User talk:West.andrew.g/Archive 4

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

Archive 6

→

Archive 10

STiki user warning's edit summary is unconfigurable and too strong

I see a problem with the user warning messages produced by STiki. The problem is with the edit summary, which (as far as I can tell) is always "User warning for addition of test/vandalism found using STiki". My concern is with the use of the word "vandalism": many people are very sensitive to having this term applied to their editing — and if there is any valid question on the motivation of a given edit, we should probably defer to the Wikipedia vandalism policy, which explicitly says that "Even if misguided, willfully against consensus, or disruptive, any good-faith effort to improve the encyclopedia is not vandalism."

I believe it would be far better for STiki and similar tools to err on the side of caution. I do appreciate the way the edit summary text for the actual reversion of article text can be configured (and, indeed, I have chosen to modify this text to call the material I am reverting "inappropriate" rather than "vandalism"). But the edit summary text for the user warning does not appear to be configurable in this way (or any other way, short of modifying the Java source) — so I've personally decided to disable STiki's user warning entirely (via the checkbox) and send any user warnings via the far more flexible Twinkle tool. Richwales (talk · contribs) 23:18, 20 July 2011 (UTC)

What would you suggest the wording be? I am open to suggestions. However, given that a STiki user has explicitly identified an edit as vandalism, I personally don't see a major issue with labeling it as such (but I would be willing to change the default summary). If there are "valid questions", shouldn't the edit be labeled as "pass" or handled via off-STiki channels (as you seem to do with all edits)? See also User_talk:ClueBot_Commons#Why_no_follow_up.3F. Thanks, -AW

I would suggest "User warning for addition of inappropriate material", or "User warning for unconstructive editing". Calling a reverted edit "test/vandalism" is, IMO, especially wrong because it implies that the editing might have been vandalism, or maybe just a test, or maybe the reverter can't tell — but my understanding of WP:VANDAL is that people shouldn't be calling something "vandalism" unless they are really, really sure it's vandalism. If the STiki user warning message's edit summary were visible and configurable in the GUI, the way the edit summary for the article text reversion already is, I wouldn't be as concerned. STiki is, in my view, highly useful for identifying and reverting misguided, mangled, or disruptive edits, even if they are not indisputably (or indisputably not) "vandalism" — things that definitely need to be reverted ASAP, and why not let people revert them using this good-quality tool? Richwales (talk · contribs) 03:15, 26 July 2011 (UTC)

Give me a day or two, and I'll distribute a new version (I prefer "unconstructive"). I probably have some other loose ends and efficiency modifications I'd like to push out at the same time. Thanks, West.andrew.g (talk) 03:28, 26 July 2011 (UTC)

Done, per today's re-compile. Thanks, West.andrew.g (talk) 15:06, 1 August 2011 (UTC)

Wikimedia Stories Project

My name is Victor Grigas and I'm a Storyteller at the Wikimedia Foundation. We're exploring new ways to explain why Wikipedia is so special and we’ve started a Wikipedia Stories Project, where we’re chronicling the inspiring stories of the Wikipedia community, especially editors and active contributors in the movement like you. I'll be traveling to Wikimania next month to collect stories for our 2011 Fundraiser. While there I'd love the chance to meet with you and hear your thoughts about Wikipedia. We’ll have a schedule of available times for you to sign up if you’re interested, but right now, we’d like to make the initial contact to gauge your interest. Please let me know by emailing me at vgrigas@wikimedia.org or responding on my talk page.

thank you,

Victor Grigas (talk) 18:54, 21 July 2011 (UTC)

STiki queues edits from approved bots?

EmausBot has come up quite a few times when I was using it. Can STiki not queue up edits from accounts with the bot userright? --Σ ^talk_contribs 01:44, 7 August 2011 (UTC)

Valid concern, I thought I had an explicit rule to prevent this -- apparently not. I am currently enjoying the great sessions at Wikimania 2011, but I will look into this ASAP and report back here. Thanks, West.andrew.g (talk) 03:16, 7 August 2011 (UTC)

In fact, the rule was not in place (but should have always been). It's now been fixed and the change is active on the STiki server (I am not going to push out a new source-code version for the minor change). A few bot-edits may still pop-up, however, if they existed in the queue prior to the time of the change. Thanks, West.andrew.g (talk) 15:08, 9 August 2011 (UTC)

I greatly appreciate your quick response, and will enjoy fighting leftover vandalism with your tool with no worry of reverting an approved bot. Thank you. --Σ _{^talkcontribs} 23:26, 11 August 2011 (UTC)

STiki level 1 warnings

Hi there, thanks for all your hard work and research that has resulted in this great tool!

