Google's Monopoly: Reduced Competition Reduces Search Quality And The Options Available To Consumers

C. Reduced Competition Reduces Search Quality And The Options Available To Consumers In General Search Services

1126. Absent Google’s exclusionary contracts, consumers likely would have had significantly better and more varied search services. Tr. 5858:23–5859:1 (Whinston (Pls. Expert)).

1127. Reduced competition in the U.S. general search services market reduces the quality and the options that consumers have, which means consumers are likely to be worse off. Tr. 5854:11–20 (Whinston (Pls. Expert)) (“[T]here’s harm because quality is super important for consumers and if competition or lack of competition reduces the quality and the options that consumers have, consumers are going to be highly likely to be worse off.”).Lack of competition in the U.S. general search market means quality changes (or lack thereof) have little impact on whether users leave Google, reducing Google’s incentives to invest, which in turn reduces the quality of the general search services consumers receive. Supra ¶¶ 571–574 (§ V.A.4.c); Tr. 5782:9–5783:14 (Whinston (Pls. Expert)) (Google’s contracts insulate Google from competition; Google need not worry about losing many customers if it fails to invest in improving its general search services.).

1128. “[Competitive markets] force firms . . . to do things that are good for consumers.” Tr. 5862:18–5863:13 (Whinston (Pls. Expert)); Supra ¶¶ 1079–1092 (§ VIII.B.2). For example, if Google were not insulated from competition, the benefit of using more data to improve search goes up relative to the cost of using more data, which means the rational business decision for Google will be to use more data and improve search to the benefit of users. Tr. 5862:18–5863:13 (Whinston (Pls. Expert)).

1129. Because quality, like price, is important to consumers, harm to quality can harm consumers even in a zero-price market like general search services. Tr. 5854:11–20 (Whinston (Pls. Expert)) (“there’s harm because quality is super important for consumers and if competition or lack of competition reduces the quality and the options that consumers have, consumers are going to be highly likely to be worse off.”).

1130. Harm to competition in general search services does not depend on there being monopoly power or harm to competition in the ads markets. Tr. 5859:2–16 (Whinston (Pls. Expert)).

1. Google Responds To (Rare) Competitive Threats By Increasing Its Search Quality

1131. Google’s conduct during rare events in the past when it faced competition as compared to Google’s conduct today shows that, when available, competition forces Google to work harder than it would have otherwise.

1132. For example, when Bing entered the market in 2009 and began to rise in quality, Google viewed Bing as a competitive threat and responded with a “Precision Code Orange.” UPX0974 at -470, -474. Amit Singhal, then-head of Search Ranking, told Google engineers that Google was facing a “serious competitive threat from Bing in ranking” and it “need[ed] to act fast and act decisively.” Id. at -474; Tr. 1815:4–17, 1815:21–1816:5, 1817:1–3 (Lehman (Google)) (discussing UPX0974; in response to Bing’s threat, Google’s search team had to come up with many ways to improve search quality); UPX0211 at -055 (“Precision is like the weather: Everyone talks about it, no one does anything. Then Bing got a big precision jump, and everything came to a head. Suddenly we actually had to focus full attention on precision.”); Tr. 5842:4–22 (Whinston (Pls. Expert)) (“[W]hen Bing was introduced, Google sat up and took notice.” (discussing UPXD104 at 65)).

1133. Google increased its focus on search innovation in response to Bing’s launch. Shortly after Bing was introduced, an internal Google strategy presentation stressed the urgency of the situation: “Bing is innovating. . . . We need to out innovate Bing[.] We have many more innovations sitting in the search quality area waiting to get out. Get them out.” UPX2086 at -512.005. Six months later, the quarterly CEO Report to Google’s Board of Directors declared, “[p]racticing what we preach – competition is good for Google, really!” UPX2087 at -697.

The Report further explained that, in response to competition from Bing, the “[r]ate of search quality improvements [was] up significantly,” there had been “[m]ore/faster search [user interface] innovation and iteration,” and Google had made “meaningful technological advances e.g. real time search, voice, instant translate, and visual search.” UPX2087 at -697.

1134. As part of its response to Bing’s launch, Google launched a search redesign that was “slightly revenue negative.” UPX0715 at -128 (describing “Skunkworks,” a redesign to the Google Search page) (emphasis in original).

1135. Similarly, when Google has faced a strong competitor in other geographies, such as in Russia with Yandex, and in South Korea with Naver, Google invested more in search. PSX00331 at -381–82, -385–90 (In a 2009 Google presentation on “Challenged Markets,” countries where Google is not the market leader, Google understood that it must “invest substantially more than standard level” and “unleash the innovation machine.”); Tr. 2718:5– 2719:24 (Parakhin (Microsoft)) (“[O]bviously, we’ve seen evidence that Google invests much more in [the Russian and South Korean] markets than in others . . . .”); Tr. 5741:20–5744:3 (Whinston (Pls. Expert)) (Yandex in Russia is an example of “circumstances where we see a stronger rival and we see a big change” in users after the institution of a choice screen.).

