The Evolution of AI in Finance

One of the things I really like to be able to do is go track down some of the friends that I’ve made over the years who have grown up with technology and grew up with computers as kids hackers, computer programmers, people who eventually became engineers and just pick apart their experience.

A lot of us had similar experiences and I think there’s a lot that can be learned from that. And sometimes it’s just learning that there are parallel experiences that led us into very technical careers. But but also I think it is important to look for the things that worked like what were the things in our backgrounds that got us good at something? And what is it that turned us into hackers and what is it that made us learn to think differently? And so anyway, today we have Jeremy Bornstein who is one of my all time favorite people.  I had the good fortune of meeting Jeremy training Aikido back in the nineties.

We were training with with Frank Doran, who at the time, was one of the senior Aikido instructors in America. Jeremy had started a company with his brother and another friend of ours in San Francisco called Xigo. Xigo was trying to do in the year 2000, essentially artificial intelligence to trade on the stock market. And and I ended up going to work for Jeremy and though the company, unfortunately didn’t work out and got shut down in the.com bubble, we still had amazing actual technology and actual customers and actual revenue, and we were doing great.

We were a victim of a sock puppet attack ended up having to shut down the company. Jeremy and I became great friends there and have been friends ever since. There’s not a lot of people who are as friendly as he is and with his diverse interests. And so we got to spend a couple of hours talking, I’d say about the first half of this is about our backgrounds, his background growing up with computers, how he got into it, how he learned the things he did. Jeremy had a super interesting career back at Apple in the Advanced Technology Group back in the nineties when there was really interesting things going on there and he invented some cool technology.

In the last half we talk about artificial intelligence, where it came from and where it’s going. Also some of Jeremy’s other interests in addition to Aikido, archery, Japanese and Western style, languages like Japanese, French, Mandarin, Spanish, and Latin.  He’s learned to play the shakuhachi and the didgeridoo.

He’s a guy who’s built massively multiplayer online games, automated trading systems, cryptographic systems and a wide variety of other things. I hope you have a good time listening to two friends have a long chat.

Katia Capprelli, the former Italian race car engineer also joined us for this conversation.

Pablos: I’ve worked on projects that in and of themselves didn’t have important world-changing merit. I built websites for car dealerships and bed and breakfasts in the ‘90s. These weren’t important but the experience of that, and learning about those industries, the tools and all those things led to being able to do other projects. For me, I always steered toward whatever the coolest project I could find at any given moment was, and I worked on that.

Jeremy: A lot of people don’t even do that. On one hand, I don’t understand that. I don’t understand it. Why wouldn’t you? You asked me what my new ideal job was. It turned out to be founding Xigo. When I started that company, I didn’t care about the idea of growing an organization or managing engineers. We didn’t have anyone else to do it. At the end, they didn’t want to do it. I was like, “I’ll do that.” It turns out it was fun and I was good at it. I had so much fun. I began to view it as instead of I had to write all this stuff myself, I could find somebody who could help me do it and then tell them about it. They would do parts of it by themselves. I could look at it and say, “That was better or whatever.” That was a revelation when I found out that I could do that.

For me, I don’t write any code. The closest I’ve come is send an email to somebody and they write some code so it’s easier. A lot of them love that because they don’t know what to build. If you have somebody who can think about what’s the right thing to build at the right time, what fits together, what’s going to sell, and what makes sense.

That’s coming back to exactly why I started. I discovered as I was hiring people for Xigo, they’re all software engineers but some people are good at the beginning parts of a project. Some people are good at the middle parts and at the end. They are different orientations and personalities. I’m much more interesting and useful at the beginning. I can handle the other stuff but it’s not my native competence. If I could get somebody else to do that, I will get somebody else to do that. Give me a room full of people who are looking for something to do and I will make sure that everyone has something to do.

Having done it all helps you to be better and trustworthy as a person. I remember at Xigo, you guys must have started about ‘98 or ‘99. In those days, computers were getting to the point where we could start to do very early machine learning type of situations. The stock market had all this data that affected the market. There was technical data, which means all the trades on the stock market, but then there was also the news. By analyzing the city, you could sometimes figure out if things were going well or going poorly, and this kind of thing.

A hundred percent of the stock trading in those days was done by humans, reading the newspaper, and looking at the stock price in the newspaper like cavemen. This was the first company that I know of that was using computers to analyze all that data. We were doing a couple of things that part of the product was letting you know things went south or things are about to go south or vice versa. The cool thing was we did two things. We built systems that could run arbitrary algorithms against that technical data, which meant that you could say, “Tell me what’s happening with this stock over its 30-day moving average and let me know if it gets too far out of bounds.”

It could watch that in real-time and then alert you. This is before smartphones because this is 2000 when we shipped. We had piles of computers that could read the news. This is an early artificial intelligence project. With 100% of the trading being done by humans, we have computers that could read the news and figure out what company is this news about. Is it good news? Is it bad news? Is it likely to move the market? We weren’t using this to do our own automated trading but we were doing all of the work you needed to be able to do that.

I’m sure that’s at the same time as brokerage houses and were doing this internally. Maybe not with the news reading aspect but with the technical aspect for sure.

I’m talking technical data.

Not necessarily the news or the way we looked at the analyst changes.

