Big Data

Nick Berry interview — A data scientist confronts his own mortality


Nick Berry has stage IV cancer. He is a first-class data scientist, but even he is not certain how much time he has left.

Without treatment, his doctors told him he has about six months to live. With treatment, including a new experimental cancer drug and chemotherapy, he could live for a couple of years. He would like to see both of his children graduate from high school. But he knows that he only has an 8% chance of living for five more years. Berry has confronted his own mortality, told his family about his chances, and shared it with friends on Facebook.

After all, he is a data scientist. He spent many years at Microsoft and Facebook, and recently he has been consulting as president of DataGenetics.

It’s a sad situation, as Berry is a beloved character in the game business. I met him years ago talking, as he was a regular at the Casual Connect (Now GameDaily Connect) game events. Berry is a very intelligent man, and he had a knack for diving into game analytics and coming up with interesting explanations. He was named one of the 50 over 50 in Kate Edwards’ latest list of the top people in the game business. And he has enjoyed blogging about things like how long it would take to count a trillion dollars (31,688 years).

But cancer is something that triggers strong emotions, and the numbers associated with survivability are unmerciful. A year ago, Berry was walking 12 miles a week and swimming 20 minutes most mornings. Now he may gasp for breath after climbing a flight of stairs. He is undergoing aggressive chemotherapy, and it is helping. But Berry knows that at some point, the cocktail of medicines will eventually poison his body and become unbearable. If anything makes him angry, it’s the bureaucracy of the medical establishment, such as getting permission from an insurance company for a treatment that actually cost nothing, as it was experimental.

Above: Nick Berry speaking at a Casual Connect event.

Image Credit: Nick Berry

Berry is trying his best to survive, and he is also making the best use of his time that he can. He took his daughter to see a show in Las Vegas. With is son, he visited the San Diego Zoo. A week ago, Berry took his family to Disney World. He visited his hometown in England with a friend. And he officiated a wedding for the first time. I was lucky enough to grab some of his time at the GameDaily Connect event in Anaheim. He was an emcee for a day, and then he went home for another round of chemotherapy.

We talked about data science, his love for games, how he is dealing with his cancer, and his legacy. I teared up a few times during our talk. But Berry made it easy to converse. He refers to his body as a “sorry sack of cells,” and he has kept his sense of humor. He was serious and precise with his words, but kept control of his emotions, like a man who has come to terms with his illness and life.

Here’s an edited transcript of our interview.

Above: Nick Berry is all about the data.

Image Credit: Nick Berry

Nick Berry: When I grew up, I was the first kid in my village with a computer. I loved writing games. I taught myself to program. I found out there was a kids’ summer camp where the kids would go for a week—it’s very popular in America, but it was new in the U.K. You’d go to this place and sit there with lots of geeky kids. I was one of the older kids there. I did that for a week, and then come back and teach and mentor. I’d go to school, and then in the summers, I’d teach at this camp, teaching assembly language and BASIC. I used to write little video games like Space Invaders. That’s how I got my start in all of this.

GamesBeat: Did you go right into games, or were you doing computer science in general?

Berry: No, games were the thing I liked most. Again, it’s that Space Invaders time. I was fascinated by it, and I wanted to write those myself. It came time to go to college, and in the arrogance of youth, I said, “I’m not going to university to do computer science. I already know more than these professors. Why do I want to go learn about punch tape?”

GamesBeat: What years were these?

Berry: I was born in 1967. In the U.K. you finish high school at 18. I decided to study rocket science at university. I did a master’s in rocket science and aircraft design. I did an apprenticeship at the ministry of defense designing fighter aircraft cockpit stuff. I was just about to go to another career, and then one of the other supervisors who taught the kids’ summer camps said, “Hey, we’re starting a software company. Come and join us.” Sure, I’ll join that.

It was related to electronic mapping. If you use any of that route software, starting here and finishing here using Dijkstra’s algorithm, we pretty much invented this idea of turn by turn route instruction stuff. It was called NextBase in the U.K., but we sold it as Automap in the U.S. We grew up to about 50-60 people and ended up selling to Microsoft, which was how I moved to the U.S.

