Could a Personal Assistant Be So Devious?
by Ariel Ezrachi and Maurice Stucke
Shoppers with Internet access and a bargain-hunting impulse can find a universe of products at their fingertips. In their thought-provoking exposé, Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy, Ariel Ezrachi and Maurice Stucke invite us to take a harder look at today’s app-assisted paradise of digital shopping. While consumers reap many benefits from online purchasing, the sophisticated algorithms and data-crunching that make browsing so convenient are also changing the nature of market competition, and not always for the better. The book raises timely questions. To what extent does the “invisible hand” still hold sway? In markets continually manipulated by bots and algorithms, is competitive pricing an illusion? Can our current laws protect consumers? The changing market reality is already shifting power into the hands of the few. Ezrachi and Stucke explore the resulting risks to competition, our democratic ideals, and our economic and overall well-being. Here is a brief excerpt.
Who wouldn’t want a personal butler? The idea of an intelligent, voice- activated helper is alluring. And yet, when considering its possible strategic usage by the winners in the Frenemy dynamic, our butler’s interests may not always align with our interests. Our new trusted alter ego, to which we outsource our decision making, may be charming, but partial. After all, as we learned earlier, being the “free” part of a multisided market, we don’t directly pay for the butler’s services. Our butler must ultimately cater to the needs of its real employer — the super-platform. Of course, we can still benefit when the super-platform’s interests are aligned with our own. But we may often be unaware of when such alignment is absent.
So what differentiates the strategies discussed earlier in our Frenemy scenario and the future use of personal assistants? We identify several key differences that affect the user experience and will likely act as “game changers” — transforming our competitive environment.
First, we note the changing interaction with the new digital helpers. Because it is human in its communications, charming in its demeanor, and funny — at just the appropriate level — we will grow to trust our companion, which has been privy to so many of our activities. Many of us already trust our favorite search engine to find the relevant results for our inquiries, Facebook to identify relevant news stories, Amazon for book recommendations, and Siri to place phone calls, send text messages, and find a good Chinese place nearby. So with an eager (and free) butler whose capacity to help us improves, we will increasingly rely on it. At first the choices will seem benign, such as asking the personal assistant for today’s weather rather than searching the web. But the more tasks the personal platform undertakes, the more we rely on the super-platform’s functions (such as its texting app, maps, and so forth). That trust, in competitive terms, can easily translate to our willingly being locked in — that is, willingly forgoing opportunities to independently consider outside options.
Second, the more we communicate only with our personal assistant, the less likely we will independently search the web, use price-comparison websites, seek independent customer reviews, and rely on other tools. The ease of voice activation and verbal communication with our butler may limit our view of the available outside options.
Today when an app is preloaded on our smartphone and integrated with the phone’s other functions, few of us download a competing app. As EU Commissioner Vestager noted, “if Google’s apps are already on our phones when we buy them, not many of us will go to the trouble of looking for alternatives. And that makes it hard for Google’s competitors to persuade us to try their apps.” Many of us stick with the default option. Likewise, as our personal assistant becomes our default, so too will its platform’s other features and functions. To illustrate, suppose we are sipping our single-malt Scotch in our summer retreat. We may ask our assistant how long a drive it is to a friend’s country home, and then ask our assistant to text our friend our ETA. We would be unlikely to pull out our Android phone, search MapQuest to figure out the distance and time to drive, and then use Messenger to text our friend.
The removal of the human element from the search activity, and partly from the decision making, transfers more power to the super-platform. The personal assistant will use its own tools and may exercise its own judgment as to prioritizing and communicating the results.
Third, the scope of data and the personalization that follows will make it harder for us to switch assistants. As Google’s CEO characterized its personal assistant, “We think of it as building each user their own individual Google.” The super-platforms, given the scope of data, opportunities to experiment and learn by doing, and control over key technologies (such as maps), can already provide us with a personalized experience that smaller providers are unlikely to match. Once we choose and train a head butler, we may tolerate mistakes rather than train a new butler from an- other super-platform.
The super-platform — through its butler — will benefit from unparalleled access to our data. As we noted earlier, the super-platforms already spend a lot of time, money, and effort to track our behavior and profile us. Now, by providing us a butler, the super-platforms will collect even more data. As repeatedly noted by developers, the hope is for these tools to accompany us in our decision making, encouraging ongoing daily interaction from chats to shopping. As a result, the information gathered about our needs and desires will be significantly enhanced — feeding the Big Data and Big Analytics machines.
As the super-platform’s power increases, so does the risk of anticompetitive activities. Indeed, the personal assistant can magnify the harmful effects of each of our three scenarios.
Think, for example, of our behavioral discrimination scenario. Our personal assistant will accumulate increasing information about us and will be aware of the extent to which we venture out and seek other options. Its aim is to deliver the right product or service at a price we are willing to pay. So the line between personalization and behavioral discrimination will blur. As we increasingly rely on the personal assistant for suggestions, it can increasingly suggest things or services to buy, and the price it has success- fully negotiated. While helping our son with his Spanish, our personal assistant might suggest a particular app or private tutor that tremendously helped other students struggling with the same issue. Because the tutoring is customized for our son, it will be harder to assess whether the price the tutor charges is the fair market price or simply a price we would tolerate. Moreover, if the tutoring service is helping other children improve their grades, we would not want our child to be at a competitive disadvantage — especially if we are all eyeing the same highly selective universities. So the personal assistant can prompt purchases that we otherwise wouldn’t consider.
Now let’s think of our collusion scenarios facilitated by smart algorithms. As our digital assistant increasingly orchestrates what we purchase at what time, it can easily become the hub upon which the spokes rely. Sup- pose the personal assistant gets a commission for every sale. Higher prices mean higher commissions. The more the personal assistant is plugged into our lives, the harder it will be to circumvent the assistant in finding a lower price. And when we independently find a lower price, our butler can always surprise us with a special deal, whose aim is to punish any discounter and undermine any attempt to cheat on the tacitly established price.
Finally, the harms we identify in our Frenemy scenario are amplified. While we would each enjoy our personalized experience and growing intimacy with our digital personal assistant, its roots and loyalty remain with its real master. As more people rely on the super-platform’s personal assistant for their day-to-day activities, so too will sellers gravitate to the super-platform. The punishment for being kicked off the super-platform is severe. Without access to the personal data stream, the independent re- tailer will have a harder time identifying a customer’s key purchasing moment (like when he needs a new oxford shirt). Because the ads we see while surfing the web will be orchestrated by the super-platform, it will be harder for the retailer to reach that customer. Even if the retailer can reach the customer, it cannot provide the increasingly customized products or services (such as tailored shirts in the styles and colors that appeal to the customer). And even if retailer can gain the customer’s attention, the personal assistant may interject with its own recommendation, suggesting that he consider a special deal by another haberdashery, one that is part of the super-platform’s ecosystem. In this multisided market, the assistant may subtly push certain products and services and degrade or conceal others, all in the name of personalization.
