Google vs. America

Four antitrust lawsuits have been filed against Google by state attorneys general, charging it with attempting and conspiring to monopolize interstate commerce. If successful, the states concerned will receive threefold the damages caused to their residents by Google's alleged monopolistic behavior.

Such a drastic action, while very unusual for the present time, is not unprecedented in Internet history. The breakup of the Bell System either resulted in or caused the growth of phreaking, depending on your viewpoint. The lawsuit against Microsoft, while its outcome was anticlimactic, did spell the beginning of the end of the domination of Internet Explorer.

It is important to note that because both cases were settled, they are not precedents in the legal sense. But even though they didn't set precedent (make rules future judges have to follow), they did have widespread social, economic, and political effects. These cases might fall into that category, as well.

They reflect a new understanding that Big Tech corporations are the colonizers, not the friends, of the common man of the Internet. Both sides of the political spectrum, despite their widening differences, have united in realizing that these corporations are responsible only to themselves and their shareholders and not to the people.

Overview of the cases

Four cases have been filed. The numbering is for convenience and is arbitrary.

• By the State of Utah, in relation to Android ("Utah-Android case" in headings, Case I in text)
• By the State of Colorado, in relation to searching ("Colorado-search case" in headings, Case II in text)
• By the Federal Government, in relation to defaults for mobile devices ("Federal-mobile case" in headings, Case III in text)
• By the State of Texas, in relation to advertising ("Texas-advert case" in headings, Case IV in text)

Numbers in parentheses refer to paragraph number of the complaint. As they have not been fully unredacted, this article can only be based on the information that is not redacted.

When reading those summaries, it is important to remember that a company does not violate the antitrust laws by securing or maintaining a monopoly because of the quality or attractiveness of its products. What is needed to prove a violation is maintaining a monopoly through anti-competitive conduct.

Utah-Android case

Filed in: Northern District of California (covering the coast of California down to about Monterey, and including the Bay Area)
Link to complaint: https://regmedia.co.uk/2021/07/08/antitrust.pdf
List of plaintiffs: Utah, New York, North Carolina, Tennessee, Arizona, Colorado, Iowa, Nebraska, Alaska, Arkansas, California, Connecticut, Delaware, District of Columbia, Florida, Idaho, Indiana, Kentucky, Maryland, Massachusetts, Minnesota, Mississippi, Missouri, Montana, Nevada, New Hampshire, New Jersey, New Mexico, North Dakota, Oklahoma, Oregon, Rhode Island, South Dakota, Virginia, Vermont, Washington, West Virginia

Colorado-search case

Filed in: District of Columbia (covering Washington DC)
Link to complaint: https://coag.gov/app/uploads/2020/12/Colorado-et-al.-v.-Google-PUBLIC-REDACTED-Complaint.pdf
List of plaintiffs: Colorado, Nebraska, Arizona, Iowa, New York, North Carolina, Tennessee, Utah, Alaska, Connecticut, Delaware, Hawaii, Idaho, Illinois, Kansas, Maine, Maryland, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, North Dakota, Ohio, Oklahoma, Oregon, Rhode Island, South Dakota, Vermont, Washington, West Virginia, Wyoming, Massachusetts, Pennsylvania, Puerto Rico, Virginia, Guam, District of Columbia

Federal-mobile case

Filed in: District of Columbia (covering guess where)
Link to complaint: https://www.justice.gov/opa/press-release/file/1328941/download
List of plaintiffs: United States, Arkansas, Florida, Georgia, Indiana, Kentucky, Louisiana, Mississippi, Missouri, Montana, South Carolina, Texas
This complaint charges Google with using "anti[-]competitive tactics to maintain and extend its monopolies in the markets for general search services, search advertising, and general search text advertising" (1). Specifically, Google has entered into exclusionary agreements with distributors of Internet-connected devices that make Google the default search engine for these devices in exchange for a cut in the revenue from advertising. Such agreements cover almost 60 percent of all searches in the United States, and almost half the rest "are funneled through properties owned and operated directly by Google." (112) For example, Google pays Apple about $10 billion so that Google remains the default search engine on iOS devices. This makes up 20 percent of Apple's income (118), and represents 36 percent of all searches in the entire United States (121).

