Following in his sister’s footsteps, Greg Bolcer attended UCI for his bachelor’s degree — lured by her description of “living on the sand on Balboa Island.” In fact, the coast helped Bolcer financially support himself, as the Fullerton native worked as a lifeguard for the city of Laguna while earning his degree in Information and Computer Sciences (ICS) at UCI. After graduating in 1989, Bolcer traded his “beachfront” office for one at UCI, working for Professor Richard Taylor as a DARPA-funded programmer in ICS. He continued working full time at UCI while earning his master’s degree in computer science and software engineering from USC and his Ph.D. in ICS from UCI.
Despite trading lifeguarding for programming, it seems Bolcer never stopped searching the horizon. Since earning his doctorate in 1998, he has founded three startups, the first of which — Endeavors — was based on technology he developed at UCI. He sold the company in 2000, and five years later, he and fellow ICS alumnus Clay Cover launched another startup, Encryptanet, based on their development of the world’s first blockchain. The company’s Paycloud service became the first and only PayPal micropayments partner. In 2008, Bolcer founded his third startup, Kerosene and a Match, which focused on visual search discovery and indexing using general-purpose GPU computing. These days, the recent ICS Hall of Fame inductee is the Chief Data Officer at Bitvore, where his day-to-day work involves a lot of AI. Bitvore processes and analyzes business news and alternative data to perform tasks such as predictive analytics and financial surveillance.
Here, Bolcer talks about his journey from a lifeguard tower to the executive suite, reminiscing about the first AI program he ever built and sharing his thoughts on whether AI is going to take over the world.
What first led you to study ICS at UCI?
My sister had been at UCI her first year as a biology major. I wasn’t exactly sure what I wanted to do other than major in some type of STEM field, but my sister said one of the best parts of going to UCI was living on the sand on Balboa Island. When we were kids, our parents let us take our towels, boogie boards and a brown-paper-bagged lunch on the bus from Fullerton down Harbor Boulevard to Balboa Island by ourselves. We used to hang out all day and then would always catch the very last bus back home. My other choices were commuting to USC, the Air Force Academy in Colorado, or commuting to Cal Poly in Pomona. I chose UCI and computer sciences. My Godfather, Bill Lochrie, who was the first Ph.D. out of engineering at UCI, told me it was an excellent choice.
You ended up at USC for your master’s degree but returned to UCI for your Ph.D. Why?
The decision to go to UCI for my Ph.D. was easy. I had been working as a DARPA programmer for ICS Professor Richard “Dick” Taylor while also working on my M.S. full time at USC, which involved early mornings, late nights, and lots of freeway driving. Dick liked the work I had done in taking over one of his previous Ph.D. student’s projects, so he took a chance on me. Most Ph.D. programs take 5 years or sometimes even double that. He arranged to have my USC coursework from my M.S. credited to the UCI program and promised to get me out fast. That — in addition to the interesting work being done on the WWW, scalable systems, and software architectures — was all that was needed to convince me that UCI was the right place.
How did you first get into AI?
My very first AI program was a solution to the 16 puzzle problem for an undergrad class at UCI. The puzzle is a square of numbered tiles in a 4×4 grid from 1-15, with a blank tile that lets you slide one tile around until you arrange the tiles in numbered order. The AI program would generate all of the possible moves, calculate how far from the solution it was, and then figure out the next best move. The first part of the assignment was to figure out how many moves to generate, how to write the best evaluation function to know you were on the right track, or to figure out a strategy for what the best next move would be. The second part of the assignment was to determine if your solution could scale to 64- 128- or 1024-tile puzzles and if it worked for all random starting points.
I used to run different strategies on each configuration and then would wait hours or days for the AI program to find a solution. I had it configured such that when it did, it would print out the series of moves. I spent a lot of time waiting to hear the printer warm up! Sometimes, when I had to wait for days, I would start to wonder if I had programmed it wrong, had a flaw in my logic, or if it was just a hard configuration. I went back a few years ago to re-run all the code, and most solutions finished in under a second. One configuration that I had stopped trying to solve after more than a week completed in about 12 seconds on a modern laptop!
One time, during an all-nighter in the computer lab, an engineering student asked me what I was working on. I told him “artificial intelligence” and learning. He asked a really good question that’s stuck with me my whole career: “At what grade level does your system learn — the equivalent of someone in preschool, second grade, or high school?” I laughed. I told him it was still learning like a muscle neuron in a dissected frog — if you poke it the right way, it’ll trigger! He had the right view though. To paraphrase a friend of mine: once we fully understand learning, we’ll stop labeling it as transfer, incremental, reinforcement, supervised, unsupervised, or even deep; we’ll just call it learning.
How does AI fit into your current work, and do you see it drastically changing how we do business?
What I do now is mostly applied AI and not the more science-fiction type of artificial general intelligence (AGI). Most classes of problems that can be solved using AI employ an AI solution that I mockingly refer to as really clever programming tricks and techniques. These tricks and techniques are applied to a random subset of data as a stopgap before some more powerful computer and algorithm comes along that can use brute-force computing across all combinations of the data to get to the same solution.
AI definitely has its applicable uses where traditional computing and algorithms struggle, but the idea that AI will take over the world any time soon is a little far-fetched. People first thought that AI and AI-powered robots were going to take over people’s jobs. There are some dangerous automation jobs where it might be safer to use robots. Likewise, for knowledge workers, there are a few repetitive jobs that might eventually become outdated. In general though, there isn’t going to be some huge displacement of workers.
