Hacker News Reader

(Disclaimer: I voted for Bernie Sanders in the Massachusetts primary).

I agree; the Hillary portion of it seemed like a bit of a silly conspiracy theory to me. They'd still need to provide evidence rather than innuendo, but it'd be more believable (albeit attract fewer HN upvotes) if the article had claimed Google has a pro-Bernie bias, since he's the candidate who likely has the most support amongst Google employees and especially engineers.

I don't think the (potential) issue is so much Google actively putting its thumb on the scale for one candidate, as the influence outside activists can have on how they create their algorithm.

Suppose that the (Hillary-endorsing) New York Times runs a front-page story saying that a study found Google search results systematically are more negative towards female candidates than they are to male candidates. The subtext of such a story would be that the NYT is pressuring Google to alter its algorithm to be more favorable to Hillary. I think there'd be a pretty decent chance Google would in fact alter its algorithm in response to such a story. I doubt they would alter their algorithm in response to a similar criticism from a conservative source.

Similarly, if one of the top results for Hillary Clinton or Bernie Sanders were a scatological reference, as with Rick Santorum, I think that Google would very quickly remove that page from its results.

Now, there's nothing wrong with getting rid of bias in search results directed at left-of-center politicians. But if Google (hypothetically, of course) is much more aggressive with eliminating bias towards left-of-center politicians than it is with eliminating bias towards right-of-center politicians, the net effect will be a general left-of-center bias in Google search results.

None of this is any kind of pre-planned conspiracy, just brain chemistry. We're wired to easily perceive bias, incivility and lying when directed at those we agree with, but justify, rationalize or ignore it when directed at those we disagree with.

It's worth pointing out that the positions of a company's executives and financial staff aren't necessarily the position of a company's line employees or engineers. Sanders is much less supportive of big business in government. And Google has put a LOT of money into government.

I think if Google execs had been directing Google engineers to make search results less favorable to Bernie and/or more favorable to Hillary, we would have heard about it.

The relevant engineers themselves would have more opportunity to change the search algorithm without generating public complaints from their coworkers. I really doubt they would, though, so conspiracy theorizing about it is still silly.

Like I said, I think the potential for political bias doesn't come from Google executive conspiracies or Google engineer conspiracies to alter the search algorithm, but rather from how the execs and/or engineers might react differently to outside demands that they change the algorithm, depending on the degree to which they sympathize with the source of the demands.

I'm just saying, the fact that Google would push politically something different than their lower level employees support wouldn't be surprising. Also, bear in mind that search ranking code is among Google's most carefully-guarded code. In fact, it's some of the only code not available company-wide.

The people who have access to Google's 'secret sauce' are likely the sort of die-hard "I bleed Google" people who would do anything in Google's best interests.

Of course, for that same reason, it's much less likely search tampering would happen... the group that could do it is very small and likely very closely monitored.

Also, the man has an ax to grind:

http://bits.blogs.nytimes.com/2012/01/05/one-mans-fight-with...

The article provides both scientifically acquired evidence and some analysis. That's more than your comment provides, IMO.

It provides evidence that search engine rankings influence opinion, which I do not doubt. From a business perspective SEO exists for this very reason.

It also provides evidence that Google as a company likely supports Hillary to at least some degree. Of this I also have no doubt.

But it uses FUD to say you should connect the dots and state that Google is actively manipulating rankings to assist a particular candidate. Of this, they provide no evidence other than "they could".

It's that last bit I take issue with. There's no analysis to support this notion.

If they did such an analysis I'd be very impressed as it would very much help discuss the deeper issues of Google's power over opinion. Essentially, Google is the portal to information for most people. They are a private company operating for profit that uses an algorithm that must be secret for it to work. I understand the unease with that situation and enjoy reading discussions about it.

However, their choice to essentially accuse Google of altering their algorithm to support Hillary is unsupported at this time. By adding it in it hurts their broader argument that gatekeepers of information should be subject to increased scrutiny and potentially to new regulations to ensure their neutrality.

Totally agree with you.

I'm also really surprised that the political leanings of corporate leaders is of such importance these days.

It's like holding every single employee of Enron culpable for the actions of Ken Lay et al.

"Hillary and Trump have received the lion's share of attention in media, social media, and such. Google searches SHOULD show them prominently."

I'm not sure I agree with this.

An alternative is easy to conceive, that all candidates that meet some external criteria (such as being on a ballot) are randomized to be favored equally in results.

I'm not convinced that it's inherently better to favor candidates with a lot of media coverage over just equally favoring them all. Sure, that's how it works now, but that doesn't mean it should be that way. There is a choice to make.

Google tries to show what the web and web users think is relevant to a search term. Not what is true, neutral, peer-reviewed, curated, balanced, etc.

It'd be interesting to see if Google could come up with a way to use machine learning to do fact finding and remove the web's biases. I strongly suspect such an effort would be seen as more dangerous than just reporting on the web.