I've noticed a minor issue with the level one warnings STiki leaves on user talk pages. Specifically, it says "Thank you" twice at the end (see this example). Also, it inserts two hyphens before the signature, which looks a bit silly for those of us who include an em-dash at the start of our signature already. —Darkwind (talk) 02:28, 15 September 2011 (UTC)

Thanks, Darkwind. I've removed the two hyphens, and removed the extraneous "Thank you", which seems to be the result of this recent template change. However, since the issue is so minor, I am not going to rebuild and push out a new version immediately, it might take a couple of days. Thanks, West.andrew.g (talk) 05:30, 15 September 2011 (UTC)

Spam detection tool

I read your paper on the spam detection tool. I would like to suggest two new features. The first, is whether a user has added a single link to multiple articles. The second, is how similar the IP of the landing site to the IP of the user adding the link. Sole Soul (talk) 06:51, 15 September 2011 (UTC)

Thanks for the read. The first one sounds viable, although it would seem to overlap a good significantly with the "times URL added in last [x]". The second one is a good one, and something already done by User:COIBot, so I didn't feel the need to duplicate the effort (academically), but it might be something worthwhile for practical use. Thanks, West.andrew.g (talk) 08:14, 15 September 2011 (UTC)

Hello, from CLEF

As promised. West.andrew.g (talk) 19:54, 19 September 2011 (UTC)

Hello there! -- とある白い猫 ^chi? 12:07, 20 September 2011 (UTC)

STiki Queue repetition

Andrew, I installed STiki (2.0) yesterday and started using it with the Cluebot-NG queue. Later, I changed to the WikiTrust queue. Then I closed STiki and later re-opened it. It started up, using the Cluebot-NG queue, which I used for a while. Then I switched to WikiTrust and noticed that all the revisions that were presented to me were the same revisions that I had seen the previous time that I was using the WikiTrust queue (and had already marked as innocent). If this isn't the right place to report possible bugs, I apologize. I have also run into a scrolling problem (under Windows Vista). That said, STiki is a very powerful tool. I am already a fan. Thanks. Peter Chastain (talk) 18:40, 21 December 2011 (UTC)

Hi Peter, I'll look into this bug shortly. I will run a trace of your classifications and see if, indeed, some of the revisions do show up twice. This is pretty core functionality for something to go wrong and unnoticed for so long -- although maybe it is specific to the WikiTrust queue -- which does not see a lot of use. This is the user-name you were using? And you logged in with the same user-name both times, right?

Can you describe what the scrolling issue is? I fret this could be a tougher fix -- its not like I actually implement scrolling, this is just something Java provides.

Also, you'll generally find that the accuracy of classifications among the queues looks like "Cluebot-NG > STiki > WikiTrust" (assuming everything is working correctly). This is of course, is dependent on a complex set of criteria, how many people have been using a particular queue lately, etc.

Thanks, West.andrew.g (talk) 18:48, 21 December 2011 (UTC)

Andrew, yes, I have always been logged in as Peter Chastain. I agree that the queuing problem is more important than the scrolling one (in which, sometimes, when I use the wheel on my mouse, part of the text in the diff-browser screen is scrolled, while other text there stays put). Your diff-browser window is useful for a quick look at differences, but when I want to explore further, the Wiki-DIFF and Page-Hist links give me a nicer view with more functionality.

"Accuracy of classifications" could have several meanings. I am not especially bothered by false positives in the queues. If I understand correctly, Cluebot gets a lot of low-hanging fruit, the obvious vandalism. I am more concerned with accuracy. I recently stumbled across an article where someone changed "Fourteenth Amendment" to "Fourth Amendment," a major change to the meaning of the article, but hard for machines to identify as vandalism. Even harder to detect are the tiny changes, e.g., to a birth date, which, while not necessarily fatal, reflect badly on the integrity of Wikipedia.

What I am trying to say is that, to root out deliberately inserted inaccuracies, it is necessary to review a lot of perfectly good edits. In evaluating one change by an editor, I look at other things that the editor has changed. Here, I think the WikiTrust queue has great potential. I don't know if you are currently doing this, but it seems to me that an editor's history in STiki evaluations should be weighted more heavily than the general survival history of that editor's changes. That is, the fact that an edit survives the scrutiny of an STicki reviewer (who is actively looking for vandalism and other inaccuracies), is a stronger endorsement of the editor's reliability than is the fact the editor has entered a lot of text that has not yet been looked at or changed.

At any rate, WP quality control is a fight waged on many fronts, and it is a good thing that STiki accommodates reviewers who find their niche in the various areas of that effort.

Thanks, Peter Chastain (talk) 11:46, 22 December 2011 (UTC)

I have run a database report of your STiki activity, and have no evidence that you scored the same RID more than once. In particular, I output the ~250 RIDs which you classified as "innocent" using the WikiTrust queue. Then, I piped them into the Offline Review Tool (ORT) and quickly cycled through them. Many of the revisions were extremely similar (i.e., inserted identical content), but indeed took place on different (but similar) articles. The insertion of Russian Characters and a few redirects were particularly common in this batch. I suspect this might be why you thought you saw the same edit more than once (I can email you the classification trace, if you like).

