This user uses STiki to fight vandalism.

User talk:West.andrew.g/Archive 5

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Archive 4 Archive 5 Archive 6


Spam link detection - SURBL or similar?

Hi. Have you examined using URL/web-host blacklists like SURBL (http://www.surbl.org) or the domains it gets info from as part of the spam link detector? Allens (talk | contribs) 03:49, 12 March 2012 (UTC)

Yes, as reported on in a CEAS 2012 paper we ran a corpus of Wikipedia spam links over a similar blacklist service provided by Spamhaus. The intersection was minimal. It appears those "attacking" Wikipedia are not those who are conducting large-scale spamming elsewhere, a point further corroborated by other evidence (lack of large scale attacks, technically naive behaviors, non-commercial linking destinations, etc.). Thanks, West.andrew.g (talk) 13:10, 12 March 2012 (UTC)

A barnstar for you!

Thanks, Lotje! I have moved the barnstar to the "awards" section of the WP:STiki page. West.andrew.g (talk) 14:04, 14 March 2012 (UTC)

keyboard shortcuts for STIKI

Please reply if Stiki has keyboard shortcuts, i am unable to find them , if it does not have keyboard short cuts then it would be a good idea to include them, regards , thanks-- ÐℬigXЯaɣ 03:01, 18 March 2012 (UTC)

It does (per documentation)! After you make a first classification using your mouse, then the: "V" (vandalism), "P" (pass), and "I" (innocent) keys are all you need to classify edits at very high speed. Just use it responsibly. Thanks, West.andrew.g (talk) 03:13, 18 March 2012 (UTC)

Also can you please let me know if there is a way to see the difference in the number of characters in the diff, (like the one shown on huggle or the page history on wiki articles), regards -- ÐℬigXЯaɣ 04:14, 18 March 2012 (UTC)

This is not a value that STiki displays, a design decision I made due to my belief that the value is a bit misleading. If you take a featured article with 99,999 characters and replace all that with 100,000 nonsense characters, MediaWiki will describe the size change as "+1". It would say the same thing if you added a comma. I think this is a poor representation of the degree of "change" on a page so I decided not to include it. Thanks, West.andrew.g (talk) 14:19, 18 March 2012 (UTC)

On the "Stiki Diff Screen" how do you scroll the page down ? is there a shortcut key for this? as of now the pg down or the cursor down button does not do it and i have to do it by leaving the keyboard and using the mouse repeatedly. Please excuse me for asking this but i tried looking for this in the Stiki documentations as well but could not find it. thanks-- ÐℬigXЯaɣ 14:57, 7 April 2012 (UTC)

The need for keyboard functionality for diff browser scrolling is recognized. This need has been filed as T#001 in the new bug tracking table over at Wikipedia_talk:STiki, and there are plans to include this in a future release. Thanks, West.andrew.g (talk) 18:57, 11 April 2012 (UTC)

Improvements to STiki

hi Andrew, I must first thank you for the tool Stiki, its fast, wonderful and mostly shows you the cases that you need to see. but i think adding one more revert button for unsourced content or comment wont be too bad. (not all of them but just the commonly used ones) you can also have a look at my talk page here [1] Thanks and regards -- ÐℬigXЯaɣ 15:50, 26 March 2012 (UTC)

Hello, and my apologies for a very slow response. The "good faith revert" button is one that has been requested many times in STiki's history -- and given recent complaints about overuse of the "vandalism" button -- one that I think should be implemented. Despite my willingness to get to this done, real-life does intervene and it won't happen any time immediate. If you could secure some consensus(?) on what the button should say, and its edit comment, if it should issue a warning/greeting, etc. -- this would be a good first step in the right direction. Thank you also for your use my tool. West.andrew.g (talk) 00:44, 29 March 2012 (UTC)
Actually, the more specific "unsourced or comment" revert may be easier to come up with a system for than a more general "good faith but clueless" revert. Comments are, after all, unsourced, so the general "unsourced material" warning template used by Twinkle should be fine (perhaps along with the "welcomelaws" template). The default edit comment can be the usual one but with "unsourced/comment" instead of "test/vandalism" and not being a minor edit. Allens (talk | contribs) 10:30, 1 April 2012 (UTC)
  • agree with allen, we desperately need few more buttons, perhaps lack of some important options is the reason, preventing people from using STiki. by the way Andrew I have created a {{STiki topicon}} Face-smile.svg-- ÐℬigXЯaɣ 12:02, 2 April 2012 (UTC)
I don't think Allens is calling for multiple buttons; instead just one labeled "unsourced/comment" -- and one more button is all I really want to include on the panel. However, this makes dealing with the "comment panel" a bit tricky -- because now shouldn't it allow the editing of the two comments that STiki is using -- and won't it be confusing if say, the "test/vandalism" comment is being displayed but one is actively pressing the "unsourced/comment" button? Thanks, West.andrew.g (talk) 16:21, 2 April 2012 (UTC)
Perhaps it would be better to remove the textbox from the revert comm panel, and move it to a settings window. This way many different reasons can be side-by-side which are attached to various buttons in the classification panel. –meiskam (talkcontrib) 14:27, 6 April 2012 (UTC)

Chiming in here as an interested editor who has seen several editors misusing STiki. It seems to me, it is very necessary to have a button for good faith edits. The button could generate a edit summary along the lines of Twinkle's like "Reverting good faith edit(s) by Joeditor to last revision by...".

