Skip to main content

Newsroom

Is Your Data Wearing a Black Hat?

Marc Resnick

The most recent Bentley Research Colloquium focused on Big Data and a broad range of issues and topics surrounding the topic. This series highlights some of the issues examined or suggested by colloquium presenters.

The potential benefits of ambient monitoring (collecting and analyzing your data) with ubiquitous computing are so promising that it’s been called everything from “the future we have all been waiting for” to “magic.” Of course, we need to take these claims with a grain of salt; there are many technological paradigms that become the subject of extreme hype in the media (and in their stock prices). But ubiquitous computing, driven by the insights of Big Data, might be the one that has the highest hopes attached to it. What is this silver bullet that will launch us finally into the future?  Consider the following two scenarios.

Scenario 1

You get out of bed and your smart floor senses your feet are cold. It turns up the heat just the right amount, adjusting for the ambient temperature in the room and your personal preferences.  At the same time, it checks your calendar and notices you will be heading for work this morning with an important meeting first thing. In response, it cranks up the coffee maker using the extra-bold, extra-caffeine blend. Because of heavy traffic, it notifies you to take a quick shower and hit the road. It could have woken you earlier, but your sleep monitor noticed you haven’t been sleeping well this week and you needed to get your full eight hours. The GPS in your car is preloaded with a path that avoids the worst of the traffic. 

Two hours later, as you leave the meeting, you get a notification that your elderly mother did not get out of bed this morning. Her smart floor monitors specifically for this risk because of her medical status. She needs to take several medications before breakfast and the pills are still sitting in her bathroom dispenser. Your phone is ready to call her medical bracelet so that you can check up on her personally. All it takes is a single swipe.

Before leaving work that evening, you log into your refrigerator to see what is available for dinner. You have almost all the ingredients for chicken parmesan and Caesar salad. The fridge verifies with your blood sensor that your cholesterol is within the limits your doctor set on your electronic health record and with your fitness band that you’ve done your exercise for the day. Noticing you had a great day at the gym, it suggests adding a glass of your favorite merlot. The missing dinner ingredients are ordered from the local grocer and scheduled for immediate delivery — to be waiting for you when you get home. As you open the front door, the oven preheats and the evening news appears on the TV.

Scenario 2 

Your alarm wakes you up 15 minutes early because of heavy traffic on your normal route to work. There is another that might avoid the traffic, but it’s devalued in your GPS algorithm because you would not pass the sponsor’s billboard. You get the service for free in exchange for this tradeoff. The temperature in the room has already been raised because your spouse got up 30 minutes ago, even though he hasn’t been in the bedroom since. The new trendy coffee blend starts brewing in the coffeemaker because 90 percent of your friends have made the switch, but it’s too bitter for you. Normally, it would add two packs of sugar, but your blood glucose level is too high this morning. As you put the car into drive, the audio system automatically plays the training program that your company assigned. 

Two hours later, as you are leaving your morning meeting, your phone buzzes. Appearing on the screen is the camera in your elderly mother’s bedroom showing you that she has not gotten out of bed. You send an alarm to wake her up. If she doesn’t move, you can call 911 with a single swipe. The alarm startles your mom awake and you watch as she reaches over to hit snooze.     

Before leaving for work that evening, you log into your refrigerator to see what is available for dinner. Kraft is sponsoring your service this week, so you get recommendations for a pasta recipe made with Kraft Italian dressing, Kraft mac & cheese, and Kraft mayonnaise. The missing ingredients are automatically ordered from the local grocery so they will be waiting for you when you get home. It checks with your fitness bracelet, notices your great day at the gym, and recommends having a glass of wine if you watch a 30-second ad.

These are extreme scenarios and neither of them is feasible today. But they are both just a few years away and will employ the same technological advances. The question is: Which direction do we want to go? 

The first scenario is what my team at Bentley calls “White Hat Design Strategy” — putting the needs of the user first and working business objectives into transactions only when they are in the user’s best interest. With careful strategic design, this approach can achieve a sufficient profit to keep shareholders extremely satisfied. The second scenario is what we call “Black Hat Design Strategy” — prioritizing business objectives and working user requirements in only where they are necessary. There are still many new benefits offered to users, but not quite as seamlessly as in scenario one. And in Black Hat design, the user makes some more tradeoffs.