One of the tricks I remember we did was we had computers. There were television broadcasts about stock market news. There was a transcription of that for deaf people. We hooked up computers to take the transcriptions of the TV broadcasting, run it through our system so we would know in real-time what was going on. Jeremy, his brother, and this buddy of ours and Lenny started the company. At that time, Xigo, being a prototypical Silicon Valley startup. There were startups for a decade before that, but a lot of the practices at Xigo ended up becoming standard practice for startups here.

The one that sticks out to me was in the year 2000, we were hiring the smartest people on earth. Every company says they’re doing that but we did it. You’re paying them six figures and most companies doing that let everybody leave work for lunch. What Xigo is doing is bring lunch in every day. I went there, “That’s amazing. Lunch is being delivered every day for free.” We had fridges full of Odwalla and stuff. It looks extravagant but the fact is the cheapest hour you get out of people is to buy them lunch.

You’re already paying them $100,000-some a year, buy them lunch and you get an extra free hour out of them every day and if you buy them dinner then you get several more hours for free. The fact is that they’re nerds. All they’re going to do is go home and sit on their computer anyway. You might as well let them hang out at work. Things like that look like bad optics, they look extravagant, people make fun of Silicon Valley. Xigo is prototypically obnoxious. We had a chill room with every video game console you could get. We had a massage therapist coming by. There was yoga and Aikido at work.

What was cool is it made it an environment that you wanted to be at. Another big issue and the barrier is a lot of transient people who moved here for work. People all moved here. Some of them for the gold rush. But some of them for the place where you get to work on these types of technical projects with other smart people, and you can have a community of that. The problem with that is you don’t have friends and family around. So, having a work environment that meets more of those needs is important. Now, we mock Facebook for having volleyball at work and Google for having a ball pit or something. Anyway, I learned a lot at Xigo. Specifically, it’s funny because it goes back to some of the things that you said. Jeremy and I knew each other from Aikido. When he hired me, he said, “We’re looking for generalists.” I’d never heard anybody say before.

I’m like, “I could be a generalist.” That was the first time in my life I had heard anybody explicitly express an interest in generalist. Normally, that would be seen as you’re not good at anything. I went to work at Xigo and I had this interesting role which was program management. It’s managing the software development. We had about 40 or so coders. I don’t know if I ever explained to you, there were two of us doing this job. Me and this girl named Leah, who I believe was a former massage therapist. I’m a computer nerd who has a lot of interest in everything that we’re building all the way down to the nuts and bolts and the 1s and 0s.

I loved it and I love knowing everything about it. I would sit there with engineers. I would ask them what they were up to and how they were doing it, and what their plan was. I would say, “What if we did it this way? What if we did it that way? Could you get squeezed a little more?” I was always trying to pack more into every release because we managed releasing software. Leah knew nothing about coding. She had almost no interest in computers. Not to disparage her at all, she was great. What she had was a clipboard with a giant list of to-do items.

Every one of them had a name and a deadline by it. She would go to every coder on the project and she’d say, “What are you doing next? When are you going to have it done?” She would write that down. The next day she’d come back, “How is it going? Are you going to have it done on time? What do you need? Who do you depend on?” She would follow the to-do list. She was arguably a better project manager than me because that was an effective methodical system, whereas I was introducing a lot of risks. I was like, “We can take the flux capacitor and cram it in here and we run it upside down.” All these kinds of things.

We’re on a two-week release cycle which in those days was aggressive. I love that. I’m like, “Rapid iteration. Let’s release.” We get to release. Has it been through QA? It’s been through QA. Has it been through operations testing? They took a look at it. I’ll be like, “Let’s launch it.” We’d get halfway to launching it. The database had fallen apart and we’re like, “Let’s roll it back.” It turns out almost every time it was my fault because I was trying to cram more code. I loved advancing things. I wanted to push things. We got a lot more stuff out sooner because of that. That lack of risk aversion came from growing up with a computer.

I learned everything by rebooting. I learned everything like crash it and reboot it. It was a safe environment to do that which isn’t true for everything. Biological stuff, there’s no reboot. I’m thankful for Xigo because it was a unique experience for me. I learned a lot. I was never built to replicate that kind of environment again in my career. It was special and unique. That was a company whereas a casualty of the time, even though we weren’t like the other companies in the so-called dot-com bubble, we got shut down because of that frenzy.

We were a victim of the crash in the brokerage bubble. That happened six months or however many months before the general crash. That’s why they stopped paying us. It’s with the alacrity which we needed.

In some sense that’s unrelated.

I don’t know.

My customers were brokerage firms. When I think about Xigo now, I think of it as an early and in some sense successful AI product. We’re using AI to build systems that could trade on the stock market. That was 0% of the trading then, and 75% of trading now at least. It falls off on any markets, at least 75% of trading. Computers are automating that for better or worse. It’s important to think over because that was an expert system, meaning we took news. We had humans read it and tagged the news to use as training data for an expert system, which is a machine learning system that can learn from that data in its own way automatically tag the new news.

Did we have humans tag stuff and have it learn to tag?

We had to make training data. We also bought news that was tagged because there were companies that would tag the news for other humans. That gave us some training data, but some of it we had to tag ourselves because there were things that humans didn’t tag.

We did it for sentiment.

