Geekynomics? Finding The Hidden Government Within Google’s Magic Money Machine

Joseph Heath, a philosophy professor at the University of Toronto, has published the timely Filthy Lucre: Economics for People Who Hate Capitalism (HarperCollins, 2009), a clever romp through formal economic theory. As even-handed as any intellectual inheritor of The Simpsons sensibility, it debunks six economic myths of the right, and six of the left. Heath writes like a Gen-X’er, coolly interjecting phrases like “Holy crap!” into formal academic commentary.

Logical leaps being my specialty, let’s jump right into paid search.

Heath’s book happens to provide brilliant tools for grasping the amazing efficiency of the Google AdWords keyword auction. But it also helps us push past platitudes about its “invisible hand” workings, to reach a better understanding of why and how the “free market” auction must be regulated.

Paid search pricing revisited

Aside from the problem of herding searchers into searching directly at a search engine made up solely of paid links as GoTo did in its 1998 launch, the click auction was a brilliant step forward in terms of efficiently pricing digital advertising. The core of it was that the highest bidder on a keyword ranked highest on the page. Bid too little, and no one sees your ad.

There was now no fixed pricing or fixed number of impressions or clicks you could force the media company to sell you by placing an ad “buy”. The pricing was set by the level of competition, and the volume dependent on users clicking. In the long term, the model led to windfall profits for Overture (now Yahoo! Search Marketing) and Google, not only due to bidding wars for high-priced clicks, but due to the marketplace’s efficiency in monetizing lower-priced inventory.

But there would be many wrinkles along the way.

The failure of fixed pricing

Google actually took a step backwards in late 2000 when it introduced the first version of AdWords, with fixed rates based on a CPM model. One problem discovered by advertisers unfamiliar with the young medium was that prices could be sky-high, especially on a per-click basis. It was back to the drawing board. Google would later bone up on the concept of a Vickrey auction, significantly altering (but borrowing substantially from) GoTo/Overture’s invention. They rolled out the “real” version of AdWords in 2002. Where keyword competition was light, advertisers found bargains.

As proof mounted that the auction model was more efficient, a certain reverence developed for it. And so into the discourse crept some of the excessive reverence for the abstract benefits of the free market often seen in the field of economics: people worshipping classroom concepts while failing to take note of their evident conflicts with not only empirical reality, but often, logic itself.

Many buyers, one market maker, one primary seller

The fact that there are all these moving parts, freedom to bid, and something resembling a high number of competitors, if breathed in by someone who inhaled a bit too much during Econ 101, might smell a little like the never-existing “perfect competition” often used as a precept by economics professors.

The search engine industry, of course, was far closer to “perfect competition” at a certain point in the 1990’s, with six or seven viable contenders for searcher eyeballs. But in the current Google-centric world, it’s indisputable that many buyers are buying media through primary market maker (platform builder). Now, nearly 50% of the product sold by that market maker is in fact the market maker’s own media. So for half of Google’s revenues, the transaction is all about “many buyers, and one seller.”

In such a scenario, you can use either classic economics or simple logic to tell you that the seller has an incentive to take away “freedoms” in the auction such as making it easy for some buyers to pick up inventory (less popular keywords) at rock bottom prices.

As it turns out, Google has a double incentive to crank up prices. Searchers like white space. On purely informational queries, searcher satisfaction goes up when the page is less cluttered with commercial messages. Google likes to either exact a decent effective CPM rate, or zero in exchange for serving no ads.

Thus far, we’ve only talked about how Google stacks up against some classical economic theory in the economic sense – roughly, as a player that behaves as you might expect a monopolist would.

But Google behaves as both a powerful player of the game, and (in a sense) its regulator. In a number of areas, discovered as the auction ran like a massive simulation game over the years, Google now intervenes in a manner similar to a government. That’s not necessarily a bad thing. Indeed, it is utterly unavoidable.

“Night watchman” at least

As Heath points out, even what are today called “conservative” economists (sometimes called libertarians) are forced by logical imperative to admit some role for government. The logic works this way: if you walk through all of the commercial transactions taking place in today’s complex capitalist economy, and tote up the cost just of enforcing contracts, avoiding obvious harm, and the like, this transactional machinery already constitutes a significant component of GDP.