Above: Nick Berry in his younger days.

Image Credit: Nick Berry

GamesBeat: Is this in the ‘90s?

Berry: It’d be 25 years ago. I could look it up. I worked on maps for a couple of years. Then I wanted a change at Microsoft. They’d just acquired a company called Electric Village, which was into gaming. The Ed Fries days, if you know Ed Fries. The internet had just been discovered, the superposition of the internet and gaming. It was what become the Microsoft Gaming Zone, where you’d dial online with your modem and play a game of chess or checkers or spades. It would matchmake you with somebody.

I joined the Gaming Zone, and the first question I asked people was, “This is interesting, but what time of day do people play the most games? Do you they play at lunch, in the evening, on the weekend?” “Well, we don’t know.” “Yes you do, you have these servers running this stuff. You should be monitoring this.” “That’s a bloody good idea. Why don’t you do that?” There really wasn’t a data science title at the time, but I started logging, every five minutes, the number of people playing chess and checkers, and I plotted that data on graphs.

I found out what other games online were doing, like Pogo and Yahoo Games, all the rest. I used to write little web crawlers to go through and pull the page every five minutes and count the number of people playing the games. It was fascinating stuff. It turns out that it’s about 7:30, 7:45 on west coast time, the superposition of the east coast and west coast. And then 2:15 in the morning is the time that’s the lowest. When we did server maintenance, we did it at 2:15 in the morning.

Even looking at the curves you could find things out. When the Super Bowl comes on, you’d see certain games drop in population. Other games would increase, because the bored spouse doesn’t want to watch the Super Bowl. They’ll go online and play a game of checkers or chess.

Above: Nick Berry was an emcee at GameDaily Connect 2019.

Image Credit: Dean Takahashi

GamesBeat: And all this came pretty easily to you?

Berry: It all seemed like common sense, which is kind of bad, because—at the time, data science was a relatively new thing. I was working at Facebook doing a lot of recruiting stuff, and people would go in there saying, “What classes should I go to?” And I’d say, “I don’t know. I’ve never been through any formal training. It just seemed natural to me.” I’d come up ways of doing things, and someone would say, “You do know there’s a name for this?” “No, I don’t.” I was just making it up as a went along.

Then I got into scraping the deluxe download games, what the prices were, and I ended up building this database of all those things. I became a subject matter expert in all this sort of gaming stuff.

GamesBeat: I remember Fighter Ace. Was that around?

Berry: Fighter Ace, and there was a space game. I forget what it was. Asheron’s Call, those early pay-to-play games.

GamesBeat: Was it a small group back then? Was it part of the games group?

Berry: It was part of the games group, yes. The Gaming Zone went through many internal name changes. MSN Game Zone, Gaming Zone. They called it Rat internally. Then it split up. The Fighter Ace team and Asheron’s Call team, the actual first-party game studios, moved out and split. It was down to the core casual games team. We did the Millennium games, games based in Windows. Windows played checkers online. That was done through our stuff. We also did the first part of Xbox Live Arcade.

This is probably slightly off record, but—in Greg’s days, we sat there and said, “This casual gaming thing is really popular.” If you asked people if they played video games, the hardcore gamers, as you know, would self-identify. But other people would say, “I don’t play games. I play Candy Crush.” They didn’t self-identify as gamers. When the first Xbox came out, we said they should put casual games on it, and the Xbox team said, “No! This is a core gaming platform.”

We said, “No, really, this is really popular. You find out that people get it home and they’ll want to be able to justify the purchase. You should have this deluxe download-like experience where you can download demos.” They didn’t want to do that. We ended up developing it internally ourselves and pushed it out there. It was a phenomenal success. Then they all said, “Well, we really always wanted to do that.” [laughs]

Again, don’t quote me on these figures, but you get the concept of phenomenal success. In the deluxe download world, PC games like Bejeweled Deluxe and all those candy games, you do the download and go through and wait for an hour and it would fuse. A good game would convert about 1.8 percent. Out of 100 downloads, you’d get one or two downloads that went through and paid to unlock. On the first Live Arcade titles, the Marble Madness and stuff, the conversion rates were like 50 percent. In the first week, it was a phenomenal success, people saying, “This is what we always wanted for Xbox Live Arcade.”