While Android is technically open-source, Google requires manufacturers of smartphones to enter into certain agreements to use Google Play and Maps and have Google share its revenue with them (128). The first is the anti-forking agreement, whereby manufacturers cannot develop alternative Android operating systems. The second is the preinstallation agreement, whereby manufacturers must preinstall Google Play, Chrome, Google search, Gmail, Maps, and YouTube on every phone so that users cannot delete them (134). It also requires that the Google search widget be placed at a specific place on the screen (138). Also, manufacturers have to "implement a Google hotword" that automatically activates Google Assistant and "ensure certain touch actions on the device’s home button directly access Google Assistant or Google" (139). The third is the revenue sharing agreement, which Google enters into with carriers and manufacturers, whereby the carrier or manufacturer gets "a substantial portion of Google’s search advertising revenues" for making "Google the preset default general search engine for all significant search access points on the device." (144) These agreements last for 2 or 3 years. If it walks away, it "loses out on revenue share not only for new mobile devices but also for the phones and tablets previously sold and in the hands of consumers." (152) These exclusive agreements, the complaint alleges, deprive a rival of the chance to compete even if it had a better search engine.

Google has also been trying to expand its empire to newer forms of Internet-connected devices like smart speakers or smartwatches. For instance, Google does not allow smartwatch manufacturers to have another search engine preinstalled concurrently with Google if they want to use its "free" operating system, Wear OS (162).

Google also enters into revenue sharing agreements with browsers, like Safari and Firefox. Internet Explorer and Edge are the only major ones in the United States to have an alternative preset (158). The browsers get up to 40 percent of the advertising revenue generated by people searching with Google as the default (157).

Texas-advert case

Filed in: Southern District of New York (covering Manhattan, the Bronx, and the suburbs to their north)
Link to complaint: https://storage.courtlistener.com/recap/gov.uscourts.nysd.564903/gov.uscourts.nysd.564903.152.0_1.pdf
List of plaintiffs: Texas, Alaska, Arkansas, Florida, Idaho, Indiana, Louisiana, Mississippi, Missouri, Montana, Nevada, North Dakota, South Carolina, South Dakota, Utah, Kentucky, Puerto Rico
The complaint charges Google with monopolizing four markets: ad servers, ad exchanges, ad networks, and ad buying tools.

TL;DR - Use uBlock Origin.

The markets for online advertising

Online advertisers generally do not directly buy advertising space from the publishers of their ads. Instead, they do so through a series of intermediaries, each of which takes its cut of the advertiser's payment. The complaint accuses Google of monopolizing the market in each and every one of them. From publisher (and therefore reader) to advertiser, these are: ad servers, ad exchanges/networks (marketplaces), and ad buying tools.

An ad server "performs three internal critical tasks related to selling ad space." It first identifies each individual user visiting the publisher's website, and links it to other information about the user. (38) Second, it connects to multiple advertising marketplaces and lets "publishers automatically route their inventory into them for sale as the users load publishers’ webpage" and "controls how the different marketplaces can access and compete for a publisher’s inventory." (39) Third, it routes ads sold for a premium to "high-value users" who are more likely to make a purchase upon seeing them (40).

Marketplaces are of two kinds: exchanges for large publishers, networks for small ones (44). Exchanges are "real-time auction marketplaces that match multiple buyers and multiple sellers on an impression-by-impression basis", an "impression" being every time a unique user sees an ad (45). Ad exchanges are only open to advertisers with a minimum number of page views, and that minimum is usually in the millions (46). Like exchanges, ad networks "match publishers' inventory with their advertisers' demand"; however, they "obscure prices within auctions" so that neither buyer nor seller knows the rate at which they take a cut of the buyer's payment. They can do that because they buy the ad space from publishers then resell it to advertisers (52). There are separate networks for web display ads and mobile app ads (53).

Ad buying tools are the agents of advertisers, who interface with marketplaces on their behalf, just like how ad servers interface with marketplaces on the behalf of publishers. Ad buying tools for large advertisers are called demand-side platforms (57). Advertisers are required to spend a minimum amount of money to use one, typically in the five figures, because a demand-side platform offers advertisers more freedom to bid and trade and is therefore much more complex (59).

It's not practical for either advertisers or publishers to bypass this system. Every time a user visits a publisher's page, the publisher's ad server sends information about the user and the ad slot to a marketplace. The marketplace requests bids for that impression from ad buying tools. Ad buying tools have to process the request, determine the price they think the impression is worth, and respond to the bid request with that price. This all has to happen in a split second, before the page is even finished loading (60).