There have been multiple claims that people were re-seeing the same edits, and I'm not trying to discredit you, but we've always traced it back to exactly this type of case. These vandalism detection algorithms, are after all, signature based. If someone is doing something repeatedly (inserting same content, massive deletions (which is what redirects look like to some algorithms)), and the algorithm doesn't like it, all of these instances are likely to land close to each other in the queue. Unless some new evidence of this issue arises, I am inclined to mark this issue as "closed." Thanks, West.andrew.g (talk) 02:41, 29 December 2011 (UTC)

STiki screen garbage

Andrew, I have a lot of garbage on my STiki screen right now. If you are interested in pursuing this, I can email you a screen photo that I took with my smartphone, along with details of my computer configuration. I hate to throw things at you that are not related to STiki core function. I think this could be a non-problem if STiki had a refresh button. As it is, I will have to stop and restart STiki. Cheers, Peter Chastain (talk) 19:21, 22 December 2011 (UTC)

Send me the screen shot and I will take a look. You can find my email on my personal webpage, http://www.andrew-g-west.com. I will have more time to look at these issues in the next couple days, when my last minute pre-holiday work and travel concludes. Thanks, West.andrew.g (talk) 19:54, 22 December 2011 (UTC)

About the same time I was getting the screen garbage, I started getting low-memory messages from Chrome. The STiki problem had something to do with re-drawing (the garbage would appear, or change, when I brought STiki out from behind other windows. I will send you the screen shots, in case you still want to look at them, and will let you know if this persists. Thanks! Peter Chastain (talk) 08:02, 23 December 2011 (UTC)

This is a tricky issue that I am afraid might be outside my scope. First, I wonder if all this might be related to the "low memory message" you saw in Chrome. I'm not sure what your machine is capable of, but I was able to count ~18 tabs open in one Chrome instance alone. I would think that would probably stress Chrome's memory allocation and make in-roads towards total system memory? If memory is hurting, this could cause some havoc with your graphics card. Have you experienced these errors again? If so, what was the system load at that time?

If this can't be blamed on something specific to your system, there is little I can do programatically to address the problem. I have never heard of anything like this before. Also, I use established Java packages (ui.Swing) to code the GUI elements, so it would not be easy to open them up and adjust anything. Moreover, it seemed like the "garbage" was not confined to STiki components, but instead to quite randomly-defined boundaries across the screen. Do other applications show/produce "garbage", or only STiki?

One thing you can do to help with this, and all bug reports, is to start STiki from the command line, (i.e., Windows calls this "command prompt"). If something goes wrong, their will be terminal output, instead of silently failing. These messages are extremely helpful to me in tracking down these errors. Thanks, West.andrew.g (talk) 03:04, 28 December 2011 (UTC)

Reverting a deleted article

I got Chloe Shimmin in the diff-browser window and went off to look at it and do some off-STiki reversion. By the time I got back to STiki and clicked the Vandalism (Undo) button, the article had been speedily deleted. When I click any of the three classification buttons (Vandalism (Undo), Pass, or Innocent) nothing happens. Peter Chastain (talk) 08:02, 23 December 2011 (UTC)

I am able to use menu functions, such as changing the queue. After I changed the queue from Cluebot-NG to WikiTrust and then clicked Pass, I got the "you've been using Pass too much" message, but it still did not advance to the next item in the queue. I ended up having to restart STiki. Cheers, Peter Chastain (talk) 08:48, 23 December 2011 (UTC)

Deleted articles/revisions can introduce some nasty corner-cases in code. Follow my above advice about running STiki from the command-line, and then copy-paste the terminal output the next time you encounter this error (be patient for a couple minutes if it does not produce anything immediately). Revert actions are forked to produce a separate thread that shouldn't affect GUI advancement, so I am a little curious to see where this error lands in code. Thanks, West.andrew.g (talk) 03:28, 28 December 2011 (UTC)

Cluebot-NG revisions in the Cluebot-NG queue

Very occasionally, when I am using the Cluebot-NG queue, I am presented with an edit that Cluebot-NG made. For an example, see revision 467440865. Is this normal behavior? Peter Chastain (talk) 15:06, 25 December 2011 (UTC)

CBNG actually scores its own edits really poorly sometimes (reason unknown), though I would assume there is some logic to prevent it from reverting itself (maybe some broader rule, like 1RR). STiki just consumes the raw scoring feed on IRC, so sometimes CBNG-queue users will see an edit that was actually made by CBNG. This isn't indicative on an error, but it seems reasonable that I write some logic to short-circuit this behavior. Thanks, West.andrew.g (talk) 03:21, 28 December 2011 (UTC)

Issue resolved (I think). The IRC server server on which the CBNG feeds operate is currently down (I've notified them on the talk page, maybe others should pile on), so not possible to test. Their may still be some old instances residing in the queue, but no new ones will be added in the future. Note, because CBNG feeds are down, I've changed the default queue to "STiki" until CBNG-IRC comes back online. Thanks, West.andrew.g (talk) 03:15, 29 December 2011 (UTC)

RB'd 0 edits, beaten to revert?