Is it like we move the market? That’s not something that came tagged. We had to do some of that. The point is people get fired up over all these complex-sounding so-called AI systems, machine learning systems now that are using neural networks which sounds tech on, smart and complicated. It’s not but there are ways of doing the exact same thing. Computers are vastly faster now. I remember us building racks of pizza box, Sun SPARC servers or Ultra SPARC, whatever they were at that time. When the iPhone came out and performed an entire rack from those days without exaggeration.

Computers got faster and we got some better algorithms for doing machine learning and things. You can see the results that can drive cars and stuff now. What gets to me is how people are extrapolating from this. It seems to me like it’s one trick and we’ve got faster at that trick. We’ve been able to apply that one trick to a few more things. People are extrapolating from this that it’s going to somehow magically also be able to do every other trick. It’s dumb. It’s irresponsible but Hollywood is using AI as to do boogeyman for stories about the future. They’re all dystopian and it all goes to shit. It’s a science fiction that doesn’t have a grounding reality in my mind.

You’re talking about artificial intelligence in general.

He’s pre-circumspect but did you read Superintelligence?

---

Artificial intelligence is always what you don't have yet.
Share on X

---

I did not.

Superintelligence is a book about this notion of humans creating a machine of any description that’s smarter than humans.

Should I read it?

It’s important to read because it’s the one that’s setting off guys like Stephen Hawking, Elon Musk, and Bill Gates who have all made comments to the effect that artificial intelligence might be a horrible thing. It is a good book. He’s an intellectual writer but everybody is baking this logical leap that is unreasonable. We’re going to get from artificial narrow intelligence. Even if you look at AlphaZero. They went from being able to play chess to being able to play Go and Chess and Tic Tac Toe or something. I forget what the third game is.

The point is AlphaZero is impressive because it can learn to play the game and then beat anyone in a few hours. It can start from zero. It could teach itself the game and figure out the best possible strategies. It’s impressive and it can do this for chess and Go. It could start from zero knowing nothing. In a matter of hours, it can learn how to play the game. It can come up with its own strategies and test them, and then it can beat anyone on earth. It’s awesome. DeepMind, which is the Google group working on this stuff, claims to be working towards an AGI. They want to create one algorithm that can learn anything. It’s got this scary fiction around it but I don’t see how that gets us to the kinds of things people are imagining. In some sense, you worked on a version of artificial intelligence many years ago.

I would not call it artificial intelligence. It’s built out of these relatively simple algorithmic components.

Would you call it machine learning?

Yes.

Would you call anything else that exists now as artificial intelligence?

One thing that I’ve heard people say is that artificial intelligence is always what you don’t have yet. Once you understand something, you can do it and it becomes a known technique that is not or knows anymore. That’s how I feel. Machine learning has gotten great. At least to me, it seems that it’s gotten great. That enables a whole class of problems to be solved that couldn’t be solved as easily for. It’s great that you get to that. That’s different from AGI. You could train a network to solve a particular problem in a particular domain. You could train a million of them, and then you could train the framework to choose between them but that’s clearly not how we work. We work in a related-ish way but not like that. I tend to think that we’re not going to get to AGI by explicitly building a structure that is an expert system like a machine learning system. If we get to AGI, we’re going to throw a bunch of stuff in a box that knows how to compose slightly more complex tasks and then put it in an environment where the end-result is a person in a box. The incentives for each piece are to combine it to produce results like that. I don’t know how to do that.

I don’t know how I feel about putting a person in a box.

That’s an easy way and then eventually, that person runs out of whatever they need to stay in the box. I’m sure people will know more about it. They will think that’s a bad description. Our brains are built out of a bunch of directing systems that we don’t understand. Some of those pieces individually looks simple but it’s compositional in a way we don’t really understand. The existence proof that you can build a thinking robot is that we’re thinking robots. We got built somehow unless you believe that there is some ineffable soul thing, which I don’t.

My career and your career are shaped by that time and space that we grew up in that Apple II moment. My daughter grew up with an iPad. The iPad is like a car with a hood welded shut. She’s not learning how computers work from it. It’s amazing. You can do all kinds of cool stuff. I wonder, what do you think the next frontier is? If you were twelve now, what would pique your interest?

I’m sure computers would still pique great interest if I were twelve. I can’t imagine a new frontier on the level of what you’re talking about. At the same time, computers are enabling those new things all the time. A while back, I had to make corporate bios for our consulting gigs. As a team, we decided what we did was we’d like to investigate areas where new technology developments made new businesses possible. That’s happening all the time.

We’re still at the beginning of figuring out what computers are. I get a lot of questions from teenagers and college kids wondering what they should do with themselves.

You can’t answer those questions.

I’d never do. A lot of them have an interest in computers and they don’t know where to start. They think I know everything which is not true, but you can’t tell them that because I want them to think I know everything. You’re professionally curious. You’re good at getting interested in new things and learning them. If somebody were trying to waste a lot of time as a teenager, what would a good thing to waste their time be?

Whatever they’re into. I can’t imagine investing my soul into something because somebody I thought was smart told me to do it. That’s what kids these days are taught in school. You do something because a smarter person than you said this is what’s it going to be. The medical field or anything like that or smart adults say that that is where you should put your time and energy into if you want a successful future. You talk about all the time how we can’t even fathom what jobs that are going to look when your daughter is an adult. We don’t even know what they are going to be. In college, I got a lot of talks from my parents about you can be a doctor or lawyer, top of the class all the time.

