Wikipedia talk:Old IP talk pages

level 4s
One of my concerns is that IPs are not static like phone numbers, after a certain number of years it becomes seriously unlikely that the current user of an IP has any connection with the editor who was warned, other than using the same ISP. So I don't see any point in keeping even level 4 warnings and block messages unless they are from recent years. Am I wrong in my assumptions about IPs? Or is there another reason why we might want to keep these?  Ϣere Spiel  Chequers  20:46, 10 April 2012 (UTC)


 * Hmmm, how do you define "recent years"? And would making the deletion (or preservation) conditional on length of time since the actions plus the level of warning complicate matters more than it simplifies them? I worry about creating a "rules maze." --MZMcBride (talk) 20:53, 10 April 2012 (UTC)
 * If I'm correct in thinking that a three year old IP is unlikely to be associated with the same address then in my view the level of warning is irrelevant, what we should focus on is the likelihood that the IP has been reassigned. Someone must have done some research on that. My view is that we tap into some such research and have a fairly simple rule - if the IP is unlikely to be associated with the same person then we blank or delete the talkpage as irrelevant. If as a hypothetical, the figures are 50% of IPs get reassigned within 12 months 75% within 24 months and 90% within three years then I'd go with blanking at three years, but even blanking at 6 years when we are 99% confident that the warnings are now associated with a different person would be progress from my point of view.  Ϣere Spiel  Chequers  21:34, 10 April 2012 (UTC)

WereSpielChequers. You are absolutely right. But the point is the other way around. If a user on a range of IPs in 2010 is spamming 'someobscurecompanywebsite.com' and gets a good handful of warnings, and is warned for that repeatedly, some get blocked and they stop before it gets worse enough to be blacklisted, then that is very likely the same editor that in 2012 is spamming that same 'someobscurecompanywebsite.com' from another IP - it is at least, even very broadly construed, a meatpuppet of that first user from 2010 - that user, or the company that hired the two employees or SEO companies who maybe even independently choose to spam Wikipedia, should know better, the company should tell their people to not spam Wikipedia (although they may think that it is so long ago, that they can have another go at it). If those talkpages on the first range are deleted, then those tracks of warnings and blocks are hidden - only admins can find that that user/company has had a string of warnings before. Of course, on that previous range there may be a completely new set of editors for which those warnings are irrelevant, for those who are looking into systematic forms of vandalism, the fact that those warnings were issued is not. Those warnings don't have to be on the talkpage (I am fine with blanking/archiving), but they should be readily available in the history (so that it does not come down to admins to follow the tracks). If those tracks are visible, I would in this case start of with uw-spam3 at the least, or even a uw-spam4im - no need for good faith there anymore, someone from company X was spamming in 2010 and was sufficiently warned for that, someone from company X is now spamming. I might even consider to now very fast go to the spam blacklist if the link is crap anyway - this user should know better, and every user can show that the user should know better (yes, I am a great fan of transparency on Wikipedia).

Say that I give you a range of IPs who have been involved in systematic vandalism in the past, do you have any idea how difficult it is to find which of those editors are warned (you have to go to ALL the talkpages, and see whether they have been deleted in the past - that work gets worse if some have been recreated), and that it is impossible for those without admin bit to see whether those editors have been warned for those actions (and that it was not for something else)? At that moment, it becomes impossible to choose whether to apply WP:RBI or WP:AGF - say that there was a range but they were never warned; you'd really have to assume that they, in good faith, had no clue that they were spamming in 2010. Sure, that IP is (likely) used by someone else - but the spammer, NPOV pusher with a specific MO, etc., is, with very, very high confidence, a meatpuppet or sockpuppet of the editor that was using that IP 2, 5, 10 years ago. And add to that, that Wikipedia has a high retention rate on spammers and POV pushers. The former make money with it (I can show you the examples of spammers who lay dormant for years, or spammers who come back weeks after their links were de-blacklisted), and ArbCom is blocking editors on the (apparently) faint suspicion that the editor is a sock of a banned editor, how long have some long term abusers been active? Do we really have to hide the tracks of possible others long term abusers who have not been recognised as long term abusers yet? Then it will all come on the shoulders of the admins, while there is such an easy solution: blank those, delete the rest.

