User talk:Kevan
If you want to report a bug, put it up on the Bug Reports page for review.
If you've got a question about the Wiki, use the Wiki Questions and another user will see it and answer it.
If you want to discuss game balance or the future of the game, try a discussion forum.
If you'd like to respond to someone who's posted here, you should do so on their talk page.
If you're from the future, there are three archives to read.
If you'd like to speak to an operator, please hold.
ENORMOUS BOT PROBLEM
Kevan, this wiki is literally being overrun with bots. It was a pain enough when there were about 10 a day, now there are sometimes 10+ an hour. There'd have to be something you can do about this? Please? DANCEDANCEREVOLUTION (TALK | CONTRIBS) 00:05, 28 March 2012 (BST)
Kevan, I agree with DDR that something should be done to help minimize spambot edits. Here are some tips from MediaWiki on combating wiki spam. The ConfirmEdit extention we have installed on this wiki could be slightly reconfigured to prevent all non-autoconfirmed users (users with less than 25 edits or one week from user creation) from adding urls. Sysops can even create a whitelist of urls which would be allowed. Use of the expression $wgSpamRegex would also dramatically help reduce spam. It's hard to make spam go away for good but with just a few adjustments to the existing wiki software, it can be greatly reduced. ~
05:34, 28 March 2012 (UTC)
- Ban the ips guys. 300 bots checked, only 50 unique IPs with an average of three registrations per visit. About two thirds were already banned by other sysops. --Karekmaps 2.0?! 10:44, 28 March 2012 (BST)
- *cough*GeneralSpiderzedVapor*cough* --•▬ ▬••▬ • •••• •▬ ▬•▬• ▬•▬ #nerftemplatedsigs 13:58, 28 March 2012 (BST)
- Oh well if its only 50 unique IPs, it must not be a problem... Manually blocking spambot iPs is a band-aid fix at best; a knee-jerk process unofficially put in place in response to the fact that some of the current spambots are inexplicably reusing IP adresses rather than proxies. It will ultimately prove fruitless. On the other hand, the suggestions I've posted above have been researched and dicussed by members of this community, and they are suggested setting by MediaWiki devs. The discussion pages for the pages I've linked above are also chalk full of other great suggestions from wiki admins that are suffering the same problems as us. These spam blocking processes are well worth looking into and implementing. ~ 15:39, 28 March 2012 (UTC)
- This is how spambots generally function actually. Botnets rarely have access to tor or other proxies and generally use userowned servers that are one use in part because it can actually check if it's already banned on an initial test account. We also actually have most of the tools you listed but it's generally irrelevant to botservs what the url is because if they're doing it for profit at this level they normally have enough domains that we couldn't really get an effective list together anyway without just up and up blocking ALL external URLs. The name of the game is backlinging but the bots don't normally check if they're actually getting the backlinks. If you want to effectively target these sites to stop them appearing you could always try reporting them to google but that won't make our spam load smaller. --Karekmaps 2.0?! 19:32, 28 March 2012 (BST)
- Blocking all external urls is exactly one of the methods that has been suggested. But only for non-autoconfirmed users and only urls not whitelisted (all commonly used domains like urbandead.com would be allowed regardless of user status). Again, manually blocking spambot IPs isn't something we be should relying on. Not when so many other options are available. It really would just be a matter of adjusting two or three already present regular expressions. ~ 04:33, 29 March 2012 (UTC)
- My point was more that none of the solutions suggested actually fix the problem. They also generally don't need his involvement. Now adding a honeypot would be a solution that might help and would require action on Kevin's part, alternatively we could try and use the already existing whitelist and blacklist functions which were actually added back when 3pwv was an actual problem but, as I stated above, they wouldn't actually help overly much in this case beyond banning all nicks ending in numbers. Also, as a note, on your link what they actually suggest is a honeypot method in addition to the other methods as those are for additional security for unlisted or advanced bots. --Karekmaps 2.0?! 10:59, 29 March 2012 (BST)
- Also worth note: $wgSPAMRegex is nothing more than another whitelist for text. It's not actually any more useful without a generic bot template spamming you. In the end it all boils down to the most inclusive lists and tricking the bot on registration. --Karekmaps 2.0?! 11:08, 29 March 2012 (BST)
- sigh...Karek, my suggestions above were the culmination of research, testing and discussion on other parts of the wiki. The community wants stricter spam blocking but its going to require a few minor changes on the server end. Let's just let the man look it over and decide without muddying the waters, please. ~ 15:21, 29 March 2012 (UTC)