I would like to take this opportunity to express the need for this change with an example from a while ago. I hate to single anyone out but this user of STiki was getting hounded by so many other editors (myself being one of them) that it drove him to place this [2] at the top of his talk page. He was repeatedly being warned[3] about misusing STiki and making false accusations of vandalism (admittedly, looking back on it, I don't think I, and maybe some other editors, handled this the best way possible). It became clear that the editor had a strong desire to continue using STiki despite the fact he knew it was going to issue false vandalism warnings in his behalf. He did not want to accuse so many other users of something they did not do, but he didn't see it as an important or compelling enough reason to stop using the tool as he had grown accustomed.

Now you may say this editor was acting irresponsibly and I would agree, but the fact is, a lot of arguing, tense words, wasted threads, false vandalism accusations and hard feelings would have been avoided if the user simply had another option at hand for an edit summary. The user now edits with Huggle and STiki, I believe, and has not caused any trouble since that I am aware. This goes to show that even in the hands of an otherwise very good editor, the tool can very easily lead to misuse just because of a single technical limitation it has. In effect, it can "bring out the worst" in some of us. Just my, for what its two cents worth. I know you have already said you intend to add such an option. Thank you for your time.--RacerX11 Talk to meStalk me 23:11, 3 April 2012 (UTC)

I agree with the idea of having a settings window (which could also handle "Rev. Queue", "Revision Filters", and "Appearance" eventually). For the default edit summary, I agree that copying Twinkle's good faith reversion message would be a good start. Note that the automatic WP:AIV reporting will need to be changed - possibly to a popup suggesting dealing with the problem user manually, including buttons to open their contributions list and talk page. Allens (talk | contribs) 14:27, 7 April 2012 (UTC)
As of today's release (2012-04-11) STiki now includes a "good faith revert" button. No warnings are placed on the user's talk page and the edit-comment is modifiable in the expected location (now tabbed!). This matter is considered closed. If you want to discuss the warning/AIV/default-comment matters, please start a new section. Thanks, West.andrew.g (talk) 17:23, 11 April 2012 (UTC)

Toolserver account

IF you expect to need a toolserver account, I suggest applying sooner rather than later. In my experience it's not a quick process; you can expect to wait at least a month. Josh Parris 01:14, 2 April 2012 (UTC)

STiki

Is there a way that STiki can run as a Wikipedia script (on my monobook.js page) or a Firefox extension or plugin? What about a Greasemonkey script? Allen (talk) 21:19, 5 April 2012 (UTC)

What do you mean by this? Have you ever used STiki in a desktop environment before? STiki is not Javascript, nor is intended to be used while using a wiki in a typical web browser. The closest thing I can imagine to your suggestion is running STiki as a Java applet embedded in a webpage -- but I am not sure what that would gain. Or perhaps I am misinterpreting your suggestion? Thanks, West.andrew.g (talk) 01:13, 6 April 2012 (UTC)
Maybe I am misunderstanding what STiki actually is.
Am I mistaken, or can STiki be used on Wikipedia? No, I have never used it before.
It sounded like a cool tool for editing wiki pages. I just didn't want to download anything.
Can you please help me understand? Allen (talk)
I have started using STiki. It is pretty cool. How do I change the "Warn Offending Editor?" box to show the edit that caused the vandalism? At least one editor has asked me what was the edit. Allen (Morriswa) (talk) 20:05, 6 April 2012 (UTC)
Again, I am little unclear on what you are asking? I quickly searched your user-talk page and couldn't find any mention of such a request. The "Warn Offending Editor?" option is simply a checkbox, how could it possible show an edit? Isn't the diff window "showing the edit that caused the vandalism?". Thanks, West.andrew.g (talk) 19:05, 11 April 2012 (UTC)
Thank you for such a valuable tool. I don't know what I was thinking about, but I think it should be this: "How do I get STiki to give me an option to watch the talk page of the user whose edit(s) I reverted?" However, as I said, at least one user has asked what kind of vandalism was being committed. I honestly have to say "I don't know". I go through so many edits with STiki, that I don't know what edit was on what page or where committed by what user. Is there some way that this can be resolved? I have a feeling that some editors are going to get really angry and -- who knows what will happen? Allen (Morriswa) (talk) 00:05, 12 April 2012 (UTC)
Okay, your query is much clearer now. First to answer your question: If you want to watch the talk page of a reverted editor you should wait until the edit is reverted and appears in the "last revert panel". From there, you can use the "(talk)" link to visit the user's talk page in your browser and watchlist it in the standard fashion. STiki does not provide this functionality directly because I don't see it as a good idea. Do you really want a watchlist containing 100's or 1000's of IP talk pages? I don't. It is convention they should come to your talk page if there is any issue.
Continuing with your other comments: I don't understand how a user could ask about "what kind of vandalism" is being committed. Vandalism is WP:Vandalism. I realize you may do many edits with STiki, but if someone complains, it should be no issue to track down the instance. Go to the complainants talk-page, STiki should have left them a warning there. This warning will indicate the page affected; visit that page's history and looking for your edit with STiki. This should be enough to produce a diff of the actions in question. Finally, why would some editors be "really angry"? If it was truly vandalism, who cares. If they have a reason to be angry, then why are you irresponsibly using the STiki tool?
I guess I shouldn't be adding the IP addresss.
I feel a bit dumb about how to find the edits. That should have been easy to figure out. Allen (Morriswa) (talk) 00:42, 12 April 2012 (UTC)

Feedback on 2012-04-11 Release

Dear Andrew, Thanks for accepting our suggestions and recommendations in your latest release above, I have tried it and I must say that I am glad to use the new version. The "rollback derived differences" was much needed, I can now clearly see what changes my revert will make. I hope the scroll down feature will soon come up. Thanks, the 100,000 milestone is fast approaching, be ready with your Champagne bottle Face-smile.svg -- ÐℬigXЯaɣ 21:29, 11 April 2012 (UTC)