Now it’s programmer or a doctor. Nobody wants their kid to be a lawyer anymore. It’s all computer programs.

You guys may have also seen people studying something because of that. I saw a lot of people taking computer science and they were like, “This stuff sucks.” I’m like, “Why are you doing it?” “To make a lot of money.”

I don’t buy that. You and I didn’t do that anyway. Nobody made me so I stuck with whatever I was curious about and interested in, and that worked out. I like your answer. The last thing I want to ask you is if you have any questions for me. These are all things I never got to ask you before.

What was the approximate cause for you leaving Alaska for the first time?

I had full starts because I tried to leave Alaska and I missed my friends so I moved back. When I got back, I realized half of my friends are already gone too so it wasn’t the same. It’s hard when you leave home for the first time because you have heavily invested in it. I thought it was important. I think it is for lots of people but making that stick was hard. I went back to Alaska and I tried to make things work there. I started companies up there.

Was that at Fort Knox, your first one?

Before that, I started a web application development company in 1995. Nobody had even seen webpages yet. I was showing them, “Webpages are home, slow and boring.” What I want to do is build applications that run on a server. I was trying to sell local businesses in Alaska on this and they hadn’t heard it.

They still don’t want to do it.

I inadvertently built one of the first web application development companies in the world. It was way too soon. One of the things I built was this website called AnchorageMovies.com. The reason is in those days, you had to call every movie theater on the phone and listened to a recording. They’d be like, “Top Gun is playing at 5:15, 6:45 and 7:50. Rated R starring Tom Cruise.” It would go on and you listen to it. You’re like, “I’ve heard Top Gun.” You wait and there are six movie screens. You’re listening to this recording. You had to listen to it sequentially to find out what time the movies were.

You had to call the other theaters and do it. You don’t have cell phones. You’re stuck in the kitchen with the phone while your mom is saying, “Give me the phone.” It was bad or you had to get the newspaper and look up the movie times in this case. I thought, “I’ll get a newspaper every Thursday when the new movie times come out and I’ll pay a teenager to type them into a web app and then there will be a webpage that shows what time all the movies are playing. You can sort by theater or movie and see where it goes. It will be amazing,” and it was.

Everybody in town who had internet used that web page. We made the logo with a Sharpie. The webpage was popular with people in Anchorage who were on the internet which is seventeen people. I realized we found this company somehow related to the newspaper ads business or something. They were a company in Colorado that had the movie times that we could buy from them. I didn’t have to pay a teenager anymore. I could get them via FTP. I updated my software, I download the movie times via FTP, and then put them on the page. No big deal. It’s the simplest web page.

I used this app to test new software development environments and programming languages and stuff. I rewrite the app. I could do it in 30 minutes with every new language. In those days, PHP didn’t exist yet. We were writing web pages and pearl and horrible stuff. What happened is I told my business partner, “Look at this. We can get the movie times for all the movies at all the theaters in America via FTP. Why don’t we make it Movies or MovieTimes.com or whatever.” My business partner is a great guy who had inherited money from his dad who ran a dry cleaning company.

He was excited about computers, but his entire worldview and his business were local. He could imagine a business in Alaska but imagining a nationwide business. Not to disparage him, that’s not how people thought in those days. Making a nationwide business was an unfathomable feat. There’s no extra work. We click a few buttons and it’s a nationwide movie. He couldn’t get behind that. I realized at that moment, I got to get out of this town. This is not where I’m going to find the people I need to be working with. I wanted to take computers and go as far and as fast as we could with them. That was the moment I realized I’ve got to get out of town. I tried one more desperate attempt to stay in Alaska.

In those days, every city had an ISP that was a couple of dozen phone lines with modems and that’s how you got online. A lot of times, there was some nerd who started it in his garage. The internet took off so all these nerds got rich running local ISP. This guy had sold his company and he was the local nerd hero because he built this ISP from the ground up. He had made millions of dollars and sold it to the telephone company. Every ISP has this story. He was a cool guy unlike the other internet guy in town who got it. I teamed up with him to start his next company that was called Fort Knox.

We were trying to do cryptocurrency in 1998. We were trying to do transaction systems on the internet. We were trying to solve problems of building resilience because you could see in those days if this internet thing keeps growing, a lot of people are going to rely on it, and it’s totally not architected for that. We were trying to figure out how do you build systems that are secure and stable amidst failures because the internet was constantly going up and down and things like that. In those days, it was less reliable. We tried to both invent systems for handling geographic failover. If your computer died, there would be one on another coast that was still running. All of this is solved these days. In those days, it was pioneering work.

We’ve created systems to defend against DoS attacks, which DDoS had barely been invented and we were working on that. The biggest one I was into was cryptocurrency. We were trying to create ways of transacting. We got excited about what was technically possible but there weren’t enough people online. There was no demand for it. Every project I worked on was ten years too early. Web application developments was ten years too early. That took off. Cryptocurrency was at least ten years too early. That is going somewhere now. Xigo is ten years too early or maybe three months too early. That’s another one way of looking at it.