Andrew, I got an edit to Iron Man match in my queue. It was [ http://en.wikipedia.org/w/index.php?title=Iron_Man_match&diff=467921054&oldid=467920972 obvious vandalism], so I clicked the Revert button. STiki gave me the message, "LAST REVERT 71.251.132.63 (contribs) (talk) RB'd 0 edits beaten to revert? check page hist". But the page history shows that the edit has not been reverted. This happened a few minutes ago, some time around 2100-2110 UTC, but it is not an uncommon occurrence. Cheers, Peter Chastain (talk) 21:23, 27 December 2011 (UTC)

It happened again, with Macherla. Peter Chastain (talk) 22:36, 27 December 2011 (UTC)

Hate to sound like a broken record here, but follow above advice regarding run STiki from the command-line. For this particular report, I will note that the "RB'd 0 edits..." message is kind of a fallback case. If the rollback does not succeed for any reason, this is what will be output -- only because it is the most frequent *actual* reason. That being said, I don't know all the possible "rollback did not succeed" messages -- and don't know that they are enumerated anywhere by the Mediawiki folks. Edit conflicts and MediaWiki server failures are among the other possible reasons, I believe. In the immediate, I can work on some more accurate messages for those few failure cases/messages for which I am aware -- and maybe some logging for ones that have not been seen before (done with talk page responses for now; off to look at some code). Thanks, West.andrew.g (talk) 03:37, 28 December 2011 (UTC)

I started STiki from a Windows Vista command window and am getting RB'd 0 edits messages (where there were recent actual edits to the article), but no messages in my command window, unfortunately. (The command window shows a command prompt, followed by my command, STiki_2011_12_02.jar, followed immediately by another command prompt. Should I be seeing that second prompt before STiki terminates?)

This is such a frequent occurrence that I have to be a bit skeptical that the most common reason for the message is that someone else reverted or edited the article, but that doesn't really matter. I don't much care why my reversion was unsuccessful. My real concern is knowing whether anybody's reversion (or subsequent edit) was actually done. I suppose I should go to the article's history whenever I see the "RB'd 0 edits" message, but I often forget to do so.

Here is what I would suggest: After an attempted reversion, STiki should check to see whether the article has been updated (by anyone) after the edit in question. If not, there is a problem. (I thought I fixed something, but it wasn't fixed.) In that case, there could be an error message (perhaps bold red typeface, but definitely more prominent than the inconspicuous "RB'd 0 edits" message), saying "WARNING: Article was not changed". Thanks, Peter Chastain (talk) 23:46, 28 December 2011 (UTC)

Hi Peter. I was perhaps a little over-zealous in my "run from the command prompt" advice for this particular problem (though it still applies to the other ones). At current, there is no terminal output for this case (nothing is "broken" -- the error case is being caught and that is displaying the "RB'ed 0 edits message"). I will code up a new version of STiki that *does* print some debugging info to the terminal. It's easy to make the message more prominent. I want to figure out what condition is causing this error to be displayed. Thanks, West.andrew.g (talk) 00:12, 29 December 2011 (UTC)

The message should also state the title of the article that was not changed. Currently, even if I check for a "RB'd 0 edits" message, I cannot go check on the article, because I didn't write down what article I was trying to revert. Peter Chastain (talk) 00:11, 29 December 2011 (UTC)

Why have you not acquired the rollback permission? I highly suggest you do this. I assume you have the "use rollback" box checked when use STiki. If you don't have the native rollback right (which you don't), STiki uses a lot more logic and bandwidth trying to simulate the behavior for you. Both methods should be sound, but the "native" version is certainly less error-prone and better tested. Regardless, I am still looking into what might be going wrong. Thanks, West.andrew.g (talk) 03:30, 29 December 2011 (UTC)

Now you should be able to download the new version of WP:STiki which was uploaded tonight. If you run this version from the terminal/command-prompt, and you encounter a case where there is an editing/revert error, there should be detailed terminal output which you can submit to me. I have also taken the time to include some code to better distinguish the "error" and "beaten to revert" cases. The corresponding messages/branches are now completely exclusive of each other for clarity; but both have been made bold/red in the GUI in order to give some prominence to the situation. Let me know your experience/output from this. Thanks, West.andrew.g (talk) 09:09, 29 December 2011 (UTC)

Awesome changes! I love the links in the Last Revert panel. I have been starting STiki from the command line. When I attempted to revert a change to Teaching grandmother to suck eggs, I got the "RB'd 0 edits" message, and the following in the command window: A rollback failed to commit. The server returned code "badtoken" and details "Invalid token". Thanks, and please let me know if there is any way I can help. Peter Chastain (talk) 11:16, 30 December 2011 (UTC)

Yes, you can help by continuing to post these error codes from the terminal window. If it frequently lands on the same case, i.e., badtoken, I'll try to implement a fix so to remedy these conditions. Either way, follow my advice to get the rollback permission and you'll have these problems less frequently. Thanks, West.andrew.g (talk) 15:33, 30 December 2011 (UTC)

Can you add, to the console message, the name of the article that failed to revert? That would make it easier for me to go back and fix those. (Of course, if the bad-token problem gets fixed, maybe all this will be moot.) Peter Chastain (talk) 11:56, 30 December 2011 (UTC)

Yes, this can be added, but it is too low priority a change to push out a whole new version at this time. It will be included in the next one, and as you say, hopefully we can fix the root of the problem instead. Thanks, West.andrew.g (talk) 15:33, 30 December 2011 (UTC)

(Moving indention level back to the left, but this is in response to your various comments...)
I have rollbacker and reviewer rights, as you do, and I'm pretty sure I have had the "Use Rollback Action" box checked for the entire time that I have been using STiki.