Thank you! Yes, your "scroll issue" should be resolved shortly. To be honest, it got kind of lost in the midst of the much larger changes. Assuming it doesn't pose any unexpected difficulty, I should roll out a fix in the next couple of days (as I also take care of all the minor issues popping up as a result of the new release). Thanks, West.andrew.g (talk) 22:40, 11 April 2012 (UTC)
This feature request was fulfilled in the 2012-04-12 release. See the release notes immediately above for more information about its operation. This matter is considered closed and ticket T#001 has been wiped from the active table over at Wikipedia_talk:STiki. Thanks, West.andrew.g (talk) 03:17, 12 April 2012 (UTC)

A barnstar for you! (x3)

Thank you all for the barnstars, I've moved them over to the "awards" section over at WP:STiki. Thanks! West.andrew.g (talk) 16:50, 13 April 2012 (UTC)

Bot activity

While performing some data analysis, I ran into your account as an oddity because some of your activity resembled bot-like activity, yet you do not have a bot flag. Upon review of the bot-like edits, I see a ton of structured edits from your account over the course of nearly a year. Are you operating a bot (external automated program that edits the wiki) through your user account? If so, please cease doing so immediately and contact WP:BAG for approval of a bot-flagged account to run your bot under. If not, how is it that you are making several edits/second for weeks at a time? --EpochFail(talk|work) 18:49, 13 April 2012 (UTC)

Hello EpochFail. Yes, my account is making structured edits. The vast majority of these are to Wikipedia:STiki/Dead_links which is a transclude from User:West.andrew.g/Dead_links (and the actual edits are made to the latter page, in my user-space). Per bot policy and as I brought before BAG (see my post), one can use bots to edit one's own user-space without the bot flag or approval. See User_talk:West.andrew.g/TalkArchive04#Suggestion_on_Dead_links_output for a little description on my intentions with these edits. Also, in the interest of full disclosure, the pages Wikipedia:STiki/leaderboard and Wikipedia:STiki/revert count‎ are also script-updated, but this only happens once daily and would seem to be an uncontroversial use case. Thanks, West.andrew.g (talk) 21:19, 13 April 2012 (UTC)
I see. My mistake. Sorry for the trouble. --EpochFail(talk|work) 21:55, 17 April 2012 (UTC)

Downloaded, but cannot find STiki... Help me, please!

Hello Andrew, Welsh Dragon edit by STiki Earlier yesterday evening I noticed a Welsh Dragon edit by STiki. I investigated your article page – and this page ... liked your idea ... downloaded, and ... nothing! Where/how will it show up on my computer? ... how do I access STiki? What should I do next? Your assistance would be much appreciated. King regards,Gareth Griffith-Jones (talk) 11:55, 14 April 2012 (UTC)

Hi Gareth, If you click the download link on wp:STiki then it will download a STiki_exec_2012_04_12.zip file into your download folder (or your download directory) you can see the download history of your browser (in tools or options) Goto that folder unzip the File STiki_exec_2012_04_11.zip and double click on the file STiki_2012_04_12.jar , this will start the Stiki application, rest is self explanatory, regards -- ÐℬigXЯaɣ 12:14, 14 April 2012 (UTC)
Hi Gareth. Everything DBigXray said above is accurate. You'll need to have Java installed on your machine for STiki to work, but this is a pretty normal package that you are likely to already have. Let us know if you have any problems and happy editing! Thanks, West.andrew.g (talk) 15:45, 14 April 2012 (UTC)

Better way to promote Stiki

See stiki talk page.--Deathlaser (talk) 17:03, 16 April 2012 (UTC)

Meta feed thoughts

I was thinking about the idea of a "meta" queue combining data from, say, the Cluebot-NG and STiki (metadata) queues. The most obvious form of such is to take the Cluebot-NG feed and check to see if its output also falls afoul of the STiki (metadata) detection system (using current thresholds or, if need be for getting some output if the two mechanisms should not intersect very much in their output, more relaxed ones). This idea is obvious enough that I'm sure you've thought of it. As a modification of it, I suggest having a fusion as follows:

  • 50% of the time, use the above double-filtered system.
  • 25% of the time, use the Cluebot-NG feed.
  • 25% of the time, use the STiki (metadata) queue.

The advantages of this are as follows (there may be others):

  • It keeps gathering data usable for improving the STiki (metadata) algorithm (including coming up with a better version for filtering the Cluebot-NG feed, if adjustments to its parameters for this purpose are needed).
  • The output will be false positives sufficiently often that people won't be just madly clicking "vandalism".
  • If the double-filtered system isn't outputting sufficiently many cases (is too stringent), it can just fall back to more frequently doing a mixture of the Cluebot-NG and STiki (metadata) queues.

Your thoughts? Allens (talk | contribs) 10:52, 1 April 2012 (UTC)

As I just mentioned over at STiki talk, the plan is for a "true" meta-queue that integrates features from all systems into a single ML model. This could be done at the feature level (again, see our CICLing 2011 paper), or by applying something similar over just the 3 output scores. We need an agreed corpus to get this done (and the CBNG folks don't like the corpus that the academics use, thus they have taken to creating their own). That seems to be a major hang-up. Thanks, West.andrew.g (talk) 01:19, 18 April 2012 (UTC)
It might be possible for at least the STiki metadata features to be used with the corpus used to create CBNG. That would enable creating at least two meta queues: one combining CBNG and STiki/metadata, and the other combining the non-CBNG queues. Those two could then be combined by the above means, at least temporarily. (By "falls afoul" in the above, I'm meaning "is more likely to be vandalism/spam/whatever than the average entry in the database", incidentally. The resulting queue could be ordered by the maximum spam probability between the two meta-queues.) I have encouraged the CBNG people to get together with you regarding databases, incidentally - I've also done some assisting with classifying their database. Allens (talk | contribs) 17:18, 19 April 2012 (UTC)

