David paused to let the video sink in before resuming. “It’s not enough to provide a general set of recommendations. Different people are motivated by different kinds of language, different styles of communication, different reasons. Let’s use another example. An employee is going to ask his manager for extended vacation time. He’d probably like to make a compelling case for granting that vacation request. What will motivate his manager? Should he mention that he’s been working overtime? Should he mention that he needs to spend time with his kids? Or that he’s planning to visit the Grand Canyon, a place that his manager happens to associate with good memories?”
“The answer,” David went on, as he paced back and forth in the front of the room, “is that it depends on the person you’re sending it to. So ELOPe customizes its analysis not just to what the sender is asking for, but for what the recipient is motivated by.”
David noticed that Rebecca Smith was standing in the doorway listening to the presentation. In a sharp tailored suit, and with her reputation hovering about her like an invisible aura, the Avogadro CEO made for an imposing presence. Only her warm smile left a welcoming space in which an ordinary guy like David could stand.
She nodded to David as she came in and took her seat at the head of the table.
Kenneth asked, “But what you’re describing, how does it work? Natural language processing ability of computers doesn’t even come close to being able to understand the semantics of human language. Have you had some miracle breakthrough?”
“At the heart of how this works is the field of recommendation algorithms,” David explained. “Sean hired me not because I knew anything about language analysis but because I was a leading competitor in the Netflix competition. Netflix recommends movies that you’d enjoy watching. The better Netflix can do this, the more you as a customer enjoy using Netflix’s movie rental service. Several years ago, Netflix offered a million dollar prize to anyone who could beat their own algorithm by ten percent.”
“What’s amazing and even counterintuitive about recommendation algorithms is that they don’t depend on understanding anything about the movie. Netflix does not, for example, have a staff of people watching movies to categorize and rate them, just to find the latest sci-fi space action thriller that I happen to like. Instead, they rely on a technique called collaborative filtering, where they find other customers just like me, and then see how those customers rated a given movie to predict how I’ll rate it. Sean’s insight was that since natural language analysis struggles to understand semantics, it would be best to start with an approach that doesn’t rely on understanding, but instead one which utilizes patterns.”
When David received nods from the audience, he went on. “That’s what ELOPe does. It looks at the language used by millions of email users. It looks at the language received by people, and how they reacted. Did they react positively or negatively? Compiled over thousands of emails per person, and millions of people, we can find a cluster of users just like the intended recipient of an email, and see how they respond to variations of language and ideas to find the best way to present information and make compelling arguments.”
Now there were some puzzled looks and half raised hands as people around the room tried to ask questions. David forestalled them with a raised hand, and went on. “Hold the questions for a second, and let me give you a simple example. Let’s imagine that a person called Abe, whenever he received an email mentioning kids, responded with a negative response.”
David gestured back and forth with his hands, getting into the example. “Now imagine that ELOPe has to predict whether a new email about to be sent would be received positively or negatively by Abe. If that new email also mentioned kids, it’s a good bet that it will be received negatively. If Abe was your boss, and you were going to ask him for vacation time, it’s probably not a good idea to use spending time with your kids as justification.”
He heard a few chuckles.
“So is there is no semantic analysis?” Rebecca asked. “We don’t know why Abe dislikes kids?”
“No, we have no idea why Abe feels the way he does,” David answered with a smile. “We just observe the pattern of behavior.”
“What if my manager hadn’t received any emails about kids?” Sean protested. “How could we predict how he would respond?”
David smiled, knowing that Sean knew the answer, and was just helping him along. “Let’s say we have another user, Bob. Bob hasn’t received any emails about kids. However, ELOPe notices that Bob, Abe, and about a hundred other people have responded similarly to most topics, topics such as the activities they do on the weekend, the vacations they take, how they choose to spend their time. Let’s say that this group of people are ninety-five percent similar. That is, across all the topics they’ve responded to, they are ninety-five percent likely to have similar sentiment in their response: negative or positive. This is what we call a user cluster.”
The executives around the room nodded in understanding, and David went on.
“If other members of the user cluster received emails about kids,” David explained, “and they all responded negatively, then ELOPe will be ninety-five percent certain that Bob will respond negatively. Of course, it’s rarely so cut and dry, and it is a statistical prediction, which means that five percent of the time ELOPe will be wrong, but most of the time it will be right. So Sean, if your boss was Bob, I wouldn’t recommend mentioning kids when you ask for vacation.”
David waited for few chuckles from the audience. “Joking aside, ELOPe is working, and we’ve tested it with users. On average, favorable sentiment in reply emails increases twenty-three percent with ELOPe turned on compared to the baseline. That’s twenty-three percent more vacations granted, twenty-three percent more people agreeing to go on dates, twenty-three percent more people getting their work requests granted.”
Rebecca stared at David. “Wait a second. Going on dates? If that’s the case, you’ve got a woman out on a date with someone she wouldn’t have otherwise been with. That sounds manipulative and risky.”
Kenneth looked startled by Rebecca’s objection, and started talking quietly to Sean, sitting next to him.
David felt his internal danger meter flare into the red, and his stomach threatened to leap into his throat. Oh, the dating example was so damn controversial. The next few minutes would make or break his project. If Kenneth and Rebecca decided against the project, it would be impossible for Sean to give him the support he needed to get his project released.
“Hold on. Maybe I chose a bad example.” David held up both hands, to forestall any more objections. “Who’s taken a Myers-Briggs personality workshop?”
Everyone held up their hand or nodded in assent as David expected. The Myers-Briggs personality work or something similar was standard fare for every manager in every large corporation. Then he continued, “Now, what was the purpose of the workshop? It’s not just to find out you were an introvert or extravert, right?”
“No, it’s to learn to work effectively with others,” Sean provided helpfully.
“Working more effectively means what?” David paused. “It means learning how others communicate and think. It means learning who is likely to appreciate a data-driven argument versus an emotional argument. It means learning who is likely to want to think out loud, versus who wants to see the arguments written down and have time to respond.” David looked at the group, forcing himself to stay upbeat and chipper even though he feared that the group opinion could easily go against him and the project at this point. “Is that manipulative? Do we take a Myers-Briggs workshop to manipulate people, or to be able to work effectively with them, and spend less time in arguments and disagreements?”