OQO, also too early. A lot of the miniaturization work we did is what makes iPhones possible now. I work in a lab where my job is to be ten years too early. That’s a long way of saying I finally got out of Alaska. I took Fort Knox from Alaska to Bay Area as a way of trying to get it into an ecosystem where it could thrive. Unfortunately, that company didn’t take off. I learned a lot doing it. Of all those companies, I think of Fort Knox and Xigo as my education.

You guys spent millions of dollars on my education. That’s what it comes up to. Looking back, we were like junior high kids. We didn’t know what we didn’t know. It’s embarrassing how ignorant we were those days, the whole industry was. It’s not fair but things are sophisticated now. In those days, we had to learn everything by hand. You had to build every computer by hand. It was hard work. Now you can start a company overnight, but in those days, everything was work.

It’s one of the things I’ve been amazed with my company. I remember in Xigo, we spent $8 million or however many million dollars on data center because we had to. My current company does not need to do this. I have as many computers as I want. I practically live in it. It makes me realize that’s the IoT Amazon should sell. They should sell a knob that you can get out in one hour. It’s more computer knobs. You make it and then turn the knob and it fires up easy two instances for you. That would be awesome. Who wouldn’t buy that knob? It just hooked up to your Amazon account. That’s a Kickstarter project. They work on computers knob. The more storage knob.

We can build an off by computers. What that knob does is it does fire up in easy two instances. That does something like calculating to a million over and over again. It’s totally useless.

They play Tic Tac Toe until you tell them to do something else.

Growing up in Alaska was good because boredom breeds curiosity. It gives you a chance to think freely. You got time. There are no distractions. My kids got a zillion exciting things going on all the time. I didn’t have that. I had dirt and then I get an Apple II and some Legos. I put dirt in the aperture then reboot after that. I’m not exaggerating by much. You had beautiful mountains, trees, fish and stuff that you take a lot for granted. It was a relatively distraction-free place and not a lot of things are vying for my attention. I got to focus on whatever was interesting to me and that happened to be the computer.

How did that change once you came to Silicon Valley?

I loved it.

How long after you moved out?

Immediately. I showed up at Aikido. That was one of the things that was exciting to me. The Aikido school I was at in Alaska had a satellite of Aikido West or Aikido lineage or whatever. Their instructor maps back to a student of Frank Doran where they’re cutting that family. Aikido West was a mother ship to my dojo. I was stoked to be going there. The cool thing about the Bay Area is I’d spent my whole life pulling teeth, dragging all my friends into these crazy computer schemes that I had. I literally took guys who had no interest in computers.

They were my buddies from hanging out. One of them was my skateboarding buddy in junior high school. I got him to buy a computer, then there were two guys. He wasn’t interested in computers. I was so enthusiastic that he couldn’t help get into it. He got a cooler one than me, so then I’d go to his house. I started companies with these guys who were my snowboarding buddies. They were great friends. They didn’t have any actual interest in computers. Now, they’re the heads of industry in Alaska in IT and stuff like that because that’s what their career ended up being because I coerced them into computer stuff.

It seemed normal to me because nobody else was into it. I had to create enthusiasm around computers and get people into it. It was pulling teeth. By the time I get to the Bay Area, a great thing here is that everyone is in on it. I’ve said bombastic things about this before, but I can go flag down a homeless guy on the street in San Francisco and started telling him my ideas for using 3D printers to print food and he’ll be like, “Let me introduce you to my buddy whose cousin is a VC.” Everybody will give you the benefit of the doubt. No matter how crazy your idea is, they will try to take it seriously and help you. They’ll try to shoot holes in it but in a constructive way.

That kind of community doesn’t exist anywhere else on earth. That’s what’s special here. There are only two things. We got computers and the superpowers that come with it are essentially bottomless, and we’ve got this community that will give you the benefit of the doubt. It’s those two things that make Silicon Valley unique. I stick on myself spiritually, this is the home for me, the Silicon Valley of the ‘80s is. I live in Seattle. Seattle is not home to me in that way. There’s this mythical Silicon Valley that lives in my mind and its components are those two things. It’s hard to find it now, some that’s even being here. We’re in San Francisco and I felt this some even in 2001 at Xigo. There are a lot of opportunists. There are a lot of gold diggers who are just here for the money because this ended up being successful financially and making all these big companies.

You’ve got a lot of people whose parents are like, “Given them computer program because it would be a good job.” They went to college and got that Computer Science degree. They ended up here, working and building enterprise software. They’re building iPhone apps and they’re missing it. They think they’re in it. They think they’re in a tech. That’s not tech, that’s software. The software industry is fine. Those are okay. Some of them are good businesses. They are products that maybe people need. That’s all fine. I’m not disparaging it but it’s not technology. If you want to move the needle on making the world better, you’ve got to invent a new technology and bring it into the world because that’s what moves the needle. That’s what’s happened a bunch of times here.

That’s why we still call it the tech industry. Most of what’s going on is not that. Most of what people are doing here isn’t that. It’s frustrating for me to be here a lot of times because of that noise. I’m cheating. I’ve been lucky because in my career, I got to know many bright and interesting people. I get to travel all over the world to track them down. I can create a community of those kinds of people myself regardless of where they are, and what any one city is doing. I don’t think that’s good. It’s important for everybody and for more people to have that. It scares me a lot to see Silicon Valley get taken over by the wrong people with the wrong mindset.

