What Do Finding Love and Managing Diabetes Have in Common?

Hint: Just get back to the basics

In 1995, I got a call from a friend who had just completed his PhD in computer science, and he wanted me to help him find a business model for a new technology he developed. His doctoral thesis was an algorithmic model that could predict what people would like based on other things they’ve expressed interest in. This was of a branch in AI called, “predictive preferences,” what we now call a “recommender engine” or “recommendation algorithm.” (Years later, Amazon would be the first company to use a similar algorithm to recommend books you might like based on ones you had previously purchased. At the time, Amazon only sold books. How quaint.)

He came to me because I also had a background in AI and had already had two successful startups under my belt, so he hoped I’d be able to explore business ideas and work with investors on funding. He showed me his prototype: a website where the user would rank at least 10 movies they liked, and from that, it would recommend other movies they might like. The feedback from his beta users said the predictions were shockingly accurate, and he was looking for a more viable business idea.

I instantly felt that online dating would be a great opportunity because, in April of that year, match.com had just launched, and everyone in the San Francisco Bay Area was using it. When I got my degree in AI in 1985, everyone was bullish on how emerging algorithms could be useful across a number of industries, and dating seemed to me to be a perfect candidate. Match wasn’t using it at all, but I also had my own bias: I had written a paper in college on the Five C’s of Romantic Relationships that almost got me into a graduate program in Psychology at Stanford, but I instead decided to remain in computer science. By 1995, I figured, AI would certainly have improved well enough to unseat Match in their dominant position.

In short, my idea was to have a dating site that simply paired people together based on common interests that we could use the algorithm to figure out. People didn’t write profiles, or see information on other people at all. They’d sign up, and provide basic info like their gender preference, age and location, and then do a very simple questionnaire that involved checking the box next to items that you did regularly: Books, magazines and newspapers that you read; movies and TV shows, musical genres and bands; types of food you ate, places you’ve traveled to, whether you go to the gym, attend a religious institution, and so on. We didn’t ask users to rate any of these things—just whether they engaged in them.

Nothing speaks more honestly about a person’s interests and their values and commitments than what they’re willing to spend time and money on.

We then paired them with others with similar interests and sent each an email saying, “You two might like each other. We suggest seeing [movie] this weekend, or going to [restaurant] to chat about [topic].” But we also gave advice based on my own research in college: Don’t just email each other; meet right away, go for walks, and above all, talk! There is no better way to build a relationship than spending time in real world settings. Just stick to the basics!

To measure the platform’s success, we emailed feedback forms that asked whether people found the site useful, and whether they found their match. We found that 80% of users found their match within the first month.

I circulated the data and our pitch deck around to the VCs in Silicon Valley, and all of them were very impressed, but said, “We’d love to stay in touch.” Of course, I knew what that meant: No. But this made no sense to me. We were clearly beating Match’s own stats, so why not us? I was close enough to one of the VCs to ask candidly what their reservations were. He said simply: “What’s your revenue model? If the platform works that well, how do you keep them paying month after month?”

That’s when I had the palm-to-the-forehead moment. Duh. Of course.

The online dating industry has changed in many ways, but its business model is still driven on the promise that they can make dating easier than doing it the old school way—the basics. The best way to sell it was to convince them that going out and meeting people is hard, and time-consuming, and frustrating, and you don’t want to deal with the hassle.

Look, if nothing else, it’s better than your not going out at all.

Today, those who are most engaged in online dating—the super-swipers—rationalize their behaviors as, “The algorithms will just get better!”

But the results aren’t any different, and the stats show that people who are actually serious about finding a partner are going back to the basics, by meeting in real life, doing activities they like, at locations they frequently go to, and being introduced by friends where they have dinner and talk. You know, the same stuff people did before the internet. And it’s the same guidelines we recommended to couples that used our unfunded-but-successful beta platform.

The problem is, the basics don't make money.

That doesn’t mean the online platforms don’t work or are doomed to failure. They’re tools—you have to use them wisely, carefully, and sparingly. Those who are most successful at dating are doing the very behaviors that our old prototype dating platform suggested back in 1995: They use the service to get introductions, but they immediately meet and go do things in the real world.

What does all this have to do with diabetes management?

Pretty much everything.

T1Ds think of their disease like dating: It’s hard, and it’s frustrating, and they don’t want to deal with the hassle. They want to put as little effort into it as possible, while acknowledging they need to be healthy. But, in an effort to avoid the real work, they seek technologies or diets or any shortcut that promises to do the hard work for them. Automated insulin pumps are a good example: Just input your basic details, and the algorithm will figure out the rest.

Indeed, automated insulin pumps will do better for patients than if they did nothing at all.

Unfortunately, these systems are able to achieve only borderline health outcomes—good enough, but not great. This is just like how online dating sites give you people who are “close, but not quite right.” They keep you going, perhaps for years, but as you age, the elevated glucose levels will still get the better of you. See my article, Benefits and Risks of Insulin Pumps and Closed-Loop Delivery Systems.

Many have simply gotten accustomed to this, and rationalize their behaviors as, “But the algorithms will just get better.” But they can’t—not with dating or glycemic control algorithms. See my article, Challenges Facing Automated Insulin Delivery Systems.

People who are actually serious about getting their glucose under control are going back to the basics: They watch glucose levels frequently, take insulin or carbs in a timely manner, log events and activities, and exercise.

The problem is, the basics don't make money.

It’s not that tech is useless, either for dating or for diabetes. Dating websites can be used as a tool to help meet someone, but the real work is experiencing another person in real-world settings to truly understand yourself and your dynamic with them. Learn your partner’s idiosyncrasies, where they’re sensitive, when they’re volatile. Learn to anticipate their needs and yours. How to react properly, and ideally, before problems start.

Exactly the same is true for T1D management: Use technology like CGMs and others as an adjunct to self-management, but don’t offload the whole process to it. You need to be fully and wholly engaged with your disease under real-world conditions and understand your personal dynamic with it. Observe glucose volatility so you can learn to anticipate it, and to react properly, and ideally, preemptively, to keep glucose levels from getting out of control. Sure, there are blow-ups now and then, but that’s part of the journey.

Like dating, diabetes management starts with the right mindset, and expectations need to be set properly. In dating, meeting someone is far different than keeping them. The storybook Disney movie that ends with the grand wedding is how we think of relationships, but in the real world, the wedding is the beginning of the relationship. Ask anyone who’s been married for fifty years, and they’ll tell you stories about the ups and downs, but what really works has nothing to do with the courting process. What works requires getting back to the basics. That’s how I came up with my research paper on the Five C’s of romantic relationships: Communication, Commitment, Compassion, Compatibility, Chemistry.

Same thing with the basics of diabetes management: starting those tasks is different from commitment towards them. As I explain in my article, Self-Identity and the Four Habits of Healthy T1Ds, you have to build experiences over time. Ask any T1D who’s had it for fifty years and longer, and they’ll tell you the same.

In both dating and diabetes, you must first adopt the mindset—the identity—of the archetype of the person you want to become, as that will drive your behaviors and your self-expectations. That person is fully engaged and invested in their partner, whether it’s a person or your life-long disease. You’re going to be together forever, so build the healthy habits that lead to happiness.

Happy Valentine’s Day!

Now, go eat chocolate and dose properly for it. Hopefully, with someone (else) you love.