In the click auction, the libertarian model broke down early on, in areas fairly basic to the conduct of a fair process. The fact that a “click” could be generated by a malicious person or a bot, first off, necessitated strong “state” (auctioneer) intervention.

Darwinism and maladaptive effects

Some of the areas a Google might consider intervening in are relatively harmless, but interesting nonetheless. Let’s focus on how common it is for seemingly self-organizing systems to break down.

“Invisible hand” thinking sometimes blithely assumes that evolutionary forces always turn out to help organisms (or social organizations) to successfully adapt.

But right in the Darwinian literature are basic examples of “maladaptive effects,” Heath writes. Taken from the standpoint of a species, an animal might develop characteristics that help it prosper or thrive in finding a mate. But as that characteristic is replicated and exaggerated, the whole species might suffer and eventually die off.

A peacock’s tail feathers are a relatively innocuous example of this. The Irish elk and its “grotesquely impractical antlers spanning 13 feet” would be a better example of a flashy trait leading to species-wide doom. “Individuals with such traits are able to survive,” Heath writes, “only because the structure of the trait is such that as they go down, they are able to take every other member of the species down with them.”

The AdWords auction is rife with such mechanisms. Under AdWords (and its contemporaries, Yahoo Search Marketing and Microsoft adCenter), there are a host of things advertisers could try to do with their ads to stand out. The worst forms of showboating might include making false statements, excessive use of caps and punctuation, and so on. Google began policing many of these things from the outset of the ad program in 2002. Google’s introduction of Landing Page and Website Quality guidelines in late 2005 was simply a continuation of this role, but the fact that the enforcement was expressed coolly in the form of a numerical score was a significant twist.

Quality Score: cyber-guvernment?

Google is not a real government – hence the coined term “guvernment” (that just happens to coincide with the name of Toronto nightclub, The Guvernment).

If it’s so obvious that Google can and must regulate the ad system, what’s the big deal?

Well, it’s less and less obvious to many advertisers, because this regulation is not being called that. By burying many policies within a numerical score that is applied inside accounts, Google behaves like a regulator vis-à-vis the auction, minus the accompanying degree of transparency and due process that we typically associate with that role.

Rank on the page, and thus your advertising economics, are determined by a Quality Score, assigned by keyword. The fact that increasingly baffling theories are circulated in the industry about how Quality Score works is due in part to the usual snake oil approach of some would-be experts, in part due to poor reading comprehension, but also in part due to Google’s opaque formula and administration of the formula.

Calls for transparency were heard loudly from the advertiser community throughout the rollout of Quality-Based bidding, accelerating Google’s response on several fronts. Google has published extensive sets of rules and guidelines. It has upped the precision of keyword quality score to a published ten-point scale. And, a diagnostic page informs the advertiser about whether keyword relevancy or landing page quality may be to blame for a low score on a given keyword. These measures fall short of the disclosure that some advertisers seek, however.

There was no perfect way to do this

For whatever reason, Google often uses technical language to describe their regulatory role in Quality-Based Bidding. Yet if advertisers are being regulated in some fashion with each and every action they take in the system, it’s only natural that they would want that function described in the traditional way (we’re applying rules to this market system we’ve built) as opposed to the current language (the system as a whole rests on global concepts like prices and quality, and all are expressed with numbers).

Continued advertiser calls for transparency and due process seem inevitable. They would be inevitable with any large system. Google has in the millions of advertisers, and many times more users. No matter how they build it or express it, the rulebook is going to seem strange or confusing to somebody. Hiding it in a scoring mechanism produces different flavors of puzzlement and controversy, is all.

Government OK, monopoly bad?

When you look carefully, Google’s role as a potentially monopolistic “single seller” is more threatening to advertisers’ bottom lines than its role as (often hidden) “guvernor” of the auction. Don’t be swayed by anyone claiming that these kinds of systems can work responsibly and efficiently without regulation. They can’t. It’s just extra puzzling because the terminology (“your keyword isn’t showing because of your ‘website quality’”; you’ve probably broken a rule but we can’t tell you which one) is otherworldly.

In future columns, I’ll focus back on the real threat to your bottom line – Google’s power to exact high click prices, which will inevitably return in an economic recovery – and what, if anything, you can do about it.