I am very confident with blanking or archiving (or replacing with a template which has somewhere a message 'there are older revids in the history of this page'), even after a very short time (2 weeks, 1 month?). And for non-static IPs, even deleting the really non-systematic ones after 1-3 months is also fine (a one-off poop-vandal may never return on the IP, never see the warning, and never engage in that type of vandalism again because they grew up (I hope)). There is no need to bother the new user with the warnings, messages or discussions, even if it was a spammer, school, or POV pusher that was using that IP at that time. A friendly welcome may there even have more effect than a page full of warnings! --Dirk Beetstra T C 05:44, 11 April 2012 (UTC)


 * MZMcBride worries about a "rules maze"? I thought that the rules would be interpreted automatically by a script, and so the problem is not that the rules are several, but that every rule is easy to evaluate.


 * I came to suggest another rule: "No IP talk page may be deleted if the IP has made >100 edits."

Reason: 100 edits (a semi-random number choice) indicates an IP of interest to the project of a decent level of seriousness, whether good or bad. It doesn't really matter if it is one person, or a string of people on a common room computer.


 * I think this would be a conforting rule to skeptics, and I suspect that it would cover an extremely small percentage of Old IP talk pages that otherwise would be deleted. --SmokeyJoe (talk) 06:04, 11 April 2012 (UTC)


 * I am afraid that the rules can not be interpreted script-wise .. that is why I am not confident that this can be a CSD criterion.
 * Hmm, there are some IP-ranges which seem to attract vandals, not sure if 100 edits (or 1000) would be any measure. --Dirk Beetstra T  C 06:25, 11 April 2012 (UTC)
 * I agree that the number of edits an IP has made is irrelevant, what matters is the likelihood that the same user is still using that IP; if the gap is so long that one can assume the IP has been reassigned then it is irrelevant whether the previous user of that IP had 1 edits or 1,000.  Ϣere Spiel  Chequers  21:11, 11 April 2012 (UTC)
 * If an IP was ever, even for a short time, an active contributor, then the talk page should not be deleted. Even if the talk page contained no significant content, that fact alone may be of intetrest when reviewing the IPs contributions.  --SmokeyJoe (talk) 23:53, 11 April 2012 (UTC)
 * @WereSpielChequers: No, it matters whether the physical person who was behind that IP has gotten a message some time ago, and who now may pretend, while using another IP, that they did never get a message. If you follow the requests for de-blacklisting on Wikipedia/meta, every now and then (once a month, approx.) there is someone there complaining that their link has been blacklisted, and that they were not informed.  It then sometimes turns out that someone actually spammed said link, using a plethora of accounts and IPs, and that many editors spent a lot of time warning said editor.  They are now indeed on another account/IP, an account/IP that did not get warnings, but playing dumb is a technique that is used by these coi spammers.  The point is not that that message is irrelevant to the user who is using the IP now (and archiving/blanking would just 'wipe' that message), the point is that the message was relevant to the physical person who was using that IP, and who is now maybe using another IP, and thát info is of importance to us.  --Dirk Beetstra T  C 05:26, 14 April 2012 (UTC)

Spamming behaviour recorded on IP talks pages is pretty sparse data
Noting Dirk's interest in using old IP talk page warnings to track long term spammers, I suspect that it is a method that could be improved upon. I think that much spam is removed without warnings being pasted on IP talk pages. Much spam, especially intelligent spam (eg a website directly related to the subject), is removed somewhat later, without necessarily investgating which account/IP added it. Doesn't committed, money-making, spamming invariably involve inserting external links? Wouldn't it be relatively easy for a bot to search active IP contributions for the addition of external link additions, and to record the number of external links added as a fraction of the IP's editcount?

