[Edit: Added a quote from the hoax news bit near the beginning some minutes after publishing the post. Also added the sentence "and this in turn may also give the 'link democracy' another chance to get it right, by removing links to the site" near the end. Also: added '[as in "cover"]' explanation based on a follow-up comment at Sphinn by Matt. Later edit: I linked the word "guidelines" to the webmaster guidelines.]

My world is getting complicated.. Should the conversation move to FF or not ??

As such my view is simple " "fact times importance equals news" – and I wonder how google can actually verify any type of linkbaiting sites that spin stories.. the users are the gulliable ones..and if a webmaster can do that.. I see no reason why not

Google has been doing this every year on April 1

I think the whole thing is a waste of time: it's not the first deceiving article and it's certainly not the last one. Search engine shouldn't decide if an article or some information is accurate, since this is subjective and impossible to determine in many situations. The job of a search engine is to find what reliable sites think it's accurate.

There are two issues here at stake:
1) Integrity of news and SEO.
2) Google as the jurisdiction or God, deciding what's true and what not

As to 1)
While the tactic was very effective in the short term for it's SEO outcome it's generally bad for the image of the site and business that did it. So in the long run it backfires.

From the SEO perspective the gaining of authority via links is partly outweighed by the loosing of authority due to bad press etc.

From the news perspective it was a dark chapter in the history of media that they just reprinted the story without checking the source or the facts.

2) Now this point is utterly ridiculous. Matt Cutts obviously did not study postmodern philosophy (poststructuralism) as we know at least since the fifties that there is no single truth out there, just points of view. So when Google starts to decide what's true and what not we will have a dictatorship like China. Google has completely no business in policing publishers whether they are writing the truth or not, whatever this might be.

Now this example just shows what we already knew before: Relying solely on popularity for judgement of worth is inherently flawed. Germany's most popular "newspaper" BILD (2 mill sold daily) is not even allowed to call itself newspaper as it contains more made up stories than real news. Anyways, the day engineers like Matt Cutts decide what's true I leave this planet. A philospher once said "I know that I don't know". Having such a powerful position like Matt Cutts makes him obviously prone to megalomania.

That said, I won't make up such stories as linkbait myself, but I just read a very similar story a few days ago online which was "true". So just dig deep enough and you find plenty of these. Lyndon is a writer and poet so he made one up.

I assume based on the Matt Cutts statements, that all link-juice to FoxNews will be disallowed...

BTW, here's an earlier case of an SEO guy creating a hoax news story in order to garner Page Rank: the Russian spammer murdered story. http://taint.org/2007/10/11/203243a.html

Google claims they are hands off, yet the sole purpose of Google News is to act like an editor.
Maybe "news" search results should be weighted by a "truthiness" scale, via wikipedia or whatever. And to answer a previous point raised, if I am looking for a news article on a given subject the last result I want to skim through is an Onion article.

Yeah, one on blogoscoped now about "Google Maps Living Room View"

Tadeusz Szewczyk – Amen.

I'm against playing god (ironic, huh?). People make their bed and must sleep in it. If people are "foolish" enough to link to an article like the one mentioned here, they deserve a spammier web. Google doesn't need to save us from ourselves if they truly believe that the web is naturally democratic.

Great reading, as usual.

> Google already has human evaluators which indirectly affect rankings – as they can influence which ranking algorithms Google picks (even when Google argues the “beliefs and preferences of those who work at Google, as well as the opinions of the general public, do not determine or impact our search results”) –

I know this is one of your leitmotivs. However, don't you think that they are referring to _particular_ search results about a given keyword, and not about _general_ algorithms that affect every search result?

Milivella, I think of it as this. Take this hypothetical example, just to clarify the point.

<the-specifics-of-this-are-hypothetical>

1. The Google ranking engineer team tries out a new algorithm. The algo tweak, let's call it #30121, will be seen by 50% of human evaluators only, it won't go live for now.
2. The human evaluators who see the algo tweak get presented a couple of search results. In these results, they check how useful the top sites are in relation to the sample query. The other 50% evaluators working on the old algo do the same.
3. The Google rankinging engineer team collects the statistics. They now have data like "the proposed algo tweak #30121 increases results quality by 5%".
4. The engineers now go live with algo tweak #30121 because of the human evaluation.