STiki Collaborator Request

You mentioned in your post on my talk page that you are looking for collaborators, I would be happy to help! -Mcfar54

I, too, would like to help, but I know next-to-nothing about computer programming. Allen (Morriswa) (talk) 01:53, 12 April 2012 (UTC)
Hello to you both (and anyone else who appears here). As I mentioned in my previous post, I'm looking for someone who can help me manage the non-technical aspects of the STiki project. This would include:
  • Issuing barn-stars to STiki users when they reach milestone edits (aiding user retention).
  • Being a talk-page watcher; combined with a general understanding of how STiki works
  • Drumming up support/advertisement in non-spammy ways (I mentioned WP:Banner)
  • Monitoring STiki users for signs of abuse (and trying to correct it)
  • Your own creative ideas about how STiki can improve its reach/experience/etc.
Ideally, I'd like to find 1 or 2 experienced individuals who are willing to get on-board for some period of time. I can provide database reports, advice, and anything else I have at my disposal -- but I'd like the person to be largely independent in their work. I don't have much to offer in return (maybe an offical "title"?) but it could be good experience for an editor. Interested parties, let me know! Thanks, West.andrew.g (talk) 03:42, 12 April 2012 (UTC)
I wouldn't mind offering my services for your consideration :) Orphan Wiki (talk) 10:27, 12 April 2012 (UTC)
No offense to the others here, but Orphan Wiki has been around a long-time and seen a lot of the tool's evolution (and has spent a lot of time across Wikipedia). I feel inclined to have him head up something like this; maybe mentoring and making use of the others who have volunteered? What do you think Orphan Wiki? Where do you think is a good place to start? Thanks, West.andrew.g (talk) 02:28, 18 April 2012 (UTC)
Thank you! Well as you said, my expertise with STiki is much more on a practical level, rather than a technical level, so I would happily watch over the STiki talkpage for queries being raised by users new to the tool (with the aid of the other experienced members, of course!) There have often been situations in which other users have complained via the same talkpage of mishandling in regard to usage of the tool. I have sometimes leant a hand in dealing with these cases before, and can continue to do so.
I also really love the barnstar idea. Although how would we go about with regard to their distribution?
Best regards, Orphan Wiki (talk) 08:45, 20 April 2012 (UTC)
Well, I think we should begin by discussing the thresholds at which a barnstar or talk-page message are deserved (first use, 100, 1000, 10000 classifications?). I'd like to hear your ideas on this. If we can set some type of fixed policy, I can write a simple database script that will determine users who cross a threshold each night. This report could then be dumped to a wiki page or your email inbox? Thanks, West.andrew.g (talk) 16:12, 20 April 2012 (UTC)

OK, well I feel that 100 is possibly a little low. I myself can amass 100 hundred edits on STiki in one sitting and in a short amount of time, if the vandalism rate is particularly high. Having not seen the stats for any STiki users, I propose something along the following lines:

  • First use = more of a welcome message, rather than an award?
  • 1,000 = first barnstar
  • 5,000 = second barnstar?
  • 10,000 = second/third barnstar?

These seem like much more satisfying levels with which to distribute an award, and it shows both trust and dedication when utilising the tool. The large difference between the edit-counts also makes an award seem more deserved and well-earned. 5,000 is a possibility, or we could have 10,000 as the next level up?

Finally, would we distribute standard Wikipedia barnstars found over at the barnstar page, or create stars that are more personal to STiki? Orphan Wiki (talk) 11:19, 21 April 2012 (UTC)

You might want to, as with the GOCE monthly drives, use the "regular" anti-vandalism barnstar (1.0 or 2.0 version) for the first one, with the higher ones being customized (once you can locate a graphic designer to customize them...). Allens (talk | contribs) 11:36, 21 April 2012 (UTC)
Yeah, that sounds like a decent idea. Orphan Wiki (talk) 11:43, 21 April 2012 (UTC)
Yes, all these sound reasonable. I'll write up a script to dump usernames who cross these thresholds each night. Thanks, West.andrew.g (talk) 13:25, 21 April 2012 (UTC)
Please check out WP:STiki/milestones, which I think is largely self-explanatory. Why don't you develop a standard "welcome message" and "congratulations you've reached XXX edits" and post them here? If everyone is pleased with them, then we can start distributing them to users each night per the new "milestones" page. Thoughts?
Additionally, that milestones page should be a good way to identify and spot-check the newest STiki users. Thanks, West.andrew.g (talk) 20:41, 22 April 2012 (UTC)
Can we make it so that rather than just saying "1" for a new user it says "1/234", or maybe "1/234? The second number is the total edits on the English Wikipedia. This should give a good idea about whether or not the person is a total newbie or just new to STiki. Yaris678 (talk) 10:45, 24 April 2012 (UTC)
Feature added. You will see this on all new reports (first one being in ~20 minutes). Thanks, West.andrew.g (talk) 04:40, 25 April 2012 (UTC)

Helping to promote Stiki

Hi Andrew apart from {{STiki topicon}} I have created 3 userboxes (see below). They automatically include users in the Category:Wikipedians who use STiki I have also tweaked the two existing Stiki userboxes [4] and [5] to include users of the 2 existing userbox into this category Please do have look, you can revert these edits if you dont agree with them, thanks and regards -- ÐℬigXЯaɣ 22:09, 14 April 2012 (UTC)