Dirk doesn't want to see IP talk pages deleted, because he looks for patterns across multiple IPs. If the data he wants were already filtered, than maybe this disagreement could be resolved? --SmokeyJoe (talk) 06:04, 11 April 2012 (UTC)


 * I am readily invoking XLinkBot (as do others like Hu12 and MER-C), and many others active in the anti-spam field are 'tagging' talkpages (users Hu12 and MER-C do a lot of that, I do it sometimes, there are others who do it as well). Of course, many spammers are not tagged - but for those goes: apparently we did not try to communicate with them or we did not try to warn them sufficiently.  Those tracks do not exist, and I might be out of line when I would blacklist their links or block their accounts if there were not earlier attempts to warn them (we only would do that then for really blatant cases - tramadol/sildenafil/phishing sites e.g. - a form of RBI, Revert, Blacklist, Ignore; no warnings needed).  This is basically just the point.  If they were warned, there are on-wiki tracks, if they were not warned, there are no on-wiki tracks.  If the tracks get deleted (even after 5 years), then it appears as if there are no on-wiki tracks.  Having tracks generated after some time (as you suggest) is not the same as warning an editor in the act (XLinkBot should warn within minutes, an IP/editor is likely to get the orange banner then, if we tag a page months later they are likely already somewhere else, even if they use an account).
 * So, I am not really using the talkpages to track the spammer, I use the talkpages to track whether the spammer was warned - and the same would go for NPOV editors, or other forms of, what I would call, systematic vandalism. Spam just makes the most obvious example, but similar would go for the early accounts of Grawp - we tried to warn them, it proved futile, so then we ban; if then would delete the early attempt to stop them by warning them (before we get to the stage that we realise that banning is the only option), another editor finding the vandalism is likely going to keep on warning (assuming good faith), not realising that maybe it is really time to go to the next stage, prolonging the abuse while blocks would already be really due.  --Dirk Beetstra T  C 06:41, 11 April 2012 (UTC)
 * Oh, by the way, I am filtering it. --Dirk Beetstra T  C 06:42, 11 April 2012 (UTC)
 * OK, clearly you know what you are doing, and it is beyond my skills. Please excuse my clumsy inputs when I start to get out of my depth.  --SmokeyJoe (talk) 23:55, 11 April 2012 (UTC)
 * No need to apologise. I actually value input from others who have a different view, it may enable other ways of thinking.  --Dirk Beetstra T  C 05:20, 14 April 2012 (UTC)


 * I'm not involved in spam filtering so apologies if this is old ground, but if you are working at the IP range level, would it help if IP ranges had talkpages as well as IPs?  Ϣere Spiel  Chequers  10:00, 11 April 2012 (UTC)
 * Difficult - if they are shared there would be a lot of innocent users getting the message. However, it may get the message through to the people who should read it, more than what happens now sometimes (the vandal is already using another IP and the talkpage message is never read).  --Dirk Beetstra T  C 10:05, 11 April 2012 (UTC)
 * I was thinking about a page for our use, not necessarily for the IPs to see. Unless of course there was a range block in force.  Ϣere Spiel  Chequers  12:13, 11 April 2012 (UTC)
 * For that, probably any sandbox would be fine: User:Beetstra/127.0.0.0/24? --Dirk Beetstra T  C 12:19, 11 April 2012 (UTC)
 * But maintaining such pages for all the incoming spam is probably a pain. We are talking on 775 780 wikis about 100 96 external links added per minute .. - of course, most of that is fine but it gives an idea of the sheer size of what we are looking at.  And then, other ranges may have a NPOV editor on it, or a serial copyvio editor ...  --Dirk Beetstra T  C 12:21, 11 April 2012 (UTC)

One Year Limit
The first criterion described under Page Deletion states that pages can be deleted if they are "Stale IP talk pages where the IP has not edited or had any edits on their talk page in over a year". I disagree with this, because that would include pages of IPs that have been blocked for a year or more, of which there are plenty on Wikipedia. This would mean that a serial vandal with a long, involved history of vandalism that has suffered long-term blocks could just have all their warnings and block notices removed shortly after those long-term blocks expired, thus making it easier for them to engage in vandalism again. Not a good idea. Nightscream (talk) 22:35, 23 December 2012 (UTC)