I see that now. Previously, I stumbled across a page that had a list of rollback users with alphabetical "jump to" links (not the page that has the search function -- which showed you did have them). Apparently this former list was out of date or something. West.andrew.g (talk) 20:26, 30 December 2011 (UTC)

In my most recent STiki session, several hours ago, I got five instances of the "badtoken" message and one (legitimate) "allreadyrolled". I will continue to keep you advised. Cheers, Peter Chastain (talk) 20:12, 30 December 2011 (UTC)

Hmm, I wonder what makes these tokens go "bad". The same code is used to retrieve the edit tokens that do succeed the other 98% of the time. Very odd. West.andrew.g (talk) 20:26, 30 December 2011 (UTC)

Curious,:how long are these edits sitting in the STiki window while you do your off-STiki work? I am not sure of how the WP folks treat/handle tokens, but maybe there is some possibility the token has expired while you did this other work? Thanks, West.andrew.g (talk) 20:46, 30 December 2011 (UTC)

RB'd 1 edit

I went outside of STiki to do an assume-good-faith rollback of Nokia, electing not to leave a warning message. (I was being generous; it could well have been intentional.) Then, in STiki, I pressed the Vandalism button, assuming that STiki would do nothing but add the offending editor to its list of people to watch more closely.) STiki issued a "RB'd 1 edit" message but (of course) did not actually revert anything, because I had made a subsequent change. STiki also sent the editor a warning message. No harm done - it was a level 1 - but I assume that STiki would want to send a warning only if it does the revert. Peter Chastain (talk) 13:13, 30 December 2011 (UTC)

Strange. I am unable to simply reproduce the error. I logged onto my no-rollback account (User:West.andrew.g.norb) and went to work on STiki. When I encountered an edit that was vandalism, I went off-STiki to do the "undo" and then came back to press "vandalism" in STiki. Each time it was correctly reported that I was "beaten to the revert" and no warning was issued. If you try this simple experiment, what happens? If that succeeds, what was so unusual about Nokia that it was not handled correctly? Thanks, West.andrew.g (talk) 15:48, 30 December 2011 (UTC)

What you did, and what I recall doing, sound the same. As I noted above, the rollback right, or lack thereof, doesn't seem to be a factor. I will watch closely to see if it happens again, though I probably will not be on WP very much until 2012. Thanks, and have a happy New Year. Peter Chastain (talk) 20:12, 30 December 2011 (UTC)

Your help desk question

I found this page where you might be able to get some responses.Vchimpanzee · talk · contributions · 19:14, 11 January 2012 (UTC)

Your recent edits

Ive seen your edits repeatedly come up on my stalk lists (irc bot), even though it was a false positive it did lead me to further investigate. I see you have made 24392 edits reporting dead links in your userspace. I think that that is a waste of server resources and should instead be logged to a text file on your computer. ΔT _{^{The only constant}} 01:47, 9 August 2011 (UTC)

I've had several reports of users utilizing the "what links here" functionality positively with respect to this listing and its archives. If my "dead links" page shows up in the listing for an article, it indicates there is a link in need of monitoring -- and requires much less manual labor than manual link inspection. That being said, if some type of consensus could be secured that this is inappropriate, I would be happy to take it down. If anything, the scope at which this operates speaks to the massive degree of link rot on en.wiki, and the fact that people should be patrolling this page to eliminate those issues. Thanks, West.andrew.g (talk) 02:00, 9 August 2011 (UTC)

RB'd 0 edits

I continue to get this message a lot. To answer the question you posed, above, the time I spend off-STiki varies, but I do tend to research things before reverting. You mentioned the possibility that the token has expired on WP while I was doing other work. Would it be possible for STiki to re-send (or re-request; I don't understand the mechanism here) the token at the point when I actually click the revert button? Or would it make sense to retry a rollback that fails with the "badtoken" and "Invalid token" errors? Peter Chastain (talk) 10:59, 15 January 2012 (UTC)

Hi Peter. I continue to be confused by this, because I have never encountered a 'badtoken' message in my own editing. Before we try to remedy the symptoms (i.e., re-requesting or re-sending the rollback token), I'd like to try to figure out the root of the problem. To this end, I will: (1) write some more debugging code and push it to you in a new STiki build in the next couple of days, and (2) write some WM/WP developers to see if they can help me understand what makes a token go bad. At this point, I don't think your editing speed is of primary concern. Also, I'm hesitant to simply "fetch a new token; try again" because I want to believe there is a reason the token went bad, and a fresh token might not reflect the diff visualized in the STiki browser. Thanks, West.andrew.g (talk) 16:33, 16 January 2012 (UTC)