- Yes and my response is from years of this plus dealing with this problem on multiple other wiki's in Kevin's role. The point is to save him time, specifically, what he needs to do to have any real effect is an autoupdated list or honeypot like the lists from stopforumspam. Everything else won't actually do much anything and with the current bots most definitely will not have an effect. --Karekmaps 2.0?! 23:08, 29 March 2012 (BST)
- sigh...Karek, my suggestions above were the culmination of research, testing and discussion on other parts of the wiki. The community wants stricter spam blocking but its going to require a few minor changes on the server end. Let's just let the man look it over and decide without muddying the waters, please. ~
- Also worth note: $wgSPAMRegex is nothing more than another whitelist for text. It's not actually any more useful without a generic bot template spamming you. In the end it all boils down to the most inclusive lists and tricking the bot on registration. --Karekmaps 2.0?! 11:08, 29 March 2012 (BST)
- My point was more that none of the solutions suggested actually fix the problem. They also generally don't need his involvement. Now adding a honeypot would be a solution that might help and would require action on Kevin's part, alternatively we could try and use the already existing whitelist and blacklist functions which were actually added back when 3pwv was an actual problem but, as I stated above, they wouldn't actually help overly much in this case beyond banning all nicks ending in numbers. Also, as a note, on your link what they actually suggest is a honeypot method in addition to the other methods as those are for additional security for unlisted or advanced bots. --Karekmaps 2.0?! 10:59, 29 March 2012 (BST)
- Blocking all external urls is exactly one of the methods that has been suggested. But only for non-autoconfirmed users and only urls not whitelisted (all commonly used domains like urbandead.com would be allowed regardless of user status). Again, manually blocking spambot IPs isn't something we be should relying on. Not when so many other options are available. It really would just be a matter of adjusting two or three already present regular expressions. ~
- This is how spambots generally function actually. Botnets rarely have access to tor or other proxies and generally use userowned servers that are one use in part because it can actually check if it's already banned on an initial test account. We also actually have most of the tools you listed but it's generally irrelevant to botservs what the url is because if they're doing it for profit at this level they normally have enough domains that we couldn't really get an effective list together anyway without just up and up blocking ALL external URLs. The name of the game is backlinging but the bots don't normally check if they're actually getting the backlinks. If you want to effectively target these sites to stop them appearing you could always try reporting them to google but that won't make our spam load smaller. --Karekmaps 2.0?! 19:32, 28 March 2012 (BST)
I've added the autoupdated DNS blacklist, and am just putting a honeypot in place now. --Kevan 17:21, 24 May 2012 (BST)
- Thanks, Kevan. It seems to be working....maybe a bit too well. I just tried creating a new account and I got a blocked by DNSBL message. Its the same IP I'm using to edit right now, too. ~ 01:09, 25 May 2012 (UTC)
- Odd, I thought it was using the same check for edits as for new accounts. Say if people think it should be switched off, or using different blacklists; it's currently just the Spamhaus and Tornevall DNS blacklists suggested by MediaWiki. --Kevan 09:12, 25 May 2012 (BST)
- Yes, it is kind of odd. For me, the culprit is Spamhaus. My IP is part of a range that is on their pbl dnsbl. Apparently its a policy for my ISP. I'll send you a private email with more info. ~ 15:30, 25 May 2012 (UTC)
- Yes, it is kind of odd. For me, the culprit is Spamhaus. My IP is part of a range that is on their pbl dnsbl. Apparently its a policy for my ISP. I'll send you a private email with more info. ~
- Odd, I thought it was using the same check for edits as for new accounts. Say if people think it should be switched off, or using different blacklists; it's currently just the Spamhaus and Tornevall DNS blacklists suggested by MediaWiki. --Kevan 09:12, 25 May 2012 (BST)
I've now switched to sbl.spamhaus.org, and have replaced the ReCAPTCHA with QuestyCaptcha. It'll only kick in on account creation, for now (so we may still see some spam from sleeper bots), but we'll see what hapens. If we want to ratchet up QuestyCaptcha to interrogate all edits that add links, let me know. --Kevan 16:30, 25 May 2012 (BST)
- The new dnsbl does seem less restrictive. I was able to create an account, at least. Questy worked like it was supposed to. I like that it is based on game related things. In the time since it has changed, we haven't received any spam edits, where regularly we would have had more than a dozen. I'm feeling good that you've foiled a good lot of them and we'll be removed from whichever target lists we were on. Cheers! ~ 00:37, 26 May 2012 (UTC)
Archive
I archived your talk page, hope you don't mind but it was starting to get out of hand after almost 3 years. It was getting obnoxiously long, more importantly I was experiencing some noticeable strain on multiple PC's I used when loading this page. You can find the new archive here. -- 
Thadeous Oakley Talk 14:31, 1 May 2012 (BST)
- Thanks. --Kevan 09:31, 25 June 2012 (BST)
This is not, the greatest game in the world...