User Box Code
Goes to top right corner of the page (see) {{STiki topicon}}
STiki logo.png This user uses STiki to revert vandalism.
{{User:DBigXray/userbox/STiki 2}}
STiki logo.png This user beats vandals using the STiki anti-vandal tool.
{{User:DBigXray/userbox/STiki}}
STiki logo.png This user beats vandals with the STiki anti-vandal tool.
{{User:DBigXray/STiki UserBox 3}}
Your contributions have now been included on the main WP:STiki page. I hope you don't mind that I copied them into my userspace for consistency and ease of maintenance. You have been credited on the individual template pages. Thanks, West.andrew.g (talk) 02:13, 18 April 2012 (UTC)
P.S. feel free to add the top-icon on pages as appropriate. Thanks, West.andrew.g (talk) 02:15, 18 April 2012 (UTC)
(talk page stalker) I'm adding one of them to my page! --Rsrikanth05 (talk)

Created a new one... if people are interested:

User Box Code
STiki logo.png This user uses STiki to patrol not-necessarily-that-recent edits.
{{User:Yaris678/STiki UserBox}}

Yaris678 (talk) 23:34, 29 April 2012 (UTC)

And I added Yaris' contribution to the main page, with attribution on the template page. Thanks, West.andrew.g (talk) 15:31, 1 May 2012 (UTC)
Cool.
Maybe it would be a good idea to do other user boxes based on what makes STiki distinctive. Something like "This user uses sophisticated algorithms to find vandalism using STiki"
Yaris678 (talk) 15:41, 1 May 2012 (UTC)

Scale, STiki, QTM

Hi Andrew,

It is good to see STiki growing in popularity. I am slightly concerned that the more popular it gets, the more difficult it will be for you to keep track of users who might be misusing STiki (either deliberately or through ignorance).

I guess this is similar to a problem they had in the earlyish days of Wikipedia, when they said "Jimbo doesn’t scale". One approach would be to insist on users having the Rollback right (as per T#005) but this has the potential to keep out good users. I guess another approach would be to give some trusted users access to your tools for checking that users aren’t abusing the system… However... I think I have a cleverer idea than either of these. I am hoping you will like it because it is taking a quantitative approach to the trust of STiki users.

I am essentially suggesting a peer-checking system. The more users you have the more beneficial this becomes. This idea is inspired by the user-based moderation system of Slashdot, but it is slightly different. The simplest way it could work is that when a user says that an edit is innocent, it is sometimes put back in the queue and fed to a different user (in the same way as if they had pressed "pass"). It can then be recorded whether the second user gave the same answer as the first. A slightly more advance version would have two new hidden queues, one for edits "innocented" by inexperienced editors to be sent to experienced editors and one edits "innocented" by experienced editors to be sent to inexperienced editors... and assume that the experienced editor is always right... unless they end up disagreeing with a lot of other users in which case there may be something weird going on. You could use this to quickly identify users who press either innocent of revert when they shouldn’t. You could make it so that the newest STiki users get a lot of edits referred on. You could also make it so that if you are identified as pressing innocent too much it increases the proportion of edits that are referred on... hopefully this will allow the system to collect enough evidence to show that it was a temporary blip and the rate can go back down... but if not the rate will go up until you reach the point where all "innocenting" by the user is rechecked.

If a user is identified as reverting too much then they need to be stopped from using STki relatively quickly. I guess some kind of intervention would be required at that point, to look at all their past edits and see what needs undoing.

An alternative/additional approach involves checking edits from the recent-changes API to see if they are reverts of STiki reverts. If one is, STiki could automatically add the new edit at the top of its queue... and then look to see what a different user makes of the edit. This wouldn’t check for over-use of the "innocent" button but it would have the advantage that if the edit was correctly identified as vandalism the first time around then it is good that it back at the top of the queue!

Yaris678 (talk) 17:39, 16 April 2012 (UTC)

How does one determine what an 'experienced STiki user is'? I would prefer to define it in terms of Wikipedia experience, rather than STiki experience. Most of STiki's problematic users have actually reviewed a high number of edit-views, seemingly because they get edit-count-itis and move at lightning speed for several days at a time until enough people have yelled at them.
I have some tools for determining when an automated attack has been launched against the STiki backend. My technique for detecting simple mis-behaving users is much simpler. I visit the leader-board, see who is new to the scene, check out their revert %'s, look at their talk-page for controversy, and maybe spot-check some of their STiki reverts. In my experience, usually the talk-page is a dead give-away.
I would prefer requiring rollback rights before implementing such redundant verification: (1) It takes time to implement, (2) Some new users aren't really doing novel work, (3) It requires me or some admin to read and interpret the daily reports, and take action against who might be bad. Entrance conditions alleviate all of this work. Maybe if we are are lucky, we could get STiki mentioned on the template one gets after they receive rollback rights? Thanks for your suggestions, regardless. Thanks, West.andrew.g (talk) 01:36, 18 April 2012 (UTC)
I trust existing users without problems would be exempt from such a requirement? (Admittedly, for myself personally, I should probably ask for rollback rights anyway to reduce load on the servers; I would think a 0.12% false positive rate with ~5000 STiki reverts should be enough, although I don't really know for sure.) Allens (talk | contribs) 02:57, 18 April 2012 (UTC)
Yes, I would grandfather them in. Rollback is quite a minimal threshold for those that do not have it (and are to any degree a constructive editor). I can't say what the magic number would be over at Wikipedia:Requests_for_permissions, but I'm willing to guess it is an order of magnitude (or two!) less than your experience. Thanks, West.andrew.g (talk) 03:04, 18 April 2012 (UTC)
You're correct, I now have rollback... Allens (talk | contribs) 17:12, 19 April 2012 (UTC)

Hi Andrew, More thoughts on this:

  1. I am pretty sure that I didn’t get a message when I got given rollback rights. I notice that Allens did… but that message doesn’t seem to be based on any template I can identify. That said, I’m sure getting STiki mentioned on one of Template:Rollbackgiven, Template:Rollbackgiven2 or Template:Rollbackgiven3 will probably get it seen by some people and it sounds like a good idea.
  2. In relation to my peer-checking idea, the obvious way of defining an experienced STiki user is that is one with Rollback rights! Alternatively, you could say must have rollback rights + 2000 non-STiki edits + 20 STiki edits... or something like that.
  3. I think such a system could be mostly automated so that it just chugs along happily until it identifies someone misusing the system… at which point it stops them from using STiki.
  4. I agree that in terms of code development time, it would be a lot easier for you to simply require STiki users to have rollback rights. I thought you were keen not to exclude potential users. Given that rollback is fairly easy to get and there are now quite a lot of active STiki users, maybe now is the time to implement T#005.
  5. I agree that it would be nice to grandfather in old STiki users that haven’t gained the rollback right. However, that should probably only be given on a temporary basis. People who are using STiki correctly should be able to get Rollback without too much fuss in a few months. Anyone who remains without rollback rights after that was probably either problematic or just not doing much editing.

Yaris678 (talk) 12:26, 20 April 2012 (UTC)

I agree with most everything Yaris said above, but I'll comment on #4 in particular. I am not keen to exclude users; but I wonder how many users this transition would actually affect. I will write up a script to determine what percentage of STiki users have the rollback right (and how many had it when they first used STiki). Maybe that would be a good starting point? We also need to consider that when (presumably, newbie) users abuse STiki, it doesn't just reflect poorly on that user or the encyclopedia -- it is really bad press for STiki. Thanks, West.andrew.g (talk) 16:16, 20 April 2012 (UTC)

Stiki Directions for use

Hi Andrew, I have updated Wikipedia:STiki#Directions_for_using_STiki Please check, Although I had not consulted you before updating the page, I felt it was needed for new users who want to understand how to use it, I was asked about it by a user so I thought of putting it here as well, what is your opinion ? . feel free to revert me anytime. -- ÐℬigXЯaɣ 14:16, 17 April 2012 (UTC)

I thought the change was a fundamentally good one. I've since changed the formatting so it matches the other table on the page and made some language changes, but thanks for your contribution. Thanks, West.andrew.g (talk) 01:37, 18 April 2012 (UTC)

STiki algorithm: Vowels?

Have you considered (for English and similar languages only, of course) using as features:

  • Number of non-vowels together (possibly non-vowel letters together, as a separate feature);
  • Proportion of non-vowels (of letters and/or of all characters)?

Vandalism using random keys tends to have a low proportion of vowels. Allens (talk | contribs) 17:40, 17 April 2012 (UTC)

These are fine suggestions. At current, I believe the "longest repeated character sequence" is the closest feature in production. As I've said, I have lots of features that need to be brought live (per the CICLing paper). Matters seem to have advanced to a point where the inclusion of minor features such as these bring very minor benefits to AUC, precision-recall, and our other performance metrics. Thus, I haven't been to eager to incrementally extend the scope of the WikiTrust algorithm, I'd rather roll out everything in bulk if/when the "meta" queue becomes a reality. Thanks, West.andrew.g (talk) 01:46, 18 April 2012 (UTC)

How to use STiki for Commons?

Hey, is there any way to use STiki on Wikimedia Commons? If not, can you please make a version for that? Best, Dipankan says.. ("Be bold and edit!") 15:09, 18 April 2012 (UTC)

Have you used STiki before? How would you suggest taking the text-diff format of a site like en.wp and making it work for the media-driven Commons? How much vandalism is going on over there? Thanks, West.andrew.g (talk) 16:18, 18 April 2012 (UTC)
Yes, I have used STiki before. It's a great tool. I'm not a technical guy. Well there is pretty more than 250 vandalism everyday... that's very minor compared to enwiki. That's going to be useful for rollbackers out there. Dipankan (Have a chat?) 15:28, 19 April 2012 (UTC)
Honestly, I'd prefer Huggle or Igloo to STiki. Also, it would be handy to revery vandalism to text based sections like in categories and file descriptions. --Rsrikanth05 (talk) 15:45, 19 April 2012 (UTC)
I don't doubt there is vandalism on Commons (with everything pointing there), but is it text-based or media-based? While STiki's metadata features might be useful in detecting where image vandalism has taken place, it doesn't really have the capability to display the change to an end-user. What is the current anti-vandalism approach on Commons? Are there any tools? Thanks, West.andrew.g (talk) 19:35, 19 April 2012 (UTC)
As of now, CAV matters on the Commons are mostly sorted manually, maybe using the Rollback right. --Rsrikanth05 (talk) 06:14, 20 April 2012 (UTC)
CAV? Sorted how? A little more information? Thanks, West.andrew.g (talk) 16:18, 20 April 2012 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────Sorted as in, if vandalism pops up, then users revert it the old fashioned way. That is for text. Click on undo manually. Also, I take back my comment. I just love the STiki interface. : ) --Rsrikanth05 (talk) 18:51, 20 April 2012 (UTC)

Stiki translations

Heya, just to let you know, Huggle, another Wiki anti vandal tool has been translated into Hindi. Can the same translations be used for STiki? I don't mind doing a little additional work. --Rsrikanth05 (talk) 18:53, 20 April 2012 (UTC)