Do you think the physical location is as imperative as it was?

I’ve been thinking about this a lot because it was imperative then. For certain kinds of things, it’s still imperative. There are still some things you could only do here. In Seattle, we have a lot going on. In Seattle, there’s more going on with cloud services than anything else because we have AWS. Google Cloud people are there, things like that. Microsoft with Azure and stuff which is surprisingly successful is also there. That’s huge and that’s essentially the future of where computing is going and stuff. There are a bunch of applications for that. Seattle is high tech. I would never choose Seattle as a place to create a crazy high-tech startup.

You can do rational things there and you can get a lot of work done. It’s a great place to work. They have hobbies, pets, strollers and shit. It’s a good place to live but they go home on weekends. They’re not as work-focused, there’s no party in, whatsoever. The Berry is constant party, constant networking, constant work, and everything all the time. I like that frenzy and that energy level. I come here, walk in a bar and three hot chicks are inviting me to their cryptocurrency bender. That happens in this city. It won’t happen anywhere else.

I’ve been traveling all over the world to places that are the Silicon Valley of Latin America, the Silicon Valley of Europe, or the Silicon Valley whatever. There’s only one time it’s been true and that’s Tel Aviv. Tel Aviv makes Silicon Valley look slow, boring, old and conservative. It’s unbelievable. They have all the right elements. They have creative but highly trained, intelligent people who have technical experience, technical training in college. They have this rabid startup community that’s a random startup generator. They’re the last ones to join social media because they were social networking. They’re social network like Israelis are all related to each other.

Everything to do in Israel is a family affair. Somebody knows somebody who’s going to help you out in an instant. It’s amazing what they can do. They suck at some things like there’s no market there. If they have to make it work, they have to bring them here or to Europe. That I’ve been inspired and impressed by. In Italy, if you have a startup and it fails, they open a criminal investigation. You’re going to get one chance at life and then that’s it. You’re not going to get another one.

That’s what happens in old cultures all over Europe. You have this case or this situation where the risk aversion is extreme. People know all the reasons why you can’t do anything. You’re arguing with 1,000 years of history everywhere you go. We don’t have that here. We don’t have history. We have a future. It’s a different thing. I’ve been thinking about it. All you need is a small community of people who will give you the benefit of the doubt. You don’t need a whole city or industry. Maybe it’s 100 people. You need to be able to go out to lunch or go out to dinner and tell somebody your idea and have them say, “What about this? What about that? Maybe we can help.”

I’m starting to think some of these places like France is. They’re going to be the Silicon Valley of Europe. The reason is because of Brexit, a whole bunch of companies are moving into France from England. All the banks and stuff are starting to ramp up France and ramp down England. There’s this economic excitement in France. They think they’re going to be doing high tech startups and stuff there. They’re all excited about it. They have a little venture scene and stuff. I was thinking, it might work because they got a critical mass. A couple of hundred people is all it takes to build that community.

If you structure it right, they’re hanging out together and they’re supporting each other. In the Berry, you see it all the time. It’s almost invisible unless you’re in it. Go to something like Y Combinator or whatever, these guys will all show up two guys with an idea and it’s like, “Those guys have a better idea,” and they joined forces. That’s the community thing. Y Combinator is a manufactured thing. It happens to be here and there are a lot of amenities you get because of that. You could do that in Paris, maybe. As long as you keep out the old guys who would tell you why it won’t work, you might have something. The problem is they have a lot of catch up to do, especially when it comes to high tech. That’s going to work for all your iPhone to apps startups and things but they have a lot of catch up to do to get to the point where they’re advancing new technologies.

---

It's not easy to find people who care about the company in the way that you want them to.
Share on X

---

That’s why I like Israel because they still have research, invention, culture and academia. The whole country is invented. They had to invent ways to grow food in a desert to feed ten million people. They’re doing it so there’s an invention culture there. China is interesting because they had a lot of economic activity over the last couple of decades. They’ve changed the incentives around startups and make it coming. There are a lot of entrepreneurial activity. They’ve done an amazing job in the scientific community.

China publishes more scientific research than the rest of the world combined. The quality is not as high on average but it’s getting there. It would be a few years at the most before we’re all blown passed up by China. That’s an amazing thing. That’s a wonderful thing in China. If they keep that up and then they could figure out how to translate that into high tech startups, there could be some cool stuff. It’s hard for me to understand what’s going on there. Have you got a sense of what’s happening in Melbourne? Did you pay attention?

I’m paying a small amount of attention on someone’s attention. The work culture is different. By default, it’s not as easy to find people who care about the company in the way that you want them to be. With that said, I have not hired anybody in Melbourne. I’m involved in a peripheral way with some startups. I’m involved with the thing called Startmate which is an incubator/accelerator program that is run from Sydney and Melbourne for a certain number of weeks each year with a different cohort of new companies in Sydney and Melbourne. It’s a twelve-week program.

In the beginning, what they did was a bunch of stuff to help the new founders figure out what their company does. They get help from experienced people to help them figure that stuff out, how to help them solve problems. At the end, they went to Silicon Valley for a couple of weeks to do pitches. What they’ve started doing over the last year or so is they do San Francisco at the beginning because there was such a cultural revolution that needed to happen for many people. That’s how we need to approach this if we’re going to play in the same place as the companies that come out of here. A lot of that was about the work culture, and how you approach thinking about the company that you’re building.