Download the new release. When a 'badtoken' is now encountered, there should be some additional terminal output, now. Send it to me via e-mail. Eventually we'll get this solved. Thanks, West.andrew.g (talk) 03:38, 17 January 2012 (UTC)

schoolblock request

Hello, In response to persistent IP vandalism I noticed that you gave that IP editor a "last warning" . Please be advised it did not work, and if you have admin powers I'd be grateful for a schoolblock. Following your last warning and yet another round of vandalism from this IP, another editor gave yet another warning. FYI, I have asked that editor why they did not impose the rootin' tootin' schoolblock. NewsAndEventsGuy (talk) 17:13, 25 January 2012 (UTC) This comment was posted by mistake. You can follow the this link to VSmith's talkpage if you are interested in details. Sorry to bother you. NewsAndEventsGuy (talk) 17:22, 25 January 2012 (UTC)

Bookspam

I just watched your anti-vandalism presentation (great!, btw), and particularly that part on link spamming and the difficulties to detect it gave me an idea. While I was watching, I realized it is not only "lucrative" to have a link to your spam website on the bottom of a popular article, but also a reference to a book of yours. Since Wikipedia is an encyclopedia, a lot of people use it as a starting point for research on a particular topic. So let's say you just wrote a book on the American Civil War. Wouldn't it be a nice "promotion" to have it refered to in every "Further reading" section of Civil War related articles? Certainly not everybody who sees a title listed in the FR section is going to buy it, but even if only a tiny fraction of readers does, there's a lot of money involved, which obviously is the incentive for spamming. (While I was writing this, I found out Wikipedia already has a name for this: it's "bookspam".) I wondered if you ever considered finding methods to detect bookspammers just the way you want to detect link spammers. --bender235 (talk) 14:17, 30 January 2012 (UTC)

Hi Bender, and thanks for the kind words about my presentation. Bookspam is an interesting problem and at first glance one that is probably much harder than typical online link spam. Consider that, (1) most things which enter publication are legitimate writings, this isn't exactly the male-pharmaceutical/pirated-software game that is usually played with online spam (though we could imagine a publisher conducting such behavior en masse). What I am saying is that in most cases, the publications are always legitimate, so it would be a matter of determining if they are appropriate in that context. (2) Most of these works probably aren't available online, making it difficult to do content/context analysis, (3) I am not sure how consistent the citation format is across Wikipedia (I am sure there is one, I don't know to what extent it is followed). This could make it difficult to aggregate all citations to a particular source. (4) While this certainly happens, I would be curious to its prevalence and real impact on the project.

I must admit, this is not at the forefront of my research at this time, but it is an interesting problem to think about. I think great progress could be made with only a few simple features, i.e., "how often does this user cite", "what percentage of those are to source [x], publisher [x], etc." Much of the heavy machinery I've used in anti-vandalism and anti-spam seem overkill or impossible here. Thanks, West.andrew.g (talk) 18:53, 30 January 2012 (UTC)

Warning???

Hello, West.andrew.g/Archive 4. Recently when I went to log-in to Wikipedia; I saw that you put a nice, big vandalism sticker on a IP Page which was my own. I had never edited with an IP. Probably you gave another guy a warning but it came on to me; as IP's periodically change their digits. By the way can I recommend a suggestion? I use STiki and found it very fast, helpful, and easy to use. Can you kindly change the background picture into a Wikipedia based picture if possible? Dipankan ^{In the woods?} 05:18, 4 February 2012 (UTC)

IP addresses that use DHCP are a fact of Internet life -- its why IP blocks are generally temporarily. While it may be interesting to think about who in your vicinity/IP-block might have committed the vandalism, I don't have much else in the way of comment here.

I am always interested in suggestions for STiki, but it is unclear what you mean by "background"? What is the current "background picture"? Thanks, West.andrew.g (talk) 05:55, 4 February 2012 (UTC)

Sorry, but there's no background picture currently. It is a Light-brownish colour theme in the background (You can see at the log-in corner). I'm trying to tell, if possible, can you change the colour to a picture of Wikipedia-related?

(a) I don't think this is easy to do in Java swing packages. (b) Why would anyone want to? The gray-ish theme is pretty standard among applications, no? Wouldn't a picture just makes things incredibly busy looking for the user? West.andrew.g (talk) 17:55, 13 February 2012 (UTC)

link updated

Hi West.andrew.g, I was updating a change in domain of "http://www.parlamento.gub.uy/leyes" to "http://www0.parlamento.gub.uy/leyes" with my AWB when I noticed your archive page /Dead links/Archive 73 (Jul 26, 2011 10:42:39 PM GMT). It looks tool generated, so I am not sure what to do and left it as it is. The dead links, however, have been fixed 1. Regards, Hoverfish Talk 17:00, 4 February 2012 (UTC)