1136. Google’s conduct in the United States today sits in contrast to Google’s conduct in response these rare instances of competition. General search services have numerous quality dimensions that consumers care about and in which Google would raise its investment if it faced greater competition. Tr. 5854:21–5856:14 (Whinston (Pls. Expert)) (identifying privacy as one quality dimension and discussing Mr. Raghavan’s testimony and an email he wrote (UPX0501) that illustrate Google making the decision not to invest in privacy due to a lack of competition); id. 5856:20–5857:4 (There are “unending numbers of ways in which consumers care about quality dimensions of search services, whether it’s latency, . . . whether there are . . . local sports scores that are provided, et cetera.”); UPX0735 at -156–57 (insufficient machine capacity is resulting in launches, which required significant resources to prepare, being “blocked”).

2. Google Offers Consumers Fewer Privacy Options And, By Default, Collects More Consumer Data Than It Would In A Competitive World

1137. One search quality dimension is privacy. Des. Tr. 22:7–9, 22:12–16, 22:19 (Fox (Google) Dep.); Tr. 5854:11–5856:19 (Whinston (Pls. Expert)); Tr. 8694:4–17 (Israel (Def. Expert)).

1138. U.S. users care about privacy when it comes to general search services. Tr. 7471:5–25 (Raghavan (Google)) (People have cared increasingly about privacy, and Mr. Pichai had mentioned privacy as a focus at Google’s 2019 I/O conference.); id. 7451:20– 7452:21 (acknowledging prior statement that users should and will consider privacy when choosing a search engine and discussing research showing that users care about privacy); UPX1069 at -661–62 (“Privacy concerns are the biggest contributor” to declining user trust in the Google brand); Des. Tr. 156:21–24, 157:3–13 (Fox (Google) Dep.) (“[P]rivacy generally is a factor that some searchers are interested in and care about.”); Tr. 2484:4–8 (Cue (Apple)) (Technology users have a significant interest in privacy.); UPX0720 at -249 (DuckDuckGo survey showing that “doesn’t collect personal info” was top reason adult Americans switch search engines).

1139. Many Americans would like to avoid the harms associated with tracking by GSEs. Tr. 1943:3–1944:25, 1947:3–1948:20 (Weinberg (DuckDuckGo)) (explaining that “generally we find that a large percentage of Americans would like to avoid these harms” and describing the results of user surveys on privacy preferences).

1140. [INTENTIONALLY LEFT BLANK]

1141. Google does not perform well with users on privacy and trust. DX0183 at -892 (2020 Google study showing that Google had “weak performance scores” (24%) in the United States on top trust drivers “[r]espects my privacy” and “[k]eeps my personal data secure.”); UPX0419 at -489 (“We saw significant trust declines in 2018 in Google specifically.”); UPX0763 at -960 (Raghavan: “We (and I) felt the need to put a finer point on privacy this year, given it’s where we’ve been getting hammered.”); UPX0795 at -084 (“<20% of users think [Google] Search delivers on their expectation of privacy.”).

1142. Users dislike Google’s data storage policies and believe Google benefits more from those policies than they do. UPX0996 at -977 (2019 Google survey of 1000 users found 76% thought Google benefits more from storing users’ information than users do and 68% felt negatively about the length of time Google stores data vs. 6% that did not.).

1143. When assessing whether to pursue a potential privacy enhancement to Google Search, Google considers whether it is losing search queries or search revenue to rivals based on privacy. Tr. 7464:19–7465:12 (Raghavan (Google)) (Whether Google was losing queries to rivals because of privacy was one factor when deciding whether Google would pursue a privacy enhancement.); Des. Tr. 231:13–232:11, 232:13–21 (Edwards (Google) Dep.) (Discovering that a large number of users were switching to DuckDuckGo because of privacy would be a reason to more quickly launch a privacy change.); Des. Tr. 185:24–186:2, 186:5–15 (Fox (Google) Dep.) (discussing UPX0419 and explaining that working on privacy and user trust is a concern for Google especially if they could lose queries to another search engine); Tr. 5854:21–5456:14 (Whinston (Pls. Expert)) (Google considers the risk that it will lose business to competitors when it decides whether or not to pursue privacy initiatives.); UPX0810 at -410, -441 (“Key risks” for enhancing private search on Google include “revenue impact” and whether regulators or Apple could request private searching to be the default setting.); UPX0740 at -956–57 (Consumer council notes with an “Action Items” heading instructing Ms. Edwards to analyze the revenue impact of a potential privacy change.).

