Debate:Should external links on RationalWiki be flagged as no follow?

This is a Debate page. Feel free to add your own spin on the story. Please keep it civil! This debate was created by signed users.

Currently every external link on the site is flagged as no follow. This applies not just to the [single bracket link], but also interwiki links such as [[wp:something]] or [[cp:something worse]], as well as all the links in the voting extension or that use the capture tag.

The question is, should we maintain this policy, remove all no-follow tags, or something in between?

Maintain current policy of flagged links

My first instinct is to opt for the status quo unless someone can come up with a good case for changing it. Wikipedia does alright with nofollow links. There have been some suggestions that it might benefit our Google rankings. I would suggest that the best way to improve Google rankings is with good articles rather than Ken-type SEO. Генгис^{You have the right to be offended; and I have the right to offend you.} 22:21, 28 July 2009 (UTC)

My concerns have nothing to do with our ranking. I personally find the whole idea of no follow distasteful. It is about receiving the benefit of incoming links to our site, but refusing to extend that same service to other sites. We become a "black hole" for the search engines, links come in but nothing comes out. The whole idea of collaboration and egalitarianism that underlies a "wiki" and even our specific site policies seems to be undermined by this action. tmtoulouse 22:28, 28 July 2009 (UTC)

When you put it like that then I can see the point. Why didn't you proffer that rationale straightaway? Генгис^{You have the right to be offended; and I have the right to offend you.} 22:36, 28 July 2009 (UTC)

I don't know. The answer is probably somewhere between laziness, not totally sure what I think about it, and wanting to see what would pop up organically without my immediate take. tmtoulouse 22:37, 28 July 2009 (UTC)

I like the current policy, at the moment I can stick down a link with no concern about how it will effect affect (delete which is inappropriate) the persons webshite for a popular search starting with [A-Z] at a search engine starting with GBY. We can link to people we would not otherwise want to help out with no fear of assisting them in anyway. We could a code or template that allows follow I suppose, that I would support as it will allow the editors to selectively choose the pages we are going to assist. I would like no follow to still not work on talkpage though, the same as comment sections on a blog. - π 23:54, 28 July 2009 (UTC)

I think it would be nice to have a tag like follow (link) /follow, only perhaps easier to type (2 or 3 letters?), we also link to many places we would like to "assist", like our blogs, and the many "good guys" out there. That's if we keep the general "nofollow" policy. ħuman 00:40, 29 July 2009 (UTC)

I didn't get it until TMT wrote his comment above. That makes a lot of sense. Sterile dog 02:00, 29 July 2009 (UTC)

Thinking about this a bit more, we already have reciprocity with our links to external sites in that people who visit here can click to an external site. The issue is really one of helping other sites with their search-engine ranking. Do we really want to boost the importance of daft Julie and other nutter bloggers or the pile of crap at Conservapedia just because we find it ridiculous or amusing? I certianly don't want to contribute to Ken's SEO machinations. I think the nofollow should stay but we should allow exceptions. As we don't want to make adding the exception too much hassle I suggest that it could be done with three square-bracket pairs which would add only two characters (sorry, it would be four for external links). It would be easy to remember but unusual enough that drive by spammers would miss it. Генгис^{You have the right to be offended; and I have the right to offend you.} 08:17, 29 July 2009 (UTC)

I like Genghis' suggestion since it would be (as I mentioned below) a sort of "Seal of Approval" (Thinking about it a bit more, could/should we maybe limit this to non-talk namespaces so that our article link policies/habits can take care of the "policing"?). Back before we had nofollow for CP links (okay, and before he realized that normal links have nofollow anyway), Ken was here every other day, spamming links to his pet articles and abusing us as a platform for his SEO wanking. I can't say that I miss that. --Sid (talk) 12:42, 30 July 2009 (UTC)

Remove all no-follow tags

• !vote to do so. ħuman 03:14, 1 August 2009 (UTC)

Other options

I understand that "no follow" is the default setting on mediawiki and that its purpose is to deter spammers - as the nofollow attribute means that their sites will not get a higher google ranking as a result of the link. Its utility is debatable as:

• Many spammers probably do not understand the issue anyway.
• Our highly active user base and our legion of eager sysops will clear out any spam quite quickly.

So why not disable it? Well, we link to a lot of weird sites and one especially weird site in particular. Do we really want to give them the extra google hits? How selective could we be with nofollow? Is it "all or nothing"? If we can do something selective then why not go for it? If not .......--Bob^{Not Jim} 20:33, 28 July 2009 (UTC)

Questions/Comments

One think that I've always wondered about this. Would disabling nofollow improve RW's google ranking?--Bob^{Not Jim} 20:35, 28 July 2009 (UTC)

No, there is no advantage in it directly, though indirectly I suppose there might be. IE people doing a tit-for-tat type arrangement with the tags. But I don't think this is common. tmtoulouse 20:37, 28 July 2009 (UTC)

It's just that I seem to recall reading that the Google page-rank algorithms gave some value to outgoing links as well as incoming ones. Presumably with nofollow set the outgoing links would be zero. But I can't seem to find where I read that now and I might well have been mistaken. (Or perhaps somebody on the internet was wrong - but that can't happen, can it?)--Bob^{Not Jim} 20:50, 28 July 2009 (UTC)