Nope, you've done all that needs to be done! This is how the dead-link reporting service was intended to be used -- people notice the article on the "What Links Here" page and conducts an appropriate investigation/removal into the (possibly) dead link. Thanks, West.andrew.g (talk) 17:54, 13 February 2012 (UTC)

Sidenote; though I am not really sure your adjustment changed the link to one that works properly? West.andrew.g (talk) 17:57, 13 February 2012 (UTC)

Feedback about STiki

I kindly request you to have a look at the suggestion I posted here:Wikipedia_talk:STiki#Flexibility --Anbu121 (talk me) 15:03, 5 February 2012 (UTC)

Invitation to events: bot, template, and Gadget makers wanted

I thought you might want to know about some upcoming events where you can learn more about MediaWiki customization and development, extending functionality with JavaScript, the future of ResourceLoader and Gadgets, the new Lua templating system, how to best use the web API for bots, and various upcoming features and changes. We'd love to have power users, bot maintainers and writers, and template makers at these events so we can all learn from each other and chat about what needs doing.

Check out the Chennai event in March, the Berlin hackathon in June, the developers' days preceding Wikimania in July in Washington, DC, or any other of our events.

Best wishes! - Sumana Harihareswara, Wikimedia Foundation's Volunteer Development Coordinator. Please reply on my talk page, here or at mediawiki.org. Sumanah (talk) 16:12, 16 February 2012 (UTC)

STiki is Erratic

STiki has reverted "theses," which is a mispelling of "these," in this Macroscope article. And I see nothing of vandalism in any of the edits that were reverted. I think this is a STiki bot issue that needs to be addressed. — Preceding unsigned comment added by 76.8.8.27 (talk) 02:26, 17 February 2012 (UTC)

STiki is *not* a bot, it is intelligent routing tool operated by humans. In this case, it was User:Arunsingh16 who decided to revert your edit. You should take up this issue with him (and I agree that this edit is not indicative of vandalism). If you notice that a STiki user has mis-used the tool multiple times (isolated mistakes do happen), please notify me. A user can have their privilege to use the tool revoked when it is not used responsibly. Thanks, West.andrew.g (talk) 02:35, 17 February 2012 (UTC)

User:Arunsingh16 has received so many complaints about his vandalism-detection methods that he has stated on his talk page: "I am sick & tired of inexperienced people who themselves have been involved in editing that has attracted discussions, telling me how my recent revert / edit was wrong." Reasoning with him has had no effect, as his talk page shows. — Preceding unsigned comment added by 63.248.139.209 (talk) 15:26, 17 February 2012 (UTC)

I'd consider a STiki block, but it seems he has already moved onto using Huggle instead. A quick scan of his edits using that tool might be warranted (a found a few questionable ones, but will not call them out here -- he seems to be a bit "unsourced revert" happy). Moreover, there was evidence he was operating STiki at a very high rate of speed -- not a good quality in a Huggle user, especially since the revert environment has a bit of "competition." Alternatively, maybe the additional warning/tagging functionality of Huggle will serve him well. Regardless, the talk-page message is inappropriate. If you feel so inclined, talk to him about it, and if that fails, perhaps another trip to an administrative noticeboard? Unless he resumes STiki use, however, this issue is now outside my scope. Thanks, West.andrew.g (talk) 15:45, 17 February 2012 (UTC)

You can see from his talk page that reasoning with him fails. I don't have a link to the administrative noticeboard. He can simply remove any of my inappropriate comments. Thanks for the information, nevertheless. I am delighted to know that my spelling-corrections will now be reverted to the old misspellings using Huggle. — Preceding unsigned comment added by 63.248.139.209 (talk) 15:57, 17 February 2012 (UTC)

See Wikipedia:Noticeboards. There are multiple places to lodge a complaint if you believe his behavior is inappropriate. It's not my area of expertise, but I would say the talk-page message might be in the "Wikiquette" domain. Thanks, West.andrew.g (talk) 16:03, 17 February 2012 (UTC)

Suggestion on Dead links output

Hi there! Seems like a nice tool that. However I couldn't find description of it anywhere? Perhaps the output would be more readable if it was slightly rearranged. Currently the output looks to me almost machine readable! First, most people are interested only in certain subjects on Wikipedia, so in my opinion it would be best to start with the article in question. The exact time parameter seems almost redundant as the entries keep coming so often. At least, maybe it could be hidden somehow or be output last in a small font? Perhaps also add headlines so people could just scan the table of contents for their favorite articles? Of course, that could be another tool/interface as well since Dead links seem to have quite a pedigree! Palosirkka (talk) 16:52, 5 March 2012 (UTC)

Hi Palosirkka. The main "gateway" to the dead links is via Wikipedia:STiki/Dead_links, which does provide a description of what is going on. Unfortunately, most people seem to arrive at the dead-links pages in archives, which does not carry the header message. Given the large number of archives, I'd probably need to write a bot to pre-pend that message to all the existing archive pages

I had never intended that page to be super human-friendly. Instead, by using the "What links here" functionality of the toolbox -- a user who sees a "dead links" pointer to an article they care about will investigate that link. This has worked (at least) a couple of times in the past -- and tends to be how most people discover this listing.

