Oxford SEO Studies Blog

Much is being made recently of the inherent unreliability of Semantically structured data, due to the conscious and unconscious biases of the writers of algorithms and meta format markup.  Mary Hodder, at TechCrunch just yesterday asserts that the inherent bias of those opening up the  semantic significance  and resonance of web published data are in fact merely asserting their own conscious and unconscious politics.  The dramatic example that is coursing through the blogosphere at present is of course Amazon’s blanket censorship of gay  literature  and nonfiction on their ranking scale by associating a semantic connection between books that describe or endorse a gay lifestyle or preference as defacto “adult” and therefore unsuitable for certain audiences.  Overnight authors ranging from Gore Vidal though E.M. Forrester (”Only connect!”) have found themselves demoted on Amazon’s book ranking due to a calculated conspiracy of both religious and irreligious bigoted prudes to target so called “Gay Literature” on Amazon by coordinating mass bad  reviews of anyone they deem of unworthy of admission to heaven.  I imagine Alan Turing’s work on computer science doesn’t rate very high any longer either.

Setting aside the blaring logical flaws in homophobic bigotry: (Pioneering psychoanalyst Henry Stack Sullivan’s early 20th century adage suffices to plug that particular pustule in one sentence: “There are no homosexuals or heterosexuals; there are only homosexual and heterosexual acts“.  While we’re at it, let’s not ignore Sigmund’s {perhaps erroneous, given more current evidence}, observation that like chimpanzees, homo sapians are equally disposed to attraction to either sex; that we are in fact bi-sexual and it is social conditioning that reinforces one sexual preference over another). 

Amazon’s response to the 10,000+ signatory petition that was sent to them by the remaining marginally sane members of the online community didn’t help matters.  Instead of acknowledging the architectural defect in their very human association of ‘Gay’ with ‘Adult’  they blamed their search algorithm and the French technical engineer who made the association.  When in doubt, blame the machines or the French, or French machines.   This, of course inspired a flurry of Orwellian panic attacks amongst the digirati , who are  united in their depreciation of anything that includes the term “Semantic”.  Who started pointing out the inherent fascism of trusting machines to do any of our rudimentary judgment calls for us.  Even associative connections should not be relegated to machines because they will learn how to appreciate poetry, take over the world and enslave us all.  Although the image of a Terminator-type robot lounging on the banks of Oxford’s Isis river reading Keats on a spring day is worth contemplating, the fact remains that in the case of the Amazon boff-up, it was a very human error, not a machine one or even a result of entrusting machines (algorithms), with too much of our thinking.

Regardless of the allocation of blame, Mary takes this opportunity to assert a stark warning:

“The ethical issue with algorithms and information systems generally is that they make choices about what information to use, or display or hide, and this makes them very powerful. These choices are never made in a vacuum and reflect both the conscious and subconscious assumptions and ideas of their creators.”

Absolutely!  Freud as a code writer.  What ontology does his cigar indicate?

“The ethics bar in creating algorithms and classification systems should be very high. In fact I would suggest that companies with power in the marketplace, both commercial and ideas, should consider outside review of these systems’ assumptions and points of view so the results are fair.”

Again agreed, which is why at Oxfordseo.com, we are grateful for the continuing  determination of the w3c collective to promote standards of markup excellence (both Semantic and otherwise), that more accurately underlines the relevance of  data in the context of the searchers’ intent and desires.  Semantic has always been in its essence, a very human response to a very human need: the goal of accessing more relevant information in a more efficient manner. Or at least better than Google does for us now.  This whole debate revolving around the fear of robots or the inherent advantages of socially generated  “human” data is irrelevant.  What we want are better tools; always.  And that means better tools for navigating the haystacks of  data on the net.

What social networking represents is an attempt to present consensual information (or “truth”), as a substitute for authoritative or accredited information. Obviously, there’s an overlap in these gross definitions, but we believe that there’s a an interesting social power struggle afoot  that underlies the reliance people voluntarily place on their sources of information, or what they consider to be true. The Real Time reporting of on the spot eye witnesses that can relay one to many accounts of real time happening events (usually tragedies!), long before a news camera crew with a predisposed “angle” on what the story is, can make a strong case for the democratisation of information gathering and relaying.

However, we  make the case that in as much as there is inherent bias (both conscious and unconscious) within any self-proclaimed authoritative account or analysis, that those biases are usually declared at the point of entry. Undeclared biases are easily caught by the vigilance of the equivalent of custom officials who have a vested interest in vetting the credentials and credibilities of authoritative accounts. In that sense there is a social dimension to that level of constant scrutiny by a “community of peers”.

This characterises precisely the history of scientific methodology (paradigm theory), as well as that of modern journalism. Both are professions which are policed by the vulture like tendency of their practioners to eat their own; or at least leave the limping and the weak behind to die, somewhere outside attention span while filling the gaps with more accurate and up to date information.

In contrast, information that is derived from the consensus of social networks is at best like getting detailed instructions from a mob. Anyone who has ever taken a road trip in Ireland and gotten lost will concur with this unique and frustrating illustration of human behaviour  revealed when one asks a stranger for directions. If you ever ask someone for directions in the Irish countryside, they will cheerfully and politely give them to you. Sometimes elaborately, sometimes complexly, sometimes abbreviated with anecdote and personal qualification, but always comprehensively and at length. The problem arises when you actually try and follow those very same directions,  because they are seldom accurate. It’s not that the Irish as a race have no sense of direction. Rather, it is the custom of the provincial Irish never to admit that they don’t know where somewhere is. It’s considered downright rude not to offer a stranger some degree of orientation even if it is entirely inaccurate and bears no relationship to the territory.

This to us, precisely demonstrates the axiom “but can you count on your “Friends” to give you reliable information”?   Everyone loves having an opinion almost as much as they love sharing it. But a mass sharing of opinions is not the same as an analysis or accountable report on an event or a phenomenon. The latter actually takes time to reflect and to harness critical faculties, which of course are an obstacle to being the first to Tweat a happening event.

To us the accuracy machismo of Real Time reporting is its own worst enemy.
We still read newspapers and scientific journals (in spite of the heralding of their demise), because we entrust a greater degree of accuracy to those who have made it their profession to account for and recount the truth of events. Never without bias or slant, obviously but with a focus on trying to reveal the truth rather than just being the first person to have an opinion on it.  Amazon had a bais and instead of blaming the French or their algorithms they should accept the culablity of their own inherent baises, as we all do.

Going back to Mary Hodder’s word of warning about overlooking the unconscious bias inherent in meta format mark (or indeed the constructing of any algorithmic-based architecture), yes by all means let’s expose the biases we all maintain and correct and refine them while allocating  more authrity and resources to orgnaisations like w3c in maintaining  a policing of those standards. But let’s not substitute our critical faculties for the consensus for of the herd.  To misquote Nietzsche, let’s not “rise above the herd”, instead, let’s take the herd with us.

“It is easy to see the truth, just rid your self of your opinions”. —Haikuin

Hey! It's Just My Cigar, OK? It don't mean a thing.

One Response to “Sigmund Freud and the Semantic Unconscious”