reverting back as so far I have not found one that was reverted by Betacommand that there was any justificaton for reverting the FUR other than the simple fact that it was made by an unauthorized bot. That does not seem to be in the best interest of wikipedia in my book. Dbiel (Talk) 21:18, 8 January 2008 (UTC)

Section break
I believe all concerns have been addressed above. – Quadell (talk) (random) 22:20, 9 January 2008 (UTC)
 * have you enabled a category crosscheck? βcommand 04:28, 10 January 2008 (UTC)
 * For what category? As I stated above, multiple times, a logo can be a valid image to use on the page of any thing that the logo represents (company, band, individual, concept, city, etc.), and there is no way a category can determine whether the article is about a thing. – Quadell (talk) (random) 13:29, 14 January 2008 (UTC)


 * one good starting point is Category:Companies βcommand 16:10, 19 January 2008 (UTC)
 * Quadell, why not compromise on this? Try and do a category crosscheck, and gradually expand the crosscheck if it works? Or is it technically difficult to do such a crosscheck? Carcharoth (talk) 18:56, 19 January 2008 (UTC)
 * its very simple to do cat cross checks. βcommand 18:57, 19 January 2008 (UTC)
 * Let me rephrase that. Betacommand, would you be happy to work with Quadell to help him get Polbot to do category crosschecks? Carcharoth (talk) 19:31, 19 January 2008 (UTC)
 * I dont speak pearl. βcommand 19:32, 19 January 2008 (UTC)
 * ROTFL! Bot owners with language difficulties. As a nonprogrammer, I find that very funny. Carcharoth (talk) 21:59, 21 January 2008 (UTC)
 * Carcharoth, the trouble with adding cat cross checks is that the bot will have the exact same effect either way. Consider: if I first run it only for Category:Companies and subcats, what will that accomplish? It won't limit it to companies -- subcats include Category:Books about companies, Category:People by company, Category:Lists of awards by company, etc. And it obviously won't cover most valid logos. So then what? I can run it again for Category:Cities and subcates (including such non-cities as Category:City founders and Category:Urban warfare), for Category:Political parties and subcats (Category:Political party songs, Category:Leaders of political parties), etc., etc. Eventually we'll be back to where we are right now, with logos for other things being tagged, and deleted, and the only way to fully cover them would be to run this bot without category restrictions. Why go through all the work of adding category restriction (that won't restrict the bot effectively anyway) only to run it without category restrictions in order to cover all bases later?
 * It's obvious to anyone paying attention that BAG is engaging in a filibuster. They'd rather not deny a bot with obvious community support, but they don't want to approve it, due to BetaCommand's vested interests. So they drag this out as long as possible. It's been a month now (and 73 kilobytes of discussion). It's a shame, since I really don't have any recourse. If I run the bot without BC's approval (and ST47's approval), good faith admins will block the bot for running "unauthorized", regardless of how much community support it has. You've got to hand it to BetaCommand: with the support of just one person, ST47, he's managed to exercise unassailable control over making sure that no bot can prevent images from being deleted by improving their rationales. Can the community overrule a BAG member? It doesn't look like it. Can the community remove a BAG member from having this authority? According to Bots/Approvals group, the only way a BAG member can be removed is by "consensus of existing members". Unlike ArbCom, Mediation, and any other group, BAG operates by consensus of its own, not by consensus of the community, though it has real power over the community. The community's support doesn't mean anything; the other opinion that matters is those of the existing in-group. BAG is broken. – Quadell (talk) (random) 00:02, 21 January 2008 (UTC)
 * There is a reason that BAG does not approve this request. Blind addition of rationales defeat the purpose of them. If you cannot accept that you need to check the image is used on a company page, this bot will not be approved. I pointed to a supercat as an example of one that you could  build off. you cant blindly use all subcats either, you should review the cats and decide what is appropriate subcats to include. Like I have stated I like the idea behind this bot, but you need to do a cross check, blindly adding rationales defeats the purpose of the rationale. you have two choices, implement the crosscheck or withdraw your bot request. βcommand 01:54, 21 January 2008 (UTC)
 * But what would be wrong with handling this as a two step process. Step one, the automating addition of a FUR and adding the page to a new category. Step two manually checking the pages in the new catetory and deleting the category entry from those that have been edited correctly (which from what I have seen of Polbot first unauthorized run would be well over 90% and probably closer to 99%) This would great aid in the real goal of making user images have the required fair use rationale. The purpose of a bot is to reduce the work required to be done by humans and Polbot does an excellent job of that with very few mistakes. I find it a much better method that the one being used by BetacommandBot which does not reduce any of the work required for adding the correct FUR and only makes it easier to find those images needing a FUR but does so by adding a 7 day delete notice which is impossible to keep up do to the vast number added at one time, and as a result, some admins simply blindly delete images that are properly being use for the sole reason of cleaning out the category after the 7 days have past. Dbiel (Talk) 16:48, 21 January 2008 (UTC)
 * Quadell, how would Polbot handle the images in Organic certification? This is why I believe a two-step process is needed. Betcommand, why can't you accept a two-step process? Bot tags - human approves. This is no different to your bot. It tags - and an admin approves the deletion. If, after a week, the image is still in the "unchecked bot rationale category", then Polbot goes back and undoes its own edit. Carcharoth (talk) 18:17, 21 January 2008 (UTC)
 * In some cases, a category cross-check (per Beta) isn't unreasonable, but I don't think it can be done automatically except in some limited cases. The non-free images in Organic certification are nice examples. As far as I'm concerned, a two-step process is great, though it should stay in the "unchecked bot rationale" category for a bit more than a week. Gimmetrow 19:42, 21 January 2008 (UTC)
 * This bot should not be an attempt to save all logos. the ability to confirm all logos needs human judgment for some cases that are not obvious, Organic certification is a perfect example bots should not be creating rationales for those images, they are not cut and dry usage. Carcharoth that is an interesting idea about reverting itself after a period of time. that would reduce some of my worry that once polbot tags them they will not be reviewed at all and just sit in yet another backlog that rarely gets worked on. βcommand 19:52, 21 January 2008 (UTC)
 * The question whether to keep a logo should be done by a human. Equally, the decision to delete a logo should be made by a human. At the moment, with the volume being deleted, I doubt all the logos are being reviewed before deletion. Also, even when the bot-checkable NFC criteria have been checked, the question of how to check the other NFCC will still remain. I have made several suggestions on how this could be done. Would you (Betacommand) be interesting in working with me and others on that? Carcharoth (talk) 21:58, 21 January 2008 (UTC)
 * Carcharoth, Ive got a few ideas that you might be interested in, but I just need more time to finish them. βcommand 03:43, 22 January 2008 (UTC)
 * OK. Drop me a note when you do. Sorry, back to the discussion. Carcharoth (talk) 11:53, 22 January 2008 (UTC)