As recently archived discussions (with respect to Portuguese) will show, translation and localization are becoming relevant issues. When talking about STiki we need to be aware of a couple things. First, there is the GUI component -- this is straightforward translation work. Second, there is the back-end component. When text occurs here, it is not likely to be "informally translatable", instead these are very exacting regular expressions (again, not terribly hard for someone who knows what they are doing).
At the heart of the backend is the "vandalism model." This is where things get tough. You see, this model contains statements like "if the editor is unregistered, add 20% to the vandalism probability" (vastly over-simplified). This model was built over the properties of 1000s of English Wikipedia edits (which were labeled good or bad). This model cannot be translated -- although we can imagine manually writing rules (like the original ClueBot) which outperform Huggle's simple mechanisms.
Finally, once someone has translated everything, they need a place to run it. STiki needs a server that is always on, with a static IP address, IRC, MySQL, etc. I keep one machine quite occupied doing this work for English Wikipedia. I simply don't have the resources to analyze all Wikimedia wikis. The toolserver might, and I encourage someone to investigate getting rights over there and doing some digging.
Once someone demonstrates to me they have bullet point #3 in place, we can begin talking about making #1 and #2 happen. Thanks, West.andrew.g (talk) 19:35, 20 April 2012 (UTC)
Woah! Why is it that, except Huggle, all CV tools are run independently? Even Igloo is. --Rsrikanth05 (talk) 06:02, 21 April 2012 (UTC)
What do you mean by "run independently"? FWIW, Huggle is able to exist in many languages so easily because it has no server-side component, everything is done on the client-side (it hooks into recent changes IRC). Thanks, West.andrew.g (talk) 13:29, 21 April 2012 (UTC)
Sorry, I meant to say unable to run independently. Okay, I didn't know that STiki uses a different method. I like the overall look and feel. However, it takes abit to start up. --Rsrikanth05 (talk) 21:22, 21 April 2012 (UTC)
I still don't know what "able to run independently" means, but that's not terribly significant. STiki takes a while to start-up because it has to talk to its own servers (in Philadelphia) to determine what RIDs to get and then ask Wikipedia for the metadata and diff for those RIDs. When it starts, its doing about 10 of these requests in parallel so information can be cached locally, making subsequent revision advances very quick. You'll notice another delay when you try to advance to a new edit after logging in a new user; same exact reason, the local queue essentially needs to reset since different users have different "ignore" classification, and some queries are also needed to determine user rights and settings. Thanks, West.andrew.g (talk) 21:57, 21 April 2012 (UTC)
Independently as in, not having to rely on it's own server. Great tool however. I enjoy it more than Huggle. : ) --Rsrikanth05 (talk) 08:07, 22 April 2012 (UTC)

Flattr

Hi, have you ever thought of having a Flattr button on your website? --Rsrikanth05 (talk) 18:41, 27 April 2012 (UTC)

This is the first I have heard of the Flattr service, as I have never seen it in the wild before. When you refer to "my website", I am also unclear if you mean my academic homepage or the STiki talk page. Regardless, I'm not sure it would be an entirely appropriate move. I am a well-funded PhD student at the University of Pennsylvania, and my work on STiki is related to my dissertation. I simply don't have the time to work on STiki as much as I would like; there is no financial barrier. Moreover, I am not interested in profiting off my software's existence. There are probably millions out there contributing to Wikimedia and other free-knowledge projects; I don't see how my contribution is any more significant. If you feel inclined to show your thanks, do so by being a STiki talk page watcher, spreading the word about STiki, or following some of the other STiki promotional ideas (this was recently posted on a talk page, not sure if it was this one, or Wikipedia_talk:STiki). Thanks for your support, West.andrew.g (talk) 19:01, 27 April 2012 (UTC)
It was just an idea. I came across the idea on Article traffic stats page. It was just a suggestion. I've been suggesting it to various people. Oh and, I really love STiki. It does things different from Huggle. I quit Counter Strike to use Huggle, and now I enjoy STiki better than Minesweeper [I love Minesweeper]. --Rsrikanth05 (talk) 06:03, 28 April 2012 (UTC)

Congrats and Cheers

Thanks DBigXRay! Award moved to the "awards" section of the main WP:STiki page. Thanks! West.andrew.g (talk) 20:52, 28 April 2012 (UTC)

CHANGELOG for 2012-04-11 STiki Release! (and 2012-04-12)

Significant changes were rolled out in this version, causing the version number to bump up 2.0 -> 2.1. Many of these were front-end modifications to the GUI. These feature requests were a little easier to accommodate than I had expected, meaning: (a) STiki has a nice organizational framework or (b) I've screwed up and I am about to be bombarded with bug reports. The major changes are as follows:

  • Good-faith revert is now available as a fourth classification option. Good-faith reverts do not post any messages to the offending user's talk page. The comment left with good-faith reverts is visible -- and can be modified -- via the (newly!) tabbed comment panel (see the screenshot). The leader-board now also contains a good-faith ("AGF") column.
  • The notion of rollback is now fully integrated. Any revert action is now a "rollback" (including good-faith revert). For users that do not have the native rollback permission, we implement rollback functionality in software. To this end, the diff browser now displays all edits that will be undone (not just the most recent one). If more than one edit would be rolled-back, this is noted beneath the article title in purple font (see the screenshot).
  • Settings are now *persistent* between STiki sessions. This includes settings such as checkboxes, form fields, and window size/placement. Only *queue selection* is not stored; this remains server-set so users can be steered away from queues that are experiencing difficulty. Configuration is stored in XML format in the user's home directory. Special thanks to User:Meiskam for his assistance with this.
  • Backend: Logic added so CBNG queue will not enqueue edits made by bots.

