What Would a Government-Operated Search Engine Look Like in the US?
By Eric Goldman
[Today's WSJ announcement that the FTC plans to subpoena Google reminds us that search engine bias issues are heating up rather than winding down. I recently posted my article recapping the last 5 years of developments on the topic.
Last week, I attended a conference at George Mason Law School on the topic, and I will post a comprehensive recap of the event shortly. I was scheduled for the last panel of the day, and it was clear that anything I said from my recent article or my 2006 article was likely to be preempted by discussion throughout the day. Therefore, I came up with a topic that was unpreemptable, but I could do this only by setting up a strawman from fantasyland. My talk notes:]
Publicly-funded search engines are sometimes floated as a way to address search engine bias. Indeed, Europe attempted to develop its own publicly-funded search engine, Quaero, although that effort has apparently collapsed.
This led me to a thought exercise: what if US government tried to build its own search engine? (We could call this hypothetical engine “The Googlement”). My talk explores a pure case of a government-operated engine. Other models, such as subsidies to a private company, might avoid some of the pitfalls but also might produce fewer of the benefits.
The Case for a Government-Operated Search Engine
Some arguments in favor of a government-operated search engine:
1) it could advance national objectives, such as prioritizing nation-specific content. This particular rationale seems hollow given the US-centric bias of existing search engines.
2) more competition is good. The marketplace relies on the invisible hand. Another well-funded competitor in the market could help discipline other market players.
3) search engines create positive externalities. We might decide this infrastructural component is too important to be left to market forces. Otherwise, due to the positive externalities, the marketplace may undersupply it.
4) the search engine will be subject to public oversight, which can ensure transparency, due process and accountability to the people.
The Case Against a Government-Operated Search Engine
On the other hand:
1) in response to the positive externality point, we don’t have good evidence of any market failure, or at least any undersupply.
2) operating a competitive search engine is expensive. Indeed, Yahoo offloaded a couple hundred million dollars of costs to Microsoft as part of their alliance. Also consider some of the second-order impacts of undertaking a major public infrastructure project like a search engine:
a) like most government works projects, the search engine would be under constant pressure to reduce costs or find new revenue sources. For example, PACER charges us to access public documents. And if the government defrayed costs through advertising, would it encounter the same church-state problems we fear with private actors?
b) the government could have difficulty recruiting best engineering talent. PACER’s 1980s state-of-the-art technology is a good example.
c) providing due process for complaints (discussed below) would be expensive. For example, Saferproducts.gov has developed complex vetting mechanisms before publishing consumer complaints.
3) consumers may fear giving their personal information to Big Brother more than they fear Little Brother. For example, in Gonzales v. Google, Google fought back against ridiculous government request for data. If the government operated the search engine itself, it could have just arrogated the data for its own purposes without even Google’s (thin) buffer.
4) who would make the editorial decisions about relevancy? I couldn’t even parameterize this problem. Would it be done by committee? Following guidelines, or just using their own judgment? Frankly, I couldn’t think of many examples of government-operated editorial publications where similar editorial judgments are made. A related question: who would decide the definition of spam?
5) decisions to downgrade, de-index or never index sites may be subject to First Amendment scrutiny.
For example, Christ’s Bride Ministries, Inc. v. Southeastern Pennsylvania Transportation Authority, 148 F.3d 242 (3d Cir. 1998) held that government bus ads are a limited public forum. In White Buffalo v. UT Austin (5th Cir. 2005), the court held that UT Austin’s anti-spam filter were OK under First Amendment because they regulated high-volume commercial email. If the judgments were about other types of communications, would the same result apply?
Numerous plaintiffs against search engines have unsuccessfully tried First Amendment complaints—as sword, not shield—against private actors. Ex: Langdon, KinderStart, Murawski v. Pataki. The many terminated user account cases, such as Estavillo v. Sony and Young v. Facebook are also analogous. These would become tenable lawsuits against a government-operated search engine.
6) We get the purported benefits only if consumers actually use the government-operated search engine. However, if the search engine fails on relevancy or its anti-spam efforts, consumers simply won’t use it. Given the significant challenges, it would be tough for the search engine to win consumers from market-driven alternatives.
Even though it’s a trip to fantasy-land, the thought exercise reinforces the merits of market-driven search engines:
* market forces will drive their quest for relevancy
* the marketplace will bear the costs rather than the public fisc
* the government doesn’t have direct access to consumer data
* for marketplace actors, the First Amendment acts as a shield for relevancy/anti-spam determinations, rather than a sword against the government