Advertising pays for most of our online experience. We consume free content and services in return for looking at some ads. Fair deal, or so it seems. Upon examination, there are some dark twists - according to Shoshana Zuboff Ad Tech is at the origin of surveillance capitalism.
The story begins innocuously enough: Google's founders had little enthusiasm for advertising. This changes when the dot com bubble bursts in the spring of 2000. They then embrace advertising to survive, bringing a new twist that makes the company fabulously wealthy and immensely powerful.
The dominant advertising model before the bust was banner ads sold by sales reps on a cost-per-thousand impressions basis (CPM). In contrast, Google sold unobtrusive text ads to appear in a sidebar. Sergey Brin is quoted as saying "It didn't generate much money." In October 2000 Google introduces AdWords, a self-service platform allowing advertisers to purchase advertisements when the search query matches their selected keywords. In other words, ads are selected when they are contextually relevant. Although ads are still sold on a CPM basis, from the outset Google claims a higher click-thru rate (CTR) than their competitors'. They also instrument search pages so that advertisers can follow ad performance in real time. In February 2002, AdWords becomes AdWords Select and changes its pricing model from CPM to CPC (cost-per-click). The ad auction model is largely copied from Overture, then a larger competitor, later acquired by Yahoo! to become Yahoo! Search Marketing. But Google makes a striking innovation: instead of awarding top spot to the highest bid, it also factors in expected CTR when deciding ad ranking. A market for prediction services is born! The context into which the ad is to be placed ceases to be the sole criteria for selection. User behavioral data also becomes hugely relevant.
With the launch of AdSense in 2003, Google moves into the advertising network business, auctioning ad space on sites of any publisher that cares to join the program.
In 2007, DoubleClick releases its Ad Exchange, thus ringing in the era of Real-Time Bidding (RTB), today's dominant internet advertising mode of operation. RTB automates the process of allocating adverts to publishers. This capability is referred to as programmatic advertising. This is not new as it was already present to a certain extent in AdWords and matured with the advent of advertising networks. However, these networks operated within walled gardens in which advertisers and publishers have reasonably long-standing relationships, if not directly with each other, then with the advertising network. By contrast, ad exchanges allow advertising opportunities to be traded in an open market. Thus, they receive requests for bids for advertising real estate, forward them, and, adjudicate the highest bid. Usually, advertisers and publishers do not interact directly with the exchange, but delegate to Demand Side Platforms (DSP) and Supply Side Platforms (SSP) respectively. In order to maximise reach, it is common practice for publishers to contact several SSPs, which in turn request bids to several ad exchanges, which, in turn broadcasts to tens, or even hundreds, of DSPs.
RTB comes in 2 flavors: Google's and the rest of the world's, gathered under the IAB (Interactive Advertising Bureau) umbrella. Google's proprietary RTB protocol and ad exchange is branded Authorized Buyers and has its origins in DoubleClick Ad Exchange, acquired in 2008. Google also embraces OpenRTB, IAB's protocol. Both eco-systems are subject of complaints to data protection authorities for infringing users' rights. At the core of these complaints is the sensitive nature of the broadcast information. RTB protocols provide fields in requests for bids for detailed descriptions of both context and user.
An OpenRTB bid request for web ad real estate, for example, may not only contain the domain together with an array of content categories to describe the site, but also the page loaded with its array of content categories and its referrer, including search string. Few people would object if this were the end of it - advertisers placing contextual ads seems logical and smart: if a user is looking at a golf site, why not show an ad for golf clubs? However, when details of the user are also included in the bid request, as they routinely are, serious privacy concerns emerge. OpenRTB provides age, gender, location, and keywords indicating interests and intent. It also identifies the user, either by a vendor-specific ID that can be reconciled by the bidder/buyer through match tables, or by a buyer ID that has previously been agreed with the exchange. Similarly, mobile devices are tracked by a unique identifier. Now we are in the realm of behavioral, or personalized, advertising: ad buyers can watch users interact with a range of web and mobile apps and tailor their ads to user actions.
In fact, buyers frequently cross-reference data they obtain in bid requests from their DSP with data purchased from a Data Management Platform (DMP) to further narrow advertising segments. DMPs, in turn, purchased that data from data brokers - often these are third-party trackers, but they could also have gathered this data with, for example, social media games or quizzes.
Revenue from advertising is the dominant publishing business model. Hence publishers are incentivised to optimize for eyeballs and click-through rates, which in turns leads to personalized clickbait. The most familiar type makes a controversial, sometimes outrageous, statement designed to pique interest. More sophisticated forms tailor the message to the individual user using Machine Learning. YouTube provides an interesting example of how this has nasty consequences for both individuals and society: there are some well-documented examples of people who became radicalized through YouTube. This happens, not because YouTube intends to create radicals, but because its recommender algorithm is designed to maximize the time spent watching videos. In other words, YouTube does not have an agenda to turn you into an extremist, it just want to turn you into a couch potato. In the process, it creates opportunities and incentives for extremists to provide content to lure people into their rabbit hole.
Calls for censorship have been gaining strength. While there is are good arguments for holding (social) media companies accountable for who they give a platform to, tackling the root causes of the spread of hate speech and fake news is even more urgent - incentives created by the current advertising market need fixing! Current EU regulation goes some distance toward this, but has so far failed to deliver a knockout blow.
To a lay person like myself, it is difficult to see how tracking users in their journey around the internet can be reconciled with existing EU legislation such as the General Data Protection Regulation (GDPR) and the 2002 ePrivacy Directive, updated in 2009. Yet this common practice has resulted in few fines so far. There seems, therefore, to be a problem with enforcement. Some will argue, on the other hand, that the legal basis for protecting citizen's privacy rights needs to be strengthened - indeed the ePrivacy Directive was intended to be replaced by a new ePrivacy Regulation - Regulation on Privacy and Electronic Communications in full - coming into force at the same time as the GDPR. However, negotiations have proved difficult and bump along from presidency to presidency.


Popular Posts