I left Microsoft and spent a year at Real Networks. I ran the analytics team there and built up the tools for that kind of stuff. Then I left there. My parents had been very sick, so I stopped working for a while and looked after them. Then I joined Facebook.

Above: Nick Berry’s most recent birthday photo.

Image Credit: Nick Berry

GamesBeat: Do you remember the year?

Berry: I was at Facebook for five years, and I’ve been gone a year. It’s all on LinkedIn. I joined there and I helped to do data science for games. My role changed to a side lead role to help recruit for Facebook. When I joined Facebook Seattle it was about 200 people, and when I left it was about 2,500. It was about giving back to the community, keeping employees happy, that kind of stuff. More of a mayoral role.

GamesBeat: That must have been a wild ride. Did it feel different from Microsoft?

Berry: Very, very different. I loved the culture of Facebook. I have to give a lot of credit to [names – 10:32] and everybody else. When they set the company up, whether through luck or foresight, they knew the company was going to be larger at some point. They set some very cultural things that are important. When you get to the inertia of a large company, you can’t change the culture, and it’s so important to that kind of company. Even the subtle things, like personal pronouns. How do “we” do this? Not how do “I” do this? We’re on the same team. Part of when you join the company, there’s a boot camp process where the noobs all get together. You meet the execs. You learn the company principles.

If you’re hired as an engineer–I went through the engineering boot camp as part of the data science stuff. You’re almost hired as a blank slate. You’re a rock star. We’re going to give you a job. They recruit all these people. So what are you going to do? Well, you don’t know what you’re going to do, because you don’t know all the things we do at this company. Everybody goes through boot camp. It’s a common experience. You all speak the same vocabulary and do these basic training exercises.

Then you get exposed to different parts of the company. Here’s a PHP bug. Here’s a bit of this. Here’s a bit of that. You get a chance to do these little tasks. About three or four weeks in, this dating starts to happen. “I have to work on billing. Who do I talk to about billing?” “Go talk to Suzy. She’s on the billing team.” “I want to work on this backend stuff.” “Talk to these people.” Naturally, people cluster together very closely. You work where you have a natural talent, where you have a passion, where you have an affinity, where you can move the needle. It works really well. You find the place at the company you’re going to join.

That sort of culture is great. It’s very flat. You have thousands of engineers working at Facebook, and their titles are just “engineer.” You asked about differences at Microsoft. Microsoft has architect, principal engineer, senior engineer, engineer, intern. You go into a meeting at Microsoft and when there’s an argument about what you’re going to do, everyone looks at the VP at the end of the table. It’s complicated.

The data level is all arguments at Facebook, and hand on heart, you can go to anybody at Facebook and say, “Hey, guys, we’ve been doing this all wrong.” “Show me.” “Well, this is how we used to do things, and I think this is the way we should do things.” “You’re right, ship it. Run a test.” Again, two engineers—if you go into a room and say, “The principal architect says to do it this way and the intern says to do it this way,” which would you go with? But if it’s engineer A and engineer B, which is the most performant?

You make a half a percent improvement in anything that’s done at Facebook, you save a million dollars a month in server costs. You’ve shipped love to 2 billion people. In Africa, you pay the data plan by the byte when you’re downloading pictures. You’ve saved money for all of those people. You move fast. Today it was your way. Good job, group hug, we’re all on the same team making the company better. Next time it could be your way.

At Microsoft, it was very vertical, very fiefdom. Get off my turf. Facebook engineers have access to the entire source code of the company. They can make corrections. There’s no way to launch a nuclear missile. You have to go through code review. But you can make a change to something that’s important. I loved that openness, the way you could—you’d be open and respectful of people, but data levels are arguments about what’s the most performant thing.



READ SOURCE

This website uses cookies. By continuing to use this site, you accept our use of cookies.