This as a general thing, not related to hoax news. And I'm not judging on this approach here (though I'd love to have more transparency by Google how they pick their evaluators). Now there are two scenarios which could play into the "hoax news" debate, for instance:

- Variant 1: Google updates the human evaluator guidelines by adding a passage that say "Undisclosed hoaxes are now considered spam according to the webmaster guidelines; please rank results which have those in prominent places accordingly." Clearly, this could then influence the results quality statistics for a particular algo tweak Google is pondering (even if that algo tweak has no relation to hoaxes), which in turn could change future rankings as Google then may or may not go live with the change, based on that. (This change then affects a whole lot of site's, including particular hoax news XYZ, which may now be effectively missing from the top 10 results where it was formerly present.)

- Variant 2: Google decides hoaxes are bad and finds some algo tweak that somehow, automatically, judges upong the "hoax" level of a story. I don't think that is very likely, it seems to be near impossible to judge, but just to continue with this example; when the human evaluators now check the search result, they now report (or don't report) improved results quality. Based on this evaluation, Google decides to go live with their new anti-hoax algo (or not).

In the first variant, the key was the update of Google rater guidelines or Google's webmaster guidelines, which raters looks at. This guidelines update could be implemented by a relatively small group of people.

In the second variant, the key was the update of the algorithm to satisfy stronger anti hoax measurements, again a decision by a small group of people. (Not small as in e.g. 5 people or anything – I would think many more people are involved in these things. But still, a limited group of people working for Google.)

In both variants, it can be of enormous impact how Google hires the human evaluators, what people they are looking for for the job, which people they ask. It is not even as simple as picking a random part of the population (which in itself seems an impossible task, as not everyone likes to work as evaluator) because not everyone may be a good quality evaluator – what if perceived short-term quality is different from actual quality? What if for instance people would prefer a more visual, spiced up and pleasing search result for a short time but that it would be less beneficial for research in the future? And is Google ever hiring a large proportion of evaluators in the vicinity of Google – e.g. friends of employees, or an above average share of California residents? These seem to be some of the questions with perhaps no easy answers. If Google would make it more transparent, outsiders could make better judgments on it, of course.

This was just the part in relation to the human evaluators – it is entirely possible Google can just completely directly and manually decide that a particular hoax news XYZ should be downranked (just like they sometimes manually ban other things they consider SEO spam), which is yet another field.

</the-specifics-of-this-are-hypothetical>

Now, while the specific points of this were hypothetical and just to illustrate the point with an example (e.g. perhaps they're not using A/B testing but simply compare against old rating results etc.), I don't think the issue itself is hypothetical. We know Google has human evaluators, we know those evaluators have guidelines which are written by humans, and we know humans create the algorithms inside Google (well, let's put an AI writing itself to the side for now for the sake of this, though one day we might ask these questions too), and I would think there's a near-100% chance that it's also true (or let's say: it would be rational) that these engineers are in some ways using the feedback given by evaluators.

Again, I don't judge this human evaluation process here* (and I may compare it with Google's publicly stated judgments), just trying to dig down to it. This is not in any way about "they should not use human input", it is impossible (again, an AI aside) not to use human input, and even PageRank is based on human (links) input.

*"here" as in this post and thread.

Money.co.uk removed their article and posted a sorry message now:

<<We published an article on 9th May 2008, 13 Year Old Steals Dad's Credit Card to Buy Hookers. We posted the story as if it were real and did not disclose that the story was a hoax until 12 days later, and this was a mistake.

money.co.uk considers itself a trustworthy and reputable source of information on financial news and information. The contractors responsible for this mistake are no longer associated with money.co.uk in any way. We have also updated our policy at http://www.money.co.uk/about.htm which confirms that fake stories will be clearly labeled as such, and we have put measures in place to prevent a similar occurrence in the future.>>

Their about page now says, "Any hoax story we publish on money.co.uk, for entertainment value, will be clearly labeled as a fake story." Which makes you wonder if they intend to publish further hoax stories after this debacle.
http://www.money.co.uk/article/1000487-we-apologise-for-hoax-story.htm

Matt Cutts posted his response to this discussion:
http://www.mattcutts.com/blog/something-is-wrong-on-the-internet/

Philipp, thanks for the very detailed reply. I want to think a bit about it.

Anyway, what I'm thinking right now is: it's clear that there must be both a "human factor" and an algo. The question is: what should be their interaction? Should the algo simulate the human factor, i.e. you evaluate five searches and algo engineers deduce an algo for all the possible search? Or should the humans add that infinitesimal, but still fundamental, piece of opinion that algo can't have? Or...?