Did you encounter a web, Facebook or Twitter advertisement seemingly tailored to your interests or related with your recent actions on the web? Chances are this was delivered to you via Real-Time Bidding channels.
I am involved in technical research and analysis of Real-Time Bidding systems and their potential influence on modern societies. However, there are also interesting policy aspects. The goal of this short note is to touch one of possible influences on democratic systems.
For a simple description of Real-Time Bidding systems (RTB), please refer to one of my works. To simplify, there are three parties involved in the system:
- The user, who enters a web site, launches a mobile app or other marvel of technology capable of displaying online ads
- A web site or mobile app which includes web ads scripts
- A script provider, who is operating Real-Time Bidding system, the Ad Exchange
- A number (tens, hundreds) of bidders, who bid for the user's attention
After detecting that a user has entered a web site (mobile app, etc.), the Ad Exchange holds an auction: it sends some data about the user to the auctions bidders. Bidders evaluate the obtained data and they decide if they want to display a message, typically an advertisement - to the user. They submit their bids, and all this happens fast (tens of milliseconds). At the end the user can see the message.
Creative uses of RTB
On a side note, Real-Time Bidding systems are not always used just to send web ads. Relatively recently, they are also very effective in serving malicious code (malware) to end-users. Sadly, a wide range of examples highlight the problem.
In general, it is widely known that even simple and perhaps innocuous differences of presenting information can potentially have influence on voters decisions. Tying this to the fact that users might have problems with distinguishing content from ads, brings to an interesting question.
What happens if users find it difficult to distinguish (e.g.) news material and actual ads? What if this happens during electoral campaign?
Real-Time Bidding provides very fine-graned targeting capabilities. Aside from receiving information on the user from the Ad Exchange (during auction phase), bidders are known to have their own databases, possibly obtained from Data Brokers. It is possible to target based on gender, education level, occupation, income, type of interests, location, and so on. It is also possible to target based on political inclinations.
"I think it's pretty safe to say that had a substantial impact on the election (...) you’re looking at swaying the votes of the very prized few who matter the most", says a representative of one of the advertising companies.
Informing the public about the aims or goals of a political party or advocate group using just another channel (next to TV, news papers or so) might not seem to be surprising. However, since there are no clear transparency messages embedded in the Real-Time Bidding messages, the users potentially have no way of understanding who is actually targeting the content to them.
Imagine a situation when external players from a country X's agency will suddenly start targeting users in a particular country Y, prior to or during the election campaign? Such campaigns could be executed cost-efficiently. And what about a situation of an agency of a country X who would register as a bidder of an Ad Exchange system?
The technical ability of influencing a democratic process by external players cannot be ruled out. As more and more ads is being served using targeting, via means such as RTB, this is a debate hard to avoid.
Real-Time Bidding provides rich capabilities of serving content to users. While there are security, privacy and transparency questions, the infrastructure and technical capabilities pose interesting issues.
It is interesting to see if - or when - we will encounter the use of Real-Time bidding in disinformation or information warfare.