1144. For example, in 2019, Google identified ways in which DuckDuckGo offered better privacy features, but Google decided not to pursue privacy initiatives in the absence of competitive pressure. UPX0811 at -420, -445 (comparing privacy features between DuckDuckGo and Google). Google employees, including executives Nicholas Fox and Benjamin Gomes, suggested that Google could and should enhance its privacy offerings. Tr. 7469:6–9 (Raghavan (Google)); Des. Tr. 158:2–159:8, 160:23–161:16, 162:25–163:16 (Fox (Google) Dep.) (discussing efforts to improve Google’s privacy offerings, specifically with features like auto deleting user data to “appeal more to people concerned about privacy.”); UPX0794 at -146 (comparison between Google and DuckDuckGo on various privacy policies); UPX0501 at -520 (June 20, 2019 e-mail from Mr. Gomes (Google) discussing proposal to meet privacy challenges from DuckDuckGo); UPX0500 at -518 (2019 email thread relating to DuckDuckGo and Google potentially pursuing additional privacy features).

1145. In response, Mr. Raghavan, who was on the Consumer Council and would later become head of Google Search, asked for data about “what impact is [DuckDuckGo] having on [Google’s] search volume.” UPX0500 at -518; Tr. 7466:1–7468:12, 7599:7–17 (Raghavan (Google)) (discussing UPX0500). Before enacting any additional privacy features, Dr. Raghavan stated he would need to see data about whether Google was actually losing queries to DuckDuckGo. Tr. 7468:2–12, 7599:7–17 (Raghavan (Google)); id. 7472:19–24 (calling DuckDuckGo “ankle biters”).

1146. In a related email, Dr. Raghavan stated, “I agree that there’s something worth exploring in this space of private search. But the working teams have to do MUCH more careful work before wasting our valuable time. I want to see evidence that that there’s a real impact on Google users, attributable to this factor.” UPX0501 at -520. Dr. Raghavan continued, “I disagree with a methodology that consists of conflating ‘people care increasingly about privacy, DuckDuckGo is making a lot of noise about it, Sundar mentioned it in I/O’ (all true statements) then concluding that this needs a product change.” Id.

1147. Ultimately, Google did not bother conducting any studies or surveys to determine if Google was losing query volume to DuckDuckGo. Tr. 7472:6–13 (Raghavan (Google)) (Dr. Raghavan could recall no follow-up surveys.). As head of Google Search, Dr. Raghavan could have requested that such a study be done, but he did not. Tr. 7472:14–18 (Raghavan (Google)). This is an example of when Google decided not to pursue privacy initiatives in the absence of competitive pressure or a “threat of losing users.” Tr. 5855:19–5856:14 (Whinston (Pls. Expert)).

1148. A 2019 presentation at Google proposed various privacy enhancements that Google could make, and identified privacy features that DuckDuckGo offered its users that Google did not. Tr. 7412:5–7413:2 (Raghavan (Google)) (discussing UPX0811); Tr. 4167:22– 4171:10 (Juda (Google)) (discussing UPX0811); UPX0811 at -408, -420, -445. One privacy option proposed in the deck was an Incognito product for Google Search, i.e., a separate website (incognito.google.com) on which users could search with enhanced privacy. UPX0811 at -411 (Incognito.google.com would “[p]rovide strong PR privacy-moment that demonstrates Google forwardness.”); UPX0768 at -159 (Mr. Fox stating “I like this idea” with regard to standalone Incognito website); UPX0732 at -243 (“Google Search incognito is a proposal that Google offers a version of search that can be reached on a standalone domain, such as incognito.google.com. . . . The biggest drawback is that Incognito mode cannot be made a default search setting.”).

1149. The 2019 privacy presentation identified key risks associated with privacy enhancements, including revenue impacts of less than $[redacted] billion/year (if the privacy product’s reach was limited) or more than $[redacted] billion (if the product became mainstream). UPX0811 at -410; UPX0009 at -470. Google was concerned it could lose billions of dollars in revenue if the privacy product was popular. Tr. 7473:14–7475:9 (Raghavan (Google)) (Google decided not to offer Incognito mode for Google Search in part because of concern that Google could lose billions of dollars in revenue if users searched in Incognito mode.); UPX0768 at -158–62 (2019 email thread discussing possibility of a standalone Google Search incognito website).

1150. Ultimately, Google did not adopt the proposals in the 2019 privacy presentation. Tr. 7416:23–7417:3, 7474:7 (Raghavan (Google)); UPX6026 at -489 (written 30(b)(6) response: “Google has considered, but has not developed or implemented, certain other features or settings that could also be considered a component of some conceptions of an ‘incognito’ or ‘private’ search ‘mode.’”). Today, Google does not offer a private browsing mode for Google Search, including for searches conducted on google.com, the Google Search Widget on Android phones, or the Safari address bar. Tr. 7473:6–16, 7475:6–21 ((Raghavan (Google)).