Disabling nofollow would at least justify our "policy" of not linking to metapedia... ħuman 21:18, 28 July 2009 (UTC)

Difflinks

How does google "see" weird links like diffs? As links to the current version? ħuman 21:21, 28 July 2009 (UTC)

A diff link is just an html page, google will see it as it appears to us. tmtoulouse 21:22, 28 July 2009 (UTC)

But will it raise the profile of the "undiffed" version, or just the basic www as a whole? ħuman 23:44, 28 July 2009 (UTC)

I would imagine google probably ignores the php arguments in a link, and so a link like http://www.conservapedia.com/index.php?title=Essay:Best_New_Conservative_Words&diff=676224&oldid=675066 would probably appear as http://www.conservapedia.com/index.php?title=Essay:Best_New_Conservative_Words. tmtoulouse 23:56, 28 July 2009 (UTC)

I don't know how Google "sees" those links, but if you check the HTML intro code of those two links have different meta data: The diff link has meta name="robots" content="noindex,nofollow" in the intro, so Google wouldn't index it either way (if I understood it correctly). --Sid (talk) 12:35, 30 July 2009 (UTC)

Yes, because Mediawiki hides diffs and old revisions from search engines that way. Nx (talk) 10:12, 31 July 2009 (UTC)

Subtlety of distinction

If we decide to go the partial nofollow route, how "smart" can the software be? Like, say, nofollowing any link in a "capture" tag? Nofollowing any diff or permalink? Nofollowing any "shortcut" (ie, wp:some shit, cp:some thing else) link? ħuman 21:28, 28 July 2009 (UTC)

I suppose it could be as smart as we want it to be. One issue is that external link formatting is not handled by a single proccess or chunk of code but divided up amongst several extensions and internal code for the MediaWiki software itself. tmtoulouse 22:39, 28 July 2009 (UTC)

Could individual links be tagged as : eg <nofollow> ... </nofollow>? ^{This message brought to you by: respond}_and honey 00:10, 29 July 2009 (UTC)

See my comment a few sections above after I write it. ħuman 00:39, 29 July 2009 (UTC)

Would it be better to "blacklist" certain sites, such as (www\.)?conservapedia.com for nofollowing and then let the rest be plain vanilla links? Also, I have the vague impression that something like 80% of the links on this site are to sites that we find in some manner objectionable (religious nutters, cranks, scam artists, etc.) This would suggest to me that nofollowing should be opt out, rather than opt in. --JeevesMkII ^{The gentleman's gentleman at the other site} 21:15, 29 July 2009 (UTC)

I... don't know. Active blacklisting like this looks like direct discrimination against a few sites to me. And while I wouldn't exactly mind blacklisting CP, I like the approach of exceptions (like the "[[[link]]]" suggestion somewhere on this page) more. It would be like a sort of "Seal Of Approval". --Sid (talk) 12:38, 30 July 2009 (UTC)

(undent) Interwiki links are not nofollowed by mediawiki, that's done by an extension. The extension has a whitelist at Mediawiki:iwnofollow-whitelist. Mediawiki 1.15 allows whitelisting external links in LocalSettings.php with $wgNoFollowDomainExceptions Nx (talk) 10:15, 31 July 2009 (UTC)

Sounds good, exactly what we need. Are we going to 1.15? If so can we get dynamic tabs so we can replace that javascript hack for the [0] tab. - π 10:25, 31 July 2009 (UTC)

Even with MW1.14 we can disable it on certain namespaces with $wgNoFollowNsExceptions Nx (talk) 07:46, 2 August 2009 (UTC)

About the upgrade: I haven't checked the 1.15 changelog thoroughly, but it seems there's nothing really important there. Since our modifications to mediawiki are very small, upgrading should be easier this time. We can upgrade if Trent wants to do it. Nx (talk) 12:38, 2 August 2009 (UTC)

NOTE

• Google states that their engine takes "nofollow" literally and does not "follow" the link at all. However, experiments conducted by SEOs show conflicting results. These studies reveal that Google does follow the link, but does not index the linked-to page, unless it was in Google's index already for other reasons (such as other, non-nofollow links that point to the page).
• Yahoo! "follows it", but excludes it from their ranking calculation.
• MSN Search respects "nofollow" as regards not counting the link in their ranking, but it is not proven whether or not MSN follows the link.
• Ask.com ignores the attribute altogether.

(from WP) so it's not mandatory anyway. ^{This message brought to you by: respond}_and honey 00:19, 29 July 2009 (UTC)

What is the difference between "follow" and "rank" in this case? Surely if it "follows" but does not "rank" that is the same as "not follow" and not "rank".--Bob^{Not Jim} 06:20, 29 July 2009 (UTC)

From wp: " ... others still "follow" the link to find new web pages for indexing ..." ^{This message brought to you by: respond}_and honey 06:25, 29 July 2009 (UTC)

It is an instruction to robots. Nice robots obey orders. Mean robots eat people and colonize Mars, from whence to attack us. ħuman 06:44, 29 July 2009 (UTC)