Few of us remember enough math to be able to precisely define "algorithm" -- but it has become increasingly clear that these little "sets of rules that precisely define a sequence of
operations" will have much to do with what media and advertising we will consume as it all becomes available across various digital platforms.
Algorithms that trawl social networks and
databases to get a beat on user interests through search patterns and other behavioral trends are being used by various "content farms" including AOL (with its 700 journalists who must wonder why
their editorial judgment can't be trusted) to predict the success of their content by dynamically monitoring user feedback and response to other stories. Last week, Google's famously secret algorithms
were adjusted in response to the kind of content generated by companies that use their own algorithms to decide what kind of content people want.
Most every online advertising company boasts that
it has secret algorithms that can predict which ads people will find the most relevant. How many thousands of online ads have run in the wrong place at the wrong time to the wrong people based on
algorithms that clearly failed? I see a couple nearly every day. And while some of them are wildly amusing, I am sure the clients featured in the ads are none too amused.
Nearly everything
you read or watch online is being number-crunched to predict what you might like to read or watch in the future. The other day, Cisco Systems released a survey that predicted TV networks will be dead in 20 years, replaced with a system of computerized content that will be
molded to your preferences. Yes, that would be algorithms deciding that since you like "Spartacus," "Californiction" and "Skins," you also like this assortment of new shows with naked women, liberal
use of the f-word and drugs. (And who knows, you might).
But can algorithms always be trusted to make the "right" decision as they mull though millions -- if not billions -- of calculations
per second?
When IBM's Watson was confronted with the answer "Its largest airport is named for a World War II hero; its second largest, for a World War II battle" on "Jeopardy," it issued
the question "What is Toronto?" ("What is Chicago?" is the right answer). Earlier this week, U.S. Rep. Rush Holt (New Jersey) a five-time champion during the trivia show's original run 35 years
ago, a former State Department arms control expert and ex-leader of the federal Princeton Plasma Physics Laboratory, actually beat Watson at a "Jeopardy"-like match held at a Washington hotel.
Meanwhile, Demand Media, the best-known content farm (thanks largely to its recent IPO) claims that the change in Google's ranking algorithms has had negligible impact on where its page show up on the
monopolistic search engine's rankings.
As a frequent Googler, while I can hardly complain about its attempt to rank better content higher up in natural search results, I am not certain -- as
Watson's failings have shown -- that semantic machine reading of content is really ready for prime time. Some companies that curate and help writers publish their work are being heavily penalized by
the shift at Google because their output is prolific, not because their content is poor quality. Other companies that provide short, to-the-point answers are also being hurt with less traffic because
their pages are not artificially inflated with meaningless back links.
With millions of dollars of potential commerce at stake, an entire black-ops industry has grown up around trying to game the
Google system in order to produce higher page rankings that pull searchers to your site rather than the next guy's. JCPenney was just the one that got publicly flogged -- but nearly everybody else
does it, too. And that is the shame of it. While companies that produce legitimate content get hurt, the bad boys who work hard at outmaneuvering Google's system come out winners. I assure you it will
only be a matter of weeks before they deconstruct the latest Google algorithms and adjust their own playbooks to creep back up the results pages.
At least Watson is supposed to be learning from
its mistakes and reconfiguring his little silicon soul to improve so that when it is put into full play by, say, the defense department, it sends missiles to take out Baharestan or Tehran -- and not
Portland or Orlando. Not sure I am so confident about Google.