This whole rant began from frustration at Churner Prize, a blog on the complete imbecility of newspaper journalists to think for themselves. We would like to subscribe to their RSS feed, but find ourselves unable to, for all feeds are piped through Feedbugger, and this gives more information to Feedbugger's owners G****e than we care to disclose (an amount not unadjacent to none at all). There is no back door to escape the information behemoth. The Churner Prize folk criticise unquestioning acceptance of perceived wisdom amongst others, yet unquestioningly accept the perceived wisdom regarding G****e.
The same frustration to UK Polling Report and the Economist's Certain Ideas of Europe, two blogs that are now off our reading list precisely because their owners have ceased native RSS feeds and tried to divert to Feedbugger. Forcing everyone to go via G****e introduces an obstacle that does nothing to help your readers, and does turn a small number of readers away. That number may be one: it is not zero.
Which brings us to a case that may go before the courts, Christian Institute -v- G****e. Ministry of Truth (a blog that handles its own RSS feeds) reports that the godbothererers might sue the advertising behemoth for rejecting their proposed commercial on the grounds that it's at the intersection of abortion and religion. MoT suggests that the CI's advert has been declined because G****e is taking a stand on the factual content of CI's link, suggesting that in its opinion, the Christian Institute promotes factually inaccurate information about abortion - there is nothing in law to prohibit discrimination on the grounds of bad science..
That's not our reading of the exchanges: the commercial is clearly about abortion, it is clearly by a religious group. And it has been rejected on the grounds that it is a religious group talking about abortion. As we understand it, G****e's statement implies that if the religious group were not talking about abortion, the advert would be fine. Or if the abortion were being discussed without religion, it would be fine. That does appear to be prima facie discrimination on the grounds of religion. No matter how wrong the Christian Institute is, the content of the website wasn't mentioned by the advert brokerage, and doesn't enter into the debate here.
Like MoT, we hope that this case does come to court, because a precedent and some case law will be very useful, and our first preference will be for a case that both sides lose.
What's wrong with G****e
All of this raises the question: what is our argument against G****e? Let us not blame the company for the failure of society, but give blame where its actions - or inactions - have led to harm. In this discussion, we focus exclusively on decisions that G****e has made, and that a competitor company need not have made, and we aim to knock at the perceived wisdom amongst geeks in the English-speaking geekery.
First, we address the canard that G****e is providing a useful service, for we do not believe this to be the case. G****e began in 1996 with a new algorithm to determine search results: rather than a naive count of the number of times a given term appeared on a page, it weighted results according to the number of in-links a page received, themselves weighted by the number of in-links received, themselves weighted in the same way, and so on ad infinitum. At the time of its derivation, the consensus was that the results were no worse than a naive count, and tended to be better. However, the developers failed to spot that this would lead to a huge growth in link clusters, sub-networks that have a tremendous number of links within themselves, few if any links out, and relatively few in. These are intended to distort the search algorithm, and generally succeed in doing so. G****e's search results have become increasingly poor as a direct consequence, and have long since failed to be usable on their own.
G****e introduced advertising in 2000. We will note, but not pursue, our philosophical objection to advertising as an intentional diversion from the task at hand. The concept behind G****e's adverts was nothing new: bidding on particular keywords had previously been implemented by Goto. The crucial difference with Goto's model was that G****e gave a small fraction of its advertising revenue to the website hosting the commercials. When this combined with the easily-compromised search results, we find that almost any search gives many links to pages that exist only to sell G****e commercials. G****e appears to do nothing to stop this, because it increases their revenue.
G****e claims that it extracts the key concepts from pages, and uses those as keywords for advertising. On occasion, this can deliver highly relevant commercials. More often, this brings out wholly unmemorable adverts, and occasionally delivers entirely objectionable text. It appears that the content extraction algorithm has not been developed.
So far, we've discussed the rise and fall of G****e's search algorithm, and the way that the company has enabled a new generation of spam sites selling its advertising. But G****e has extended its reach beyond web search and advertising brokerage. Its first significant acquisition was Dejanews, an archive of Usenet postings from 1995. While the preservation of the archive could reasonably be seen a social good, the subsequent conflation of Usenet and closed discussion groups has led to a diminution of Usenet, and a flood of spam messages pointing to sites containing G****e commercials. G****e appears to do nothing to stop this, because it increases their revenue. Even though Usenet is not the web, it has been directly polluted by G****e's decisions.
The company has made many other purchases. These fall into three categories: advertising and support services (Doubleclick, Adscape, anti-spam software); venues for advertising (Blogger, You Tube, mapping software); and methods to track users (Feedburner, Urchin analysis, Genius). We have no particular objection to an advertising service buying support services. We have no particular objection to an advertising service buying its own billboards. It's the last category that we find most troublesome, because of G****e's approach to privacy.
That lack-of-privacy policy in full
From its actions, G****e has made clear its belief that it is entitled to every piece of information in the world. It maintains a history of every search from a given IP, and cross-checks that against a long-lasting cookie. Originally, these cookies had an expiry date in 2038; this was changed circa 2006 so that the cookie would expire after two years of inactivity, and some still see this as too long. More importantly, the company still holds complete records of searches. G****e's published privacy policy allows it to cross-reference information it gathers from other products, whether owned by G****e or someone else. This is done in order to present adverts that its users are more likely to follow, an action that boosts G****e's revenue.
Being generous people, we give G****e the benefit of the doubt in this area: while we do not expect G****e to abuse the information it gathers, we do not trust them not to do so. No such benefit can be given to the corrupt political regime in the area where G****e has chosen to locate itself, for it has already engaged in one fishing expedition to retrieve all search terms for a week. Having all this information in such an alien jurisdiction presents an unacceptable risk to our privacy.
Furthermore, G****e believes that it is entitled to sell advertising against everything possible. We do not give G****e (or anyone else) permission to process our emails so that it can sell adverts against them, yet it still does so. Indeed, we specifically include a signature in our emails as saying that the automated processing of emails to influence the selection of advertising is not allowed. Not that this makes a blind bit of difference; adverts are still served, and the email is still retained forever.
The final criticism, and one that is less rational yet more pervasive than the above, is that we find the company exudes arrogance in every communication. It claims not to be evil, yet it has chosen to imbue its structure with actions that we deem to be (at best) morally ambiguous. Not allowing people to insist that information about them is deleted is simply rude; collusion with the illegitimate and anti-democratic regimes that purport to rule over California creates the evil of failure to stand up for that which is right. But in particular, we believe that creating such huge databanks on people is a profoundly evil project; it purports to know more about the person than the person knows about themself. G****e casts itself as an omnipresent deity, yet it is made by humans, and that presents us with a theological contradiction that we cannot resolve in the company's favour.
For all these reasons, we choose not to engage with G****e at all. We choose not to list ourselves in G****e. We choose not to see advertising brokered by G****e. We do not send messages that we know will be intercepted by G****e. We do not wish to contribute to G****e's bank of data about us, for we must assume that it will be tapped into by nefarious elements. Knowledge is power, and we wish to keep some power over the corporate behemoth.
What would it take for G****e to win our trust? At the very least, marked improvements in its privacy policy, a physical and cultural move away from the Untied States, a shift to the effect that it is the servant - and not the master - of the users. We do not expect this to happen overnight.
Until that happens, we are unable to follow RSS feeds from those who insist on using Feedbu**er. And we shall publish notices informing people of why we do not read them.
A follow-up post was published on 10 May.