I wonder if that’s the thing where you have to cherry-pick the right people.

Definitely, you have to pick the people who are likely to be able to be mentored. I would say some people have direct competition with them. They don’t want to learn anything from anybody. They don’t want to learn that thing they need to learn because that deficit is somehow built into their personality or their outlook or something like that. You want to pick the people who you think are teachable, who have some great things going for them, possibly not even including the idea of the company. The basic outlook of this person can succeed in doing this thing, but everyone needs to learn something in some partially unknown priority order. How can we help that happen?

Usually, I try to have a couple of companies that I’m an advisor for. All of the successful situations for me are when I have a good working relationship with the founder, whoever I’m advising. My input is valued that it affects what they do, they’re feeling supported. In other cases, you know that the dynamic is not worth doing at all. There’s no point.

There’s a selection at various levels because there are people who are not even going to bother applying. Some are not going to move forward and some are clearly going to move forward no matter what.

If you had to imagine stacking any of those startups up against Silicon Valley startups.

They have a different set of things to learn.

Now is the time when we talk about learning Japanese.

Over the past few years, I’ve been teaching myself French. My French is not decent.

French is difficult to learn from people that learn English first from what I hear because of the palette and the tonality of it. Did you struggle with that?

Ask my girlfriend who was lovingly accepting my flaws trying to speak French in a way that was helpful to me and painful to her but hopefully, she can have a conversation with me without cringing. When I learned Japanese, it was from the beginning Japanese textbook. There was a language lab I can go to listen to dialogues.

Are you self-taught in Japanese?

Not for Japanese. The point that I was going to make is that all the vocabulary study that I did was looking at lists of vocabulary and trying to remember crap which sucks. I did not do that for French. For French, I started reading novels. I understood enough for the grammar to be able to guess at most things or figuring out most things. I started reading books in French. When I came to a word I didn’t know, I would make a flashcard for that word, possibly with the context of the sentence that it was in. I would look up in the dictionary and they get the dictionary entry in French, so everything all in French until I had accumulated a bunch of these flashcards. The thing is that’s new when I started learning Japanese. We now have this technology called spaced repetition software. The psychological principle for this is that when you make a memory, that memory is accessible to you at first, and then gradually will decay over time until it’s harder and harder to retrieve that memory.

Somebody figured out that if you reinforce that memory just as it’s about to become inaccessible, then it’s a strong reinforcement. Learning looks like you learned something and then it decays. It’s about to go learn it again or you remind yourself, and then this time it’s reinforced to a higher level and then it decays even slower. Over time, if you reinforce it at the exact right time, you can learn some of the few repetitions. This is the insight that spaced repetition software is based on. On my phone, I have a program called Anki which has thousands of French words that I’ve put in. When I’m bored or have a time in the day, I can go and do this.

It’s pouring vocabulary into your head the easiest possible way. I had been thinking that I need to refresh my Japanese vocabulary and it will work with recognition in terms if you see a Kanji, also production, because what you do in that case is it will tell you the reading of the character or the word, you’re supposed to imagine writing it and rewrite it. You then hit the button like, “Show me the answer,” and it will show you. You can say I did that right or I did it wrong. If you did it wrong, the computer knows however that you want to reinforce that memory. If you said okay then it says okay. I know how long I’m going to let you see it for another four months or something like that. All of these things were together.

That would work well for our language. My Italian, I never use it. I never use my Japanese anymore. To have every so often refresh, that would make much sense for a language. Having this conversation when we were in Prague, I was talking to someone about language. Everyone assumes that because I came here to pass the age of eleven, that Italian should be solid in my head, but it’s not. That’s not how it works for me. If I don’t use it, I lose it. A lot of people seem to believe this linguistic idea that if you learn a language and you’re fluent in it up until age eleven that you’ll have it forever. The basics, that’s true but my vocabulary is upsetting. It’s appallingly bad.

Here’s what I recommended, switch your phone to Italian. Whatever news you’re reading to, which by the way, you shouldn’t read news because it sucks. If someone says, “Did you hear about this or that?” If you want to read something, look for the Italian version and read it.

For a long time, I read manga in Japanese and that’s all I read for three years. I was interested in that. There are many different types and it’s easy to find a niche thing. It’s a graphic novel. If you don’t get a word, you still get what’s happened. For years, my Japanese was better than my English or Italian because I was reading so much novels in manga. Novels in Italian are horrible in my opinion. All of the novels that were available to me in Italian were not the ones that I wanted to read or are interested in. I’m one of those Europeans that refuses on-premise to switch my units over. I’m forcing him to learn Celsius.

I switched when I moved to Australia.

I pretend that I don’t know Fahrenheit to force other people.

For both French and for units, I made it a point to not think about the correspondence.

It feels like it’s 12 or 13. I don’t try to convert. I go by what it feels like. Sometimes, they correlate. I know that 30 is about the high 80s or about 90. I know that because I know them both by feel and you can correlate it, but I don’t know how to do the conversion because that’s not useful to me. Language is interesting to me because I have many bits and pieces of different ones. I remember vividly certain things from each language. I would say in Italian the best thing that I know is logistical things like how to explain something to someone or how to give directions. In Japanese, I know conversational things like how to describe things.