As always, thank you for your continued use of the STiki tool. Please report any bugs you encounters. Shortly STiki will past the 100,000 revert mark, and this should be a great milestone in the software's history! Thanks, West.andrew.g (talk) 16:46, 11 April 2012 (UTC)

Piggybacking on this release, a few minor bug fixes and feature requests were fulfilled on new build released 2012-04-12. Notables fixes include:
  • The "activate hyperlinks" setting is now correctly read and applied from the user configuration XML. (T#007)
  • Functionality has been added so the diff-browser can be scrolled using the keyboard. The keys are PG_UP and PG_DOWN (dramatic scrolling) or UP_ARROW and DOWN_ARROW (more subtle). The "classification" panel must have focus for this functionality to work (i.e., in a state where pressing "v" would fire a "vandalism" revert). (T#001)
  • Good-faith reverts will no longer be marked as "minor" (as vandalism/ spam edits are), per the policy of WP:Minor. Because native RB cannot be marked "minor", good-faith reverts must use "standard" editing calls to simulate rollback functionality ("software rollback"), incurring some processor/bandwidth overhead.
Thanks again, West.andrew.g (talk) 03:10, 12 April 2012 (UTC)

Helping with Stiki

Hey, I would quite like to help with stiki but I don't really know how I can. I could do some beta testing if needed. Cheers, Willdude123|Ƹ21ɘbublliW (talk) 13:09, 28 April 2012 (UTC)

Do you have any technical/programming capabilities? If not, then hang around as a talk-page lurker, keep using the program, and maybe cooperate with User:Orphan Wiki on this new project to recognize new STiki users with welcome messages, and long-term users when they cross major usage thresholds (i.e., 1000 classifications). You've been around for a while, and I appreciate your use and support. Thanks, West.andrew.g (talk) 19:28, 28 April 2012 (UTC)
I'm learning some Python but apart from that, not really. What kind of technical skill would be needed?Willdude123|Ƹ21ɘbublliW (talk) 21:13, 28 April 2012 (UTC)
No technical skills are needed. That is just generally how the workload breaks up: technical tasks (programming, bug fixing, etc.) versus less technical ones (awarding barnstars, talk-page watching). If you want to help out and fall in the "non-technical" bin, then those are the types of task you/we should be looking for. Nothing too pressing at the moment; although you could help with keeping this "STiki welcome message/barnstar" effort moving. Thanks, West.andrew.g (talk) 15:41, 1 May 2012 (UTC)
What things would you like help with on the technical end of things? While my main programming language is Perl (I did my bioinformatics dissertation research using it), I can understand what is happening in a Java program and probably do some edits to it. Allens (talk | contribs) 17:50, 1 May 2012 (UTC)

The Welcome Message

OK, so we're thinking about developing a welcome message for newer users of STiki and also a series of barnstars, recognising the attainment of a certain quantity of edits, the amounts of which we have discussed above. The first that I will help to address is the former. I thought it'd be best to begin a fresh thread, to make things easier to sift through in sections, and also to isolate certain points of interest and thus maximise focus and attention.

A welcome message will have to be built around a number of links to various pages within our STiki community. For a start, I think we should all think about which pages would be the best to include (as internal links) in the STiki welcome template here, before the scribing of the template commences. Obviously the pages of the most relevance and use to a newbie will be inserted. Thoughts? Orphan Wiki (talk) 17:09, 28 April 2012 (UTC)

I think the most obvious links should be to: Wikipedia:STiki, Wikipedia_talk:STiki, Wikipedia:STiki#Using_STiki, Wikipedia:STiki/leaderboard (maybe to get them feeling a little competitive, to encourage future use), Wikipedia:Vandalism (and any other policy-related links), and Wikipedia:STiki#Userboxes_and_awards (to encourage support through user-boxes). This should be a good start; draft something up! Thanks, West.andrew.g (talk) 19:37, 28 April 2012 (UTC)
OK, well I've written up a first draft over at my sandbox. This template follows the basics of a standard Wikipedia welcome message; short, simple and to the point. There are, of course, the options of adding a picture or making the design look a bit more exciting (although I'm not a master of formatting).
Any suggestions? Orphan Wiki (talk) 11:50, 29 April 2012 (UTC)
For ease of maintenance, I copied your sandbox over to Wikipedia:STiki/welcome template and made a few modifications there (so others should feel free to edit). Principally, I wasn't a big fan of the "when to use rollback" wikilink. Of the several reasons to use rollback, STiki only really addresses one of them (vandalism) and the description of vandalism they provide isn't nearly as elegant as the other link we provide to WP:Vandalism.
Content wise, I think it is pretty reasonable. It would be nice to make it a little more aesthetic. Maybe someone could come along and "box it up" (in the style of most barnstars) and work the STiki logo in there somehow? With that, I think its in decent shape to distribute (though anyone is free to take an editing pass). Thanks, West.andrew.g (talk) 16:02, 1 May 2012 (UTC)
Great, and thanks for the tweaks. Have we to wait until someone "boxes" it up before distributing the welcome message then? Orphan Wiki (talk) 22:13, 1 May 2012 (UTC)
I just recently boxed it up and added the image. I think its fine to begin nightly distribution and see what kind of feedback we get.
When designing the "milestone" box(es), it might be easiest just to take the 'anti-vandalism barnstar' template, and add the STiki image and a bit of positive STiki wording to the body. Thanks, West.andrew.g (talk) 19:12, 2 May 2012 (UTC)
OK, I'll begin with the nightly distribution straight away. When some free time arises over the next few days, I'll also look into the barnstars. Orphan Wiki (talk) 22:29, 2 May 2012 (UTC)

BAG Approval

I saw in your block log that you once got blocked for running scripts without BAG approval. Did you get BAG approval? How? --Tomtomn00 (talkcontributions) 12:25, 29 April 2012 (UTC)

Short answer, "no", my account does not have BAG approval. Long answer found here. Thanks, West.andrew.g (talk) 15:44, 1 May 2012 (UTC)