 * No, because if the editor is blocked, or the editor is involved in long-term abuse (even shortly), the page should not be deleted anyway - those are important tracks. --Dirk Beetstra T  C 06:56, 5 February 2013 (UTC)

Why don't we create a bot that removes old IP talk pages?
If nothing has been written on a user talk page for a certain amount of time, say one or two years, and there is nothing of value on the talk page (maybe just a welcome template or maybe even just a user warning), why isn't the page removed automatically? Can't we have a bot doing this? —Kri (talk) 11:54, 18 January 2013 (UTC)


 * Practically very difficult to do, a bot would have to check whether the editor removed warnings which are of older interest, and see whether these warnings are significant - e.g. many XLinkBot warnings regarding YouTube videos are irrelevant, however there are some cases of YouTube spamming - if the editor was spamming, the page should not be deleted, however, the rest could be. A bot can not make that call, only a knowledgeable user.  Sometimes users get tagged as suspected socks - and we should not delete talkpages of sockpuppets, however if the suspicion is wrong, the page could be deleted.  Simply, there are no 'machine readable' conditions, and no-one should bot-wise (or like a bot) delete them.  There are no really big reasons to delete old IP talkpages; the only reason to delete would be that the old warning is not of interest to the user who is now using the same IP, but that could also be achieved by archiving the page after some time and replacing it with a carefully worded welcome template.  Also, some IPs were assigned to different people in the past, and are still - having the tag that it is such an IP is not bad to have.  A bot archiving the pages would be fine, and I thought that we had such a bot?  --Dirk Beetstra T  C 06:55, 5 February 2013 (UTC)
 * Oh, I should add to this, that sometimes there are interesting discussions on IP talkpages (or in old revisions of these) - discussions that helped the wiki forward. Such pages don't have warnings on them, still the discussion may be of interest.  How would you make a bot consider that?  There are no warnings, so just delete it?  --Dirk Beetstra T  C 06:57, 5 February 2013 (UTC)


 * A-ha. Seems complicated. —Kri (talk) 16:33, 5 February 2013 (UTC)
 * That's why we don't have a bot doing it :-D. Bot blanking would be fine though.  --Dirk Beetstra T  C 11:50, 6 February 2013 (UTC)
 * If I were to filter pages manually then have a bot automatically blank the pages would this be acceptable? The bot would only blank pages, it makes no "decisions" but because it has the bot flag it could do so without creating the "new messages" banner. I could make a bot that does that in AWB very easily. James086 Talk  18:49, 13 February 2013 (UTC)


 * I actually thought we had something like that .. but I think that would be acceptable (even for IPs used by a spamming, banned, blocked sockpuppet), and I think it should be a bot in 'minor edit mode', otherwise the 'you have new messages'-bit gets reset and the 'new' user would be unnecessarily drawn to the talkpage. For me, everything on a talkpage of an IP that has not been used for over 3 months and seems non-static could be replaced by a template, regardless what is in the history of that account (and it is easy to revert anyway).  It should however take care of already placed templates (for example, what I think is sharedIPedu, e.g.) - that would be good to keep.  Or maybe the time to blanking should be dependent on what is behind the IP (an organisation's page like a school could be kept longer than that of a random ISP).  --Dirk Beetstra T  C 06:08, 16 February 2013 (UTC)


 * The way I was intending to run it was to blank if the following criteria were met:
 * Take the list from here: Database reports/Old IP talk pages (note the criteria for the generation of that list are: the IP has never been blocked and has not edited in the past year and where the IP's talk page has not had any activity in the past year, has no incoming links, and contains no unsubstituted templates)
 * If there is a conversation with another editor present: skip
 * If there is only a welcome template: skip
 * If there is a long history of warnings: skip
 * Otherwise: blank (of course if there's something not covered by these, it's still me deciding whether to blank so I have human judgement)
 * Were you thinking the bot should leave OW (or similar) for any editors that come across the page later? James086 Talk  11:21, 16 February 2013 (UTC)


