« Randy Picker on Google's "Don't Be Evil" Motto | Main | The Iraq War – A Humanitarian Success? »

November 23, 2008


TrackBack URL for this entry:

Listed below are links to weblogs that reference More Google and Evil:


Feed You can follow this conversation by subscribing to the comment feed for this post.

Uzair Kayani

Professor Picker offers a brilliant critique of a company that I love. I understand the three-pronged critique as follows.

Google is more or less a monopolist in the search engine arena. As such, we are worried that it will extract monopoly rents from its main customers- advertisers that buy keywords or buy Google ads that are displayed on other sites. It will charge advertisers more for showing their ads than it would have if other competing search engines could offer those advertisers a cheaper price.

These advertisers buy ads on Google because they expect to attract customers to their advertised products (say a research paper or a car). The ads typically show up on the right-hand side when you run a Google search.

Monopolies like Google can extract exorbitant prices from their customers. However, Google-the-monopolist says that it doesn't use its monopoly power, since it sells all its ads via auctions.

Loosely, an auction market is supposed to be a perfect free market. The monopolist auctioneer has no opportunity to set prices. The bidders at the auction compete with each other in how much they are willing to pay. That is, the customers (bidders, here Google-ad buyers) rather than the producers (auctioneers, here Google) decide the price. Monopoly problem solved.

Here are the problems with the happy auction story:

(1) Google sets minimum bids ("reserve prices" in auction talk) so that people who bid on particular keywords still don't get the those keywords after winning the auction. That is, Google doesn't reward the winning bidder if the winning bid is "too low."

(2) Google limits the quantity of ads that show up for any search. It controls the number of "slots." For example, if 15 people won 15 different auctions, and every keyword they bid on appears on a specific search query, you would expect all of them to show up with the search results. Every advertiser would get what s/he paid for. Instead, only a subset of the keyword ads show. Perhaps the bidder doesn't get her money's worth.

(3) Google forces two other entities to bundle its services with theirs: ASK and AOL are the victims (or more accurately, their consumers are). Here, the theory is that bundling allows Google to sell its services in areas where it would otherwise be uncompetitive. The customer who likes AOL but hates Google may still end up using Google services, because he likes AOL + Google more than any alternative.

I don't understand auction theory well, but I suspect that setting reserve prices (criticism # 1) could help reduce the opportunities for other bidders to collude against or intimidate a bidder that would otherwise win. It would guarantee incremental bidding. If the reserve price were unknown and random, then the rightful winner wouldn't care, but colluders and intimidators might be deterred, because they wouldn't know where their chicken game might land them. Below the reserve price they would pay nothing, but above it, they would pay a whole lot. And since they don't know the reserve price, they will suspect that Google keeps changing that price arbitrarily. Thus, any potential colluders or intimidators will be forced to bid incrementally rather than aggressively. Incremental bidding (enforced by the random reserve price) might ensure that other bidders are not scared off.

Second, limiting the number of slots (criticism # 2) could increase the value of every individual ad to the advertisers, because a glut of ads might make each individual ad relatively worthless. A random display of ads (where the ad you paid for shows up only sometimes) may help all advertisers on net, because it avoids the glut problem that would make all ads (and Google itself) less attractive. Put differently, unlimited ads could create a tragedy of the commons. Limited (but randomly assigned) slots might avoid this problem.

Bundling (criticism # 3) also appears to be a serious charge, but I suppose if there is no independent market for AOL and ASK services (that is, a market apart from their search services where they compete with others, say in e-mail or social networks) then we don't have a problem. If no one would buy their e-mail or other services without their search services, then we are not hurting them by bundling those unmarketable services with Google.

These are all highly speculative points but I was fascinated by the discussion.

Uzair Kayani

Oops. A minimum bid (the least you must bid) is not usually the same as a reserve price (the least you must pay to win), though I guess it could be. I need to get an auction theory book.

The comments to this entry are closed.