Stephan Nedregaard

Links in regular content not followed by search engines anymore

Wednesday, January 19, 2005 10:15:34 AM

I was made aware of this article by Google:
http://www.google.com/googleblog/2005/01/preventing-comment-spam.html

We have already implemented this in all content areas (journal posts, journal comments, etc.). The reason we do this is to avoid spammers polluting the site.

We are working on a better way of finding out which links this should apply to, as I think links should be followed in the journal post itself, as long as the article is serious.

Journal comments: Fixed!Changes to "Most visited" on the journals front page

Comments

Moose # Wednesday, January 19, 2005 10:15:34 AM

How will you determine whose article is serious?

Rijk # Wednesday, January 19, 2005 10:15:34 AM

Maybe content-to-links ratio? If there is little text, just links in a journal post, the post isn't likely worth finding...

Moose # Wednesday, January 19, 2005 10:15:34 AM

Rijk, but that solution would rule out a link-logg, à la Anne's HREF, or Mark's ex-B-Links...

Jonny Axelssonjax # Wednesday, January 19, 2005 10:15:34 AM

Links to opera.com are serious, everything else kind of dubious...

Rijk # Wednesday, January 19, 2005 10:15:34 AM

Nothing stops people from having a linklog, my idea just means that the linklog journal entries will not be found by Google e.a., and the linked sites don't benefit from my.opera.com's Pagerank.

Jonny Axelssonjax # Wednesday, January 19, 2005 10:15:34 AM

At least one notorious journal spammer has been removed, which is great. But I have noticed that the comment link isn't updated when this happens, so the line goes "N comments, last by <spammer>". Minor issue, but should be fixed.

Moose # Wednesday, January 19, 2005 10:15:34 AM

Well, my sheer humanity surpasses the electric eclecticity of the automated scriptorium at her.opera.com. I take one look at "random 10 posts", and I see three ALL-CAPS posts, written in MX Spanish, all identical. They will sure pass through filters like this.

ho-hum,

M.

Stephan Nedregaardstephan # Wednesday, January 19, 2005 10:15:34 AM

I noticed them. I just didn't bother to do something about it as there were only three.

There are a lot of things we plan on doing to prevent people from posting the same post three times (probably due to a human error), these will be introduced in the next major version.

We could back port these solutions, but generally don't because that would mean implementing it twice, and we're really focused on getting a new (and MUCH better) version out!

Stephan Nedregaardstephan # Wednesday, January 19, 2005 10:15:34 AM

Minor issue that we can't do much about right now, I'm afraid. We will however fix this for the next release.

Stephan Nedregaardstephan # Wednesday, January 19, 2005 10:15:34 AM

I said:

"We are working on a better way of finding out which links this should apply to, as I think links should be followed in the journal post itself, as long as the article is serious."

As this generated so much interest, I'll share some of my thoughts with you. These do not neccessarily represent what will be implemented in the new site:

- Recommendations and reporting by other users:
Based on your statistics regarding being reported or recommended by other users, your posts will be given a nofollow value.
- Number of commenters: If more than x people commented on your post, we will not do a nofollow.

Any other ideas?

Moose # Wednesday, January 19, 2005 10:15:34 AM

Well, now there are 7 posts from one guy, including 5 identical, and 2 very similar. This is just the beginning. Not that I monitor the journals - I see this sort of thing every time I click the journals link.

M.

Moose # Wednesday, January 19, 2005 10:15:34 AM

Well, you know what's coming if you allow other people to determine whether a given journal author's links are to be followed?

Humans are spiteful, vile, and malicious; they love mudslinging. They will stop at nothing to drive down someone they dislike. There are some lessons to be drawn from the Amazon customer reviews, for example.

You are better off trusting a machine, and improve the machine, if need be. Don't leave anything to humans. They are not worth it.

M.

orinoco # Wednesday, January 19, 2005 10:15:34 AM

I'm sorry Moose, I can't let you do that

/Hal 9000

orinoco # Wednesday, January 19, 2005 10:15:34 AM

Is there any way you can add support for <a href="http://my.opera.com/orinoco/journal/122">bookmarklet links</a> & the class="OpBut" attribute used in the forums? Being able to set these up as working hyperlinks will make customisation tips much easier for users to share through their journals.

Thanks!