 * Similar. I am thinking of the scenario of an editor who edited using said IP, now moved on, but did trigger any form of a 'message'.  However, they did never see that message, so the new message flag is set.  Now you come along, and because the IP did not edit for a year, and all other criteria are met, you blank the page.  Now someone pops up and uses that IP.  He will see a new-message banner and think 'hey, curious', and go to the page, but find nothing.  That might be very confusing.
 * I would consider having a template that is slightly more informative than OW for those cases (though I don't know how to word it). That template should then include the same message as OW, I think.  Thoughts?  --Dirk Beetstra T  C 12:10, 16 February 2013 (UTC)
 * Note, this is also fine for long laundry lists of warnings, no use to leave 50 warnings on those pages (I do sometimes blank those, noting the old messages and add fresh ones, even on regularly used pages), even for sockpuppet pages. Those lists are particularly bitey for innocent new editors who now start using such IPs, and I thought that that was the whole thing that we tried to avoid.  --Dirk Beetstra T  C 12:13, 16 February 2013 (UTC)

Just an idea at this stage but is this closer to what you had in mind? Also I think this template should be substituted when applied so that it doesn't block the database report from picking up the same page a year after another warning is given. James086 Talk 13:04, 16 February 2013 (UTC)

Hmm .. I was more thinking:

That database report is strange. If there are unsubstituted templates, but 2000 warnings, then I would say, blank all the warnings if that IP was not used for a year, leaving the unsubstituted stuff. Also, when this template is there, and then 50 more warnings are added, it would not be picked up .. and not be blanked. That database report should just report all IP talkpages that have not been used for over a year. And on that one should do a consideration. With blanking it does not matter at all whether the user was blocked in the past or not, the page could still be blanked as there is no history that goes lost. One could even consider to blank everything after 3 months (i.e. 3 months after last edit or 3 months after last block expired, whichever is later), as long as the pages do not get deleted. We might want to think about what to do with those typical things like spam-tracking templates, sockpuppet-tracking templates (for those, archive in stead of blanking, or collect them in a collapsible box??). There is absolutely no reason to bother the current user of that IP with it, though having the tracks is handy every now and then.


 * here e.g. the sharedIPEDU-type templates

--Dirk Beetstra T C 13:30, 16 February 2013 (UTC)


 * I prefer your wording (messages vs warnings). I would be using that database report only because it is available (it's left over from when people were deleting based on this guideline, now essay). I can ask MZMcBride/Legoktm for another report that is less restrictive if this goes ahead. Also I won't have the bot adding the more complex headings because I'm a very limited programmer. Complex pages I would handle manually. James086 Talk  14:11, 16 February 2013 (UTC)

Which Speedy Category?
If an IP talk page meets all the criteria for deletion described in this article, which Speedy category would you recommend using to have the page deleted? Robman94 (talk) 22:30, 4 February 2013 (UTC)
 * As I have said above, this is never a CSD-type of decision, but I presume it would be a 'general cleanup'-type of delete, preferably with a properly chosen edit summary when applying the tag - otherwise it becomes a group-MfD (no need to do one by one, find 100 and do it). I still don't see the need to delete, blanking (or archiving) and replacing it with a nice welcome text is very much doing the same without any chance of losing important information, and it also takes away the, maybe scary, old warnings or similar. --Dirk Beetstra T  C 09:33, 5 February 2013 (UTC)
 * I see your point, so maybe the article should be modified to discourage deletion in favor of page blanking. Robman94 (talk) 17:25, 5 February 2013 (UTC)
 * It already does: "If you have any reason to believe that the talk page will one day be valuable, err on the side of keeping the page (consider to blank the page, and replacing it with an appropriate 'welcome' message - referring that there may be important messages in the history of the page)." --Dirk Beetstra T  C 11:46, 6 February 2013 (UTC)

Removing old warnings from IP talk pages
There is a discussion at WP:BTR on the approach to have a bot to do this work. 103.6.159.84 (talk) 06:08, 25 January 2017 (UTC)