Adding a table of contents doesn't make a whole lot of sense. To prevent editing conflicts (given the frequent updates), every single line in the report is itself a section. Thus, a table-of-contents would double the length of every report page!

I agree that someone could use this raw data to generate new interfaces into it, and this could certainly be helpful for the community. To be honest, though, this is not at the top of my priorities at this point. Thank you for your comments, West.andrew.g (talk) 22:31, 5 March 2012 (UTC)

I found about the Dead links patrolling for recent changes ending up to the page without the intro. Thanks the info and the tools. Palosirkka (talk) 13:15, 6 March 2012 (UTC)

STiki not starting up on OS X

Hi. I'm running OS X 10.5.8, and get the following problem when trying to start up STiki:

java -jar *.jar
Exception in thread "main" java.lang.NoClassDefFoundError: java/awt/Desktop
at gui_panels.gui_button_panel.<init>(gui_button_panel.java:91)
at executables.stiki_frontend_driver.<init>(stiki_frontend_driver.java:201)
at executables.stiki_frontend_driver.main(stiki_frontend_driver.java:157)

Unsurprisingly, clicking on it directly also doesn't work. Allens (talk | contribs) 04:07, 12 March 2012 (UTC)

Hmmm... The java.awt.Desktop package is responsible for interfacing between Java and your OS applications. For example, if one clicks on a hyperlink in Java, it will open in the default browser of the user via this class. A little forum digging revealed you most likely need to update your Java as the one that ships with 10.5.* is a bit out of date. Seemingly, visiting the Apple Downloads site, searching for "Java", and installing "Java for Mac OS X 10.5 Update 4" will fix your woes. Let me know how it goes. Thanks, West.andrew.g (talk) 13:43, 12 March 2012 (UTC)

Thanks! Not being a Java programmer (I prefer Perl), I had little idea what was inside your program and what was in the local library/libraries/whatever. I'll give that a try. Allens (talk | contribs) 13:58, 12 March 2012 (UTC)

No luck; same error. I also tried installing Update 10, and that didn't work either. Update 5 said it had already been installed. Admittedly, I haven't done a reboot, but the updates didn't say one was needed. Are there some debugging options that I could try running it with that would help get more info? Allens (talk | contribs) 14:31, 12 March 2012 (UTC)

Oh. In terms of the java version, java -version yields:

java version "1.5.0_30"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_30-b03-389-9M3425)
Java HotSpot(TM) Client VM (build 1.5.0_30-161, mixed mode, sharing)

Allens (talk | contribs) 15:11, 12 March 2012 (UTC)

I am continuing my digging. It seems things would be greatly simplified if you could get Java version 1.5 to version 1.6, but I assume this didn't happen with the Apple update because your machine is not "Intel-based or 64-bit capable"? What is frustrating is that this error is being thrown at start-up, rather than when you actually try to click on hyper-links in the program (a problem I have seen in the past). Give me a little while to look at this, I might be able to compile a version that does not have this class dependency. Thanks, West.andrew.g (talk) 15:40, 12 March 2012 (UTC)

That's odd, it is an Intel mac - it is not 64-bit, however. Allens (talk | contribs) 16:29, 12 March 2012 (UTC)

This page seems to define a solution to the issue (albeit a bit hacky). Give me a little while to integrate this into my code-base, test it, and push a new version out. I'll ping you on your talk-page when I get it done. Thanks, West.andrew.g (talk) 15:53, 12 March 2012 (UTC)

Take whatever time you need (I know what it's like being a graduate student!); I certainly have enough other things to work on... Allens (talk | contribs) 16:29, 12 March 2012 (UTC)

Done. New version uploaded at WP:STiki. Let me know how it goes, since I can't really test the fall-back functionality of the library. Thanks, West.andrew.g (talk) 17:05, 12 March 2012 (UTC)

Afraid not working quite yet (exception inside the executable this time):

Exception in thread "main" java.lang.NoSuchMethodError: executables.stiki_frontend_driver.setIconImages(Ljava/util/List;)V

at executables.stiki_frontend_driver.<init>(stiki_frontend_driver.java:231) at executables.stiki_frontend_driver.main(stiki_frontend_driver.java:157) Allens (talk | contribs) 17:32, 12 March 2012 (UTC)

Give it another try, I changed the file behind (so no Wikipedia edit was made), and I am pretty confident this new version should fix your issues. Let me know. Thanks, West.andrew.g (talk) 19:49, 12 March 2012 (UTC)

Works! Thanks very much... Allens (talk | contribs) 20:10, 12 March 2012 (UTC)

Some bubble tea for you!

awesome, thats what i was looking for, thanks ÐℬigXЯaɣ 03:19, 18 March 2012 (UTC)

also do have a look here http://en.wikipedia.org/wiki/User:DBigXray/STiki_UserBox_3 :) -- ÐℬigXЯaɣ 03:19, 18 March 2012 (UTC)