Your English teacher is giving you a bad grade on the paper.

It’s just dumb things like I told you I read so much manga so I know many ways to describe battle.

I was thinking about the school class stories where it’s about what happens at school.

How to talk about the cute boy in your class. There are things that I vividly remember. I think it’s just repetition like in Japanese, I had a ton of friends in school that were Japanese. We had a lot of international students. In Italian, it was mostly logistical. All of that repetition, that part of the language did it for me. I know food words in Farsi and that’s it. I used to be decent at Farsi and now I know only food words.

How did you start learning Farsi?

I spent a long time in a relationship with a guy from Iran. His mother became a second mother to me. She and I still are close. At some point, it was inconvenient for me to not know Farsi. I went to Tehran on three different occasions and I loved it. It was inconvenient to go and not know Farsi. Farsi is one of those languages that’s impossible to learn in writing if you are not familiar with Arabic. Everything looks the same to the untrained eye. All of the characters that are small differences in characters that make them completely mean something different. We scrapped the idea from the get-go that I would ever learn how to write it. I learned how to speak. I got good at doing the regular day-to-day parts of language like who I am, where I’m from, where I’m going, what I’m doing, intentional based things, logistical stuff.

When I left Iran, all of it disappeared. There’s no reason to reinforce it and it wasn’t usable. It wasn’t interesting and it wasn’t something that my brain even thought about. It was almost the mechanism completely shut itself off after I left Iran because it was three months after I left Iran and I tried to speak. I went into a Farsian store to get bread but I tried to speak and none of it was there. It was almost as if I left the country and my brain was like, “No, we’re done. We did what we needed to do at the moment and now it’s over.” I thought that it was interesting because I don’t think that it had ever happened to me with a different language. I’ve never had that experience where I leave a place and the language stops. That’s weird. I think language is an interesting thing.

I’m fascinated.

I want to do some self-experiments on this.

Is it the hardest part to maintain vocabulary?

In Italian, that’s the hardest part for me. I’ve told you this before. My comprehension of Italian, French and Spanish is good. The problem that I have is that I understand the mechanism of putting sentences together, but it’s hard for me to find the words that properly convey because Italian is complex. There are a lot of different words that convey similar things. We have specific words for specific things. In the Italian language, there are seventeen different ways to describe a way that you love somebody, for example. When you’re speaking a language of French, Italian or Spanish, there are many ways to describe things. It becomes convoluted when you’re trying to figure out how you’re going to string a sentence together. I’ve lost many of those nuanced and highly descriptive terms for things.

I’ll hear somebody say it, I’ll know what it means, and I’ll know exactly what they’re going for, but being able to string that sentence together in a nuanced way that makes sense is difficult. In Italian, that’s the challenge that I faced. In Japanese, I found that my vocabulary is quite good still in certain things. I could have a great conversation with somebody about the weather. I can have a great conversation with somebody about a book or something like that. Being able to delve into different subjects is incomprehensible for me because I don’t know those terms. The Japanese is one of those languages that has a fast way to describe everything. I was talking about how in Latin-based languages there are seventeen different ways to describe everything. Each of them is equally nuanced and mean somewhat different things. It’s difficult because at some point, you lose all of that. I’m sure learning French was difficult to wrap your head around how many ways you can describe that you love somebody.

There are many things where the nuance is something that I have to accumulate by constant exposure.

That’s great that you have a native-speaking girlfriend. There’s no better way to learn the language than completely immersing yourself with somebody.

The funny thing for us is that I do all sorts of reading in French like novels. The thing that I don’t learn from the reading is always the social register of the words. When I’m speaking at home, sometimes I’ll say something and then Leslie will say, “Don’t say that.” She tried to explain to me how it sounds when I say it. It sounds like you’re a rich white boy trying to sound like a dock worker and making fun of the dock worker.

That’s exactly it with me in Farsi. Your standing socially dictates how you describe that. There was a term that I picked up online that my ex could have said easily and it would have been funny, but if I said it, it was condescending because I’m white and female and whatever. That’s another thing. Slang is impossible to learn if you don’t have a native speaker. It becomes one of the most important parts of your lexicon in any language, being able to understand the regional slang of any place. In Italy, it changes all the time. In any language, I’m sure it does but having somebody to teach you what people that are saying, otherwise you sound like you came out from 1876 giving formal address. That was how it sounded when I first learned Japanese when I would talk to people overly. They are like, “Where are you from?”

We can wrap this up. Thank you.

Thank you, Pablos.

I hope that people get something out of it. I did.

It was fun.

Important Links:

• Jeremy Bornstein
• Superintelligence
• DeepMind
• Startmate
• Anki

Mentions:

• Chris DeSalvo
• Danfuzz Bornstein
• Lenny Raymond
• Leah Kennedy

About Jeremy Bornstein

Experienced engineering executive and technologist. Proven builder and leader of highly effective teams. Demonstrated ability to recruit, motivate, and inspire technical staff. Builder of engineering organizations which collaborate effectively with the entire business.

Specialties: Distributed system architecture, systems analysis, mentoring, debugging, interviewing

Recorded on April 5, 2018

The post Building the Right Thing at the Right Time – Jeremy Bornstein appeared first on .