Just wondered whether you used "You search and find nothing" as a tribute to the song of the same name in Beneath a Steel Sky? --Rosslessness 21:53, 21 June 2012 (BST)
- Loved the game at the time, but no, just a coincidence. --Kevan 09:31, 25 June 2012 (BST)
Also
|
Borehamwood |
|---|---|
|
|
Now it's dead, can we open up new character creation again?
I'd love to do some three year old repairs. --Rosslessness 15:55, 23 June 2012 (BST) 
3 people like this comment.
- Afraid not - the contract with Channel Four specifically involved removing access to new characters once the promotion was over. Same deal for Monroeville. --Kevan 09:31, 25 June 2012 (BST)
- Ooh, there's an interesting titbit. As you're so chatty, can I ask how many man hours went into setting up the new cities? --Rosslessness 17:16, 25 June 2012 (BST)
- Two or three days, including the research? I wasn't billing it per hour, so I don't really remember. --Kevan 13:30, 29 June 2012 (BST)
- Including the layout planning? How did that work? Its not all Randomly generated surely? (Last question I promise)--Rosslessness 14:16, 29 June 2012 (BST)
- Partly. Apart from the unique locations and some manual tinkering, the maps were just drawn out as zoned areas, and the urban areas were generated randomly within the spaces assigned to them. --Kevan 14:21, 29 June 2012 (BST)
- Including the layout planning? How did that work? Its not all Randomly generated surely? (Last question I promise)--Rosslessness 14:16, 29 June 2012 (BST)
- Two or three days, including the research? I wasn't billing it per hour, so I don't really remember. --Kevan 13:30, 29 June 2012 (BST)
- Ooh, there's an interesting titbit. As you're so chatty, can I ask how many man hours went into setting up the new cities? --Rosslessness 17:16, 25 June 2012 (BST)
- as you seem to have some time to answer stuff: are we likely to see any new cities or attempts to attract new players? --Honestmistake 23:15, 30 June 2012 (BST)
- Dammit, you broke Kevan D: --RadicalWhig 09:35, 8 July 2012 (BST)
Hi Kevan
Would it be possible to get a few wiki-things installed? Namely, InputBox. I don't have a clue how easy or difficult this would be. There was a discussion on this, and people also wanted collapsable tables; I'm not sure if they require installing anything. We gave things a try on our end but couldn't get anything to work. Karek said something about jQuery. I have no idea what that means. Thanks. -- AHLGTG THE END IS NIGH! 02:49, 25 July 2012 (BST)
- It means he'll have to do a version update to be able to implement that particular feature as it is in standard MediaWiki releases. InputBox shouldn't have this issue and looks to be designed with backwards compatibility in mind up until major version differences. --Karekmaps 2.0?! 06:03, 25 July 2012 (BST)
Actually Kevan, on some Show Preview testing on Wikipedia, I couldn't get it to work the way I wanted to. I wanted to be able to enter text into the field, hit enter, and have it take me straight to that page, in the User namespace, with a particular user name, and as a subpage—say, to User:A Helpful Little Gnome/what I typed in. Doesn't seem to work like that. Oh well. Hope I didn't waste any of your time. Thanks. -- AHLGTG THE END IS NIGH! 01:12, 31 July 2012 (BST)