What I’ve found in our industry — selling software AI solutions and datasets into the financial markets — is that our customers are solving problems in a way that was never practical before. Instead of replacing people, companies are using the people they already have to tractably understand at a deeper level the information already out there for competitive advantage. It would take hiring hundreds, thousands, or even tens of thousands of people to be able to get to the same level of understanding — something no sane company would even attempt to do.
Businesses utilize AI to uncover insights about their data and information that has been there all along, but nobody knew how to take advantage of it or scale it up. One of my colleagues likes to describe it as giving everyone an “Iron Man” suit. I think of it as finally being able to understand and exhaustively take advantage of latent information. Instead of a 16-puzzle, where you can generate all the moves needed to get to a solution, it’s a billion-piece chess or Go game with all sorts of really crazy alternate pieces with different rules of their own. It’s that new scaling to larger problems that is driving business.
What has been the key to your success with technology startups?
Startups are difficult. You sacrifice a lot to pursue something you really believe in and are passionate about. The dirty secret about startups is that you can have the best technology, team and idea and still not succeed. My expertise is in early-stage product development and venture-funded companies. What that means in practice is that you have an idea and the ability to prove or validate that idea with someone who might eventually buy a product from you. Furthermore, the approach you take should be unique, and potential users/customers should see value in your specific approach. If you are too early with your solution, you risk running out of time and money; if you are too late, you risk competing against others who have more time and money than you solving the same or a similar problem.
The trick is, no matter how well things are going or how poorly, not to get too enamored with your success or too disappointed in your failures. As high flying and high profile some of my startups have been, not all of them were able to make the transition to the next level. The only key to that is to never lose focus on making progress to your next company milestone and understanding when you can’t. Every company has been a learning experience. The best part of my current company is that I am surrounded by colleagues who have had far bigger successes, and I learn things from them every day.
What are some of the more interesting things you learned while at UCI?
The two biggest things I learned working at the lab had to do with insights into how software scales. One of my colleagues, Ken, had been working on open hypertext systems at the time. The users who wanted the technology kept insisting that a good hypertext system should never show any errors to the user and that all the data should always be well connected. Enforcing that restriction created a barrier for adding and linking content. It turns out that if you allow “broken links” and things to change out from underneath links that someone else created, you remove one of the biggest barriers to building a highly scalable, self-growing system. That was 1991.
The second insight came a year or two later, when one of my officemates, Roy, was tapped to help work on the HTTP protocol standards including 1.0 and 1.1. When the WWW first started, you could count the number of web servers on your hands and toes. As the number grew to 200 and 2,000 and eventually 2 million, Roy was thinking about how big a task it would be to upgrade all 2 million web servers to include the new HTTP 1.1 features. He came to the conclusion that it probably wouldn’t matter because in a year or two, there could be 200 million web servers. How right he was. I think that year he formalized the Apache Software Foundation, whose projects are used in over 95% of all websites and applications, even today. That lead to my research into what the world would look like with 2 billion, or even 200 billion, web servers all collaborating as an Internet of Things. That was 1995. All those collaborations led to three hosted workshops at UCI on internet-scale technologies between 1998 and 2000 on decentralization, internet-scale notifications, and internet-scale naming and namespaces.
Those insights taught me to look at problems and solutions in a different way and to leave open the possibility that sometimes human factors drive the best software solutions. Being part of the lab, workshops, and university/industry collaborations just reinforces that UCI really was a special place to be.
What is your best memory of UCI, and who was your favorite professor?
My best memories are from the software engineering research lab. There were so many interesting people doing research on so many interesting things that just being part of the group felt like we were changing the world. The research was mostly funded by DARPA, the NSF, and a few companies. We collaborated with other universities like the University of Colorado at Boulder, the University of Massachusetts at Amherst, Purdue and several others. We worked with Boeing, Lockheed and Northrup Grumman in addition to various government and military agencies. We constantly interacted with a variety of Silicon Valley and other high-tech startups. This took a lot of travel. I was always first in line to volunteer. Being able to understand how university, government, military, research, and high tech startups work and to seamlessly move between those communities is one of the skills I developed at UCI that has helped me in my career.
My favorite professor is Dick Taylor, not just because he took a chance on me for my Ph.D., but because of one my funniest memories. An officemate, Jim, was working on a standard called WebDAV to make the web more writeable. Dick, Jim, and I all went up to Silicon Valley to visit Netscape right after they had done their initial public offering in August 1995. As we were walking through the parking lot, we admired all the new fancy cars. We all were “oohing and aahing” over one when Dick leaned over to take a closer look, setting off the alarm and causing everyone in the building to look out to see what was going on. We had a really successful meeting and ended up working with them for a couple years after that, but when Jim and I shared the story back at the lab, the story lived on for years to come.
What was your reaction to learning you were being inducted into the ICS Hall of Fame?
Truthfully, I was slightly confused at first. Every year, I have been actively involved in nominating people. I’ve joked with the organizers that they should take some of my nominations more seriously because I have a pretty good track record of nominations that get accepted. When I saw the email, I thought to myself, “I don’t remember nominating anyone this year.” It took a short while for me to realize that it wasn’t one of my nominations that got accepted.
Any words of advice for ICS students?
ICS is truly a special place in the world. There are amazing people doing amazing stuff that is well known all over the world. While other universities may have greater name recognition, UCI in general and Information & Computer Sciences specifically have been quietly doing the things that truly make a difference in people’s lives. ICS has become students’ top choice. Once you are here, find a project and find your passion. Everything else in your career will benefit from that.
— Shani Murray