You may think that no company would choose a Black Hat approach if a White Hat one is available. But the market tells us otherwise. The following outlines the problems with Black Hat design, but also shows how, driven in part by Big Data, it’s already being used in today’s market — and why it will continue to proliferate, unless we make a conscious choice to change directions.

Priorities

The first problem with Black Hat design is that advertising revenue is prioritized over user choice. We already see ad-based services spreading throughout social media through sponsored links. Or entertainment media funded through streaming that requires watching ads before and periodically throughout the material you want to see. There is nothing wrong with making that tradeoff consciously. 

What sets White Hat design apart is that it helps us learn how to create very accurate and sensitive models of users. If these are unobtrusive, they can be used to show consumers advertising for products they really might be interested in at that specific time and place. This type of advertising is the most valuable for the marketer, but also for the consumer. The problem with Black Hat versions of these ad-based services is that the ads are set by auction and the targeting is not always as precise as it could be. The same ad follows you wherever you go. For example, I went to one store’s website because a student of mine used it as the basis of a homework assignment. Now an ad for it shows up on Facebook, Amazon, CNN and everywhere else I go online. I have nightmares about it.

Collaborative Filtering

Collaborative filtering is the general term for algorithms that use the behaviors and choices of a user’s social network to predict what he or she might prefer. The error that Black Hat strategies make is that while we do indeed share a lot of interests with our friends, this is a huge overgeneralization. There are some contacts that we trust for food or apparel, others we trust for financial advice, and still others for work-related activities or medical choices. We can see collaborative filtering overgeneralization errors cropping up already. Many of the ads that we see on our social-media feeds are based on what our contacts on that system have responded to. Facebook selects sponsored links to plug into our feeds based on what our Facebook friends have engaged with. Or profile aggregators combine behaviors from all of the networks, websites, and a variety of other sources to compile extensive user profiles and use the aggregate to make recommendations. The problem is that the aggregate is too great of a melting pot. We are all individuals. We don’t necessarily like the same TV shows or music as our friends.

White Hat strategies allow the user to select whose behavior they want recommendations to be based on for each channel or product category, separately. Perhaps I want recommendations for restaurants that have been reviewed highly by my social friends on Yelp, but I want books recommended by a trusted social-media marketing expert whose blog I subscribe to. As with ad-based consumption services, these recommendations are more targeted and more likely to be acted upon, benefitting both the marketer and the user. And with the increase in customer experience, the social-media provider benefits as well.

Third-Party Oversharing

A third error we see with Black Hat strategies is that they use information that the consumer may not be comfortable sharing with particular third parties. It may be OK to share blood-sensor data with a user’s doctor or pharmacist, but perhaps not with Kraft when deciding what foods to recommend she eat for dinner. White Hat strategies err on the side of discretion, only sharing with third parties what the user has opted in to share. One user might be willing to have a chip in his car monitor his driving behavior in exchange for discounted auto insurance when he drives safely. It is very easy to ask first.

Black Hat strategies, however, don’t seem to understand the importance of opt-in defaults. When Samsung recently warned users that they should not talk about private information near their smart TVs because it will be captured and shared with third parties, there was an understandable protest. This kind of sharing should always be opt-in and easy to customize. Marketers should never require consumers who may not be technically savvy to fiddle with complicated privacy settings or else give up in helplessness. We might be willing to accept the tradeoffs of third-party exchange if the benefits are valuable. But we demand that choice.

Summary

The differences between White Hat strategies and Black Hat strategies might seem obvious. But the prevalence of Black Hat design in today’s media landscape suggests that it is not as obvious as we think. The sad part of this story is that most companies that adopt Black Hat strategies think that the revenue and profit potential is higher, justifying their choice. But in the long run, it is just the opposite. When companies create effectively targeted marketing, behavioral suggestions, business transactions, and even enforced constraints such as those discussed in the scenarios above, many consumers are open to the possibilities — as long as they feel aware, in control, and competent to maintain that control.

Professor Marc Resnick (Experience Design) has more than 25 years of experience in human factors and usability. He has applied human factors to a variety of domains, including enterprise systems, health-care information systems, consumer products, and social networks.

Big Data Series

The Promise and Threat of Big Data: Inside Bentley's Research Colloquium
Digital Health Data Matters for Cancer Survivors
Are Wearables Destroying Your Privacy? 
When Googling Goes Bad
Finding the Signal in the Noise of Big Data
The Trouble with Big Data When It Comes to Women on Corporate Boards
Is Your Data Wearing a Black Hat? 
Join the Transforming Tech Industry