1151. Private browsing mode is far from the only privacy feature that Google might reconsider if there was more competition. Google has opted not to create preset privacy bundles of low, medium, and high privacy settings for users despite the recommendation of its chief marketing officer. Tr. 9060:6–9061:9 (Fitzpatrick (Google)). Google considered, but rejected, an easier way for users to choose how long their data is saved. Id. 9052:1–9054:16. Google has the capacity to better protect particularly sensitive user data, but has opted not to. Id. 9061:10– 9063:12 (discussing UPX0981 at -847). Google has also opted not to add a VPN (virtual private network) to Incognito Mode. Tr. 9063:14–9068:10 (Fitzpatrick) (Google)) (discussing UPX0981 at -847).

1152. By default, Google logs and tracks search history for signed-in users for 18 months. Tr. 7454:8–15, 7459:20–23 (Raghavan (Google)); Des. Tr. 185:22–186:14 (Edwards (Google) Dep.) (User identifiers associated with search queries are “by default retained for 18 months.”). These default settings ignore Google’s research that shows 74% of Google’s users prefer that Google store user data for one year or less. UPX0996 at -978; Tr. 9055:11–9058:7 (Fitzpatrick (Google)) (reviewing UPX0996 at -978); UPX0794 at -146 (chart comparing DuckDuckGo’s and Google’s privacy).

1153. On Google, even if the user finds the default settings, the shortest auto-delete interval that a user can chose for their data is three months; 36 months is the only alternative Google offers. DXD-31 at 014. Google’s research, however, shows that 49% of users prefer that data storage last no more than a month and 17% thought Google should not keep their data at all. UPX0996 at -978; Tr. 9055:11–9058:7 (Fitzpatrick (Google)) (reviewing UPX0996 at -978).

1154. By default, Google also collects and logs search history for signed-out users and Google ties that search history to a unique cookie assigned to the user’s device or browser. Tr. 7456:16–23 (Raghavan (Google)); UPX6030 at -596, -598–99 (written 30(b)(6) response: “If multiple queries are associated with the same identifier, then they may be retained in that form for up to 18 months.”).

1155. The search histories that Google logs and tracks include a user’s search queries, the locations where they entered searches, the websites they clicked on, and granular interaction data such as their mouse movements, page hovers, and finger scrolls. Tr. 7460:2–7461:21 (Raghavan (Google)); UPX6026 at -469–70, -483 (written 30(b)(6) response).

1156. Before 2019, Google kept users’ search history forever, with no option for users to have their search history auto deleted. Tr. 7461:22–7462:14 (Raghavan (Google)); Des. Tr. 186:22–187:9 (Edwards (Google) Dep.) (The retention of user identifiers was “indefinite.”).

1157. Google offers an “Incognito mode” for Chrome, but this mode is not “truly private,” which requires Google to describe it in “really fuzzy, hedging language.” UPX0981 at -847.

1158. Google Search’s privacy settings are hard to find and adjust. Tr. 9011:4–22, 9050:10–19 (Fitzpatrick (Google)) (Changing users’ auto delete defaults takes as many as 10 clicks.); Tr. 638:3–23 (Rangel (Pls. Expert)) (discussing “considerable choice friction” in changing Google’s default privacy settings).

1159. Google violates privacy expectations by using personalized search history to target users with “creepy” advertisements, even when the users are not on Google Search. Tr. 7463:17–7464:7 (Raghavan (Google)) (Google monetizes user search history to personalize ads when the user is on YouTube, Gmail, and Google Discover); Tr. 3678:9–3679:16 (Ramaswamy (Neeva)) (“One of the most important ways a search engine knows its ads are effective is by observing [users’] behavior on other sites, especially sites on [the user has] clicked on the ads for. . . . [I]t leads to people having this vague sense of unease about their behavior being watched, or people use words like tracked.”); Des. Tr. 231:20–232:7, 232: 10–13 (Baker (Mozilla) Dep.) (discussing UPX1070 at -313 and users’ belief that “online tracking” is “creepy”); UPX0790 at -677 (quoting Google’s then-CEO Eric Schmidt, “Google’s policy is to get right up to the creepy line but not cross it.”); Tr. 1943:3–1944:9 (Weinberg (DuckDuckGo)) (discussing harmful effects of Google’s ad targeting).

By contrast, Google rarely uses personalized search history to serve organic results. Des. Tr. 188:9–15, 188:17–190:19, 201:2–15 (Edwards (Google) Dep.) (Google Search is “extremely lightly personalized” and search history is only used “on a very small percentage of queries.”); UPX1044 at -718 (“[P]ersonalization only rarely happens as part of the ranking process” because the query “generally provide[s] all that’s necessary to deliver good results.”).