New section break
So seriously, given the above, why is this request being denied? I'll state here and now that I will happily confirm such edits by Polbot, and nominate any non-notable articles at AfD (or merge non-notable cases to lists). Well, I might just make a list, but still, I agree that not all these stubs about companies need logos. It looks like advertising until a more substantial article appears. Having logos for Joe Bloggs pizza parlour is embarassing. Carcharoth (talk) 11:53, 22 January 2008 (UTC)
 * Im waiting for a response about the bot reverting itself. and a full response from the operator. βcommand 16:24, 22 January 2008 (UTC)
 * I don't think the bot reverting itself is a useful feature. The bot could be removing a valid rationale from an image, causing the image to be deleted. – Quadell (talk) (random) 19:32, 22 January 2008 (UTC)
 * So just rap what ever you tag in HTML comments and when you revert yourself just remove what ever you added between the HTML comments. βcommand 19:38, 22 January 2008 (UTC)
 * I don't understand your previous sentence. – Quadell (talk) (random) 19:53, 22 January 2008 (UTC)
 * Oh, I see what you mean. I think it would be irresponsible for Polbot to "revert herself", if reverting means (in 99% of cases) removing a valid rationale from an image, bringing into non-compliance. – Quadell (talk) (random) 19:58, 22 January 2008 (UTC)
 * I am attempting to work out a compromise with this bot, there have been several ideas that have been suggested, you ignore all of them, What I am saying is that you either need to implement a category crosscheck or self reversion after certain time period. the best compromise that I have seen is the self reversion, as I also think that blind addition of rationales is irresponsible, and is against our non-free policy. βcommand 20:16, 22 January 2008 (UTC)
 * The self revert sounds like a good idea to me, as long as the time frame is long enough to allow for manual checking. The problem with BetacommandBot is that it tags over a 1,000 images at one time with a 7 day warning which does not allow enough time to review the tagged images within that 7 day window resulting in image deletions by some admins solely based on the tag which is a bit unfair when the image may have been uploaded over a year ago when the rules were different and all that is needed is an updated FUR with all the necessary information already available on the image page. To delete an image simply because the FUR does not contain a link to the article is stupid. Dbiel (Talk) 20:33, 22 January 2008 (UTC)
 * I think it was a questionable decision to allow BetacommandBot another run without first approving this bot - by doing so a large amount of work that could have been automated / semi automated is going to have to be carried out manually. In order to progress this request, would it be possible to allow a short trial run, and from there establish the error rate? If the error rate is low, around 1% for example, then I don't think an automatic revert should be necessary. Addhoc (talk) 19:04, 15 February 2008 (UTC)
 * Marking request as expired as it was either never transcluded to BRFA or has had no attention for some time.  Richard 0612  12:21, 19 February 2009 (UTC)


 * The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made in a new section.