Google's monopolistic conduct

Ad buying tools and ad servers should loyally represent and serve the interests of their clients in the fast-paced world of online advertising. Ad exchanges should be neutral meeting places of ad buying tools and ad servers, and should be forced by keen competition not to take more than a reasonable share from every transaction made. Ad networks should act as middlemen, connecting buyers to sellers and charging for the introduction. Yet Google, by monopolizing all four markets, "is pitcher, batter, and umpire, all at the same time", taking advantage of the complexity of the market to excessively enrich itself (4).

Google's exchange, AdX, charges sellers "19 to 22 percent of...clearing prices" (48) and its network, Google Display Network or GDN, charges sellers "around 32 to 40 percent of each transaction" (54). It can afford to charge these prices not because it provides a superior service, but because it uses underhanded tactics.

When Google first launched AdX in 2009, it required small advertisers using Google's ad buying tool, Google Ads, to also use AdX, and required large publishers who wanted to receive bids from them to trade in AdX and use Google's ad server; it would "represent the buy-side, where it extracted one fee, as well as the sell-side, where it extracted a second fee," and the trade would take place in a market it represented, "where it extracted a third fee". (119) In addition, it required those who wanted to use GDN or advertise in Google Search to do so through Google Ads (121). Using more than one ad buying tool would result in extra cost for no discernible advantage for, as we have seen above, the operations of an ad buying tool cannot be parallelized. Once the trap was set, Google programmed Google Ads to route bids only to AdX, without regard to which exchange offered the lowest price, thus acting against the advertisers' best interest (124).

Google also used a practice the industry called waterfalling between 2009 and 2016. This forced "publishers to route their ad space to a single exchange, one at a time, rather than all at once." (132) Since a publisher could only route its ad space to a single exchange, it could not access an advertiser that bid a higher price for the same ad space, but on a different exchange (133). It also began a process called Dynamic Allocation in 2010. Under this, AdX would be entitled to send live bids to Google's ad server, but all other exchanges would only be able to send static bids based on an average of the prices it previously paid for similar impressions. Then, if it wanted, AdX could outbid the competing exchange "by paying just one penny more" (136). Google Ads won over 80 percent of AdX auctions and other competing exchanges could only scrounge for leftovers AdX deemed not worth it (139).

It also encrypted user IDs in its ad server and ad buying tools so publishers and advertisers could not compare IDs and figure out who was who, since the same person was known to the publisher and the advertiser by different IDs. Yet, internally, the hash or ciphertext of the ID was always the same (146). While Google claimed this was motivated by "privacy", no compunctions of privacy prevented Google from sharing user IDs with AdX and Google Ads (156), any more than it stopped them from secretly backing up WhatsApp communications to Google Drive while claiming WhatsApp was end-to-end encrypted (169). This information asymmetry enabled Google to use its inside information to create programs to increase its own profits and tell publishers and advertisers that the programs were for their benefit, since they couldn't know better (155).

Frustrated with the rigged game, the practice of "header bidding" was innovated by competitors in 2014. Header bidding involved putting a piece of code into the header of a webpage that would command the visitor's browser to directly request bids from multiple exchanges, then send the highest bid it received to the publisher, which would forward it to the Google ad server (186). By 2016, it had become the dominant practice for online advertising, and some publishers' ad revenue doubled pre-header bidding levels (188). Google desired to quash header bidding for three reasons: keep prices high, continue insider trading, and forestall competition (191).

Seemingly admitting defeat, Google introduced a new system called open bidding. While ostensibly permitting other exchanges to return bids, they still faced stacked odds on open bidding. First, they could not directly access the user's page (197). Second, publishers who sold ad space in them were punishable by a 5 to 10 percent penalty (198). Third, publishers continued to be forced to route their inventory only through AdX (199). Fourth, AdX was secretly given a special "prioritization" so it would beat even those bidding higher (200).

The bottom seemed about to fall out when in March 2017, Facebook announced its support for header bidding (204). The Big Tech giants began playing a game of chicken, each threatening to air the other's dirty laundry, but in September 2018 they reached a truce. Google would charge Facebook a lower 5 to 10 percent fee and let it bid directly into its ad server for publishers in exchange for not getting involved in header bidding and shushing up about the deal (219). It then lied about it to prospective customers, claiming that all buyers had an equal shot when it obviously knew that wasn't true (233).

Signing a corrupt bargain with its biggest rival was not the end of it. It tried no less than ten ways to kill header bidding. The end result of this action would be to establish a walled garden - "a closed ecosystem out of the otherwise-open internet" (260).

The same dirty tactics used by Google also trickle down to the ad buying tools market. Advertisers have no choice but to use Google Ads to reach the majority of American consumers efficiently.