My Conversation with Sam Altman

Yes, the Sam Altman of Y Combinator and Open AI.  We even got around to Harry Potter, James Bond (and Q), Spiderman, Antarctica, and Napoleon, what is wrong with San Francisco, in addition to venture capital and the hunt for talent.  Here is the transcript and audio.  Here is one excerpt:

ALTMAN: I think our greatest differentiator is not how we identify talent, although I will answer that question, but the fact that we treat our own business — we run Y Combinator in the way that we tell our startups to run as a successful startup, which almost no venture capital firm does.

Almost every venture capital firm gives advice they never follow themselves. They don’t build differentiated products. They are not network-affected businesses. They don’t try to build a brand and a community. And they don’t try to make something that gets better the bigger it gets and have the scale effects that anyone would tell you they want in a business.

We at Y Combinator always say we want to get a lot bigger because this is a network effect, this is a network that matters. Most venture capital firms will say out of one side of their mouth, “Oh no, smaller is better,” because they don’t want to work more. Then they’ll tell all their businesses, “The network effect is the only thing that matters.”

Many people are as smart as we are, think about the world in similar ways. But I think we have internalized that we run our firm the same way we tell our startups to operate, and we view the most important thing that we do is to build a network and a network effect.


COWEN: Let me play venture capital skeptic, and you can talk me back into optimism.

ALTMAN: I might not.

COWEN: Let’s say I say, tech has had a stream of big hits: personal computer, internet, cell phone, mobile. You’ve had a lot of rapidly scalable innovations become possible in a short period of time. We’re now in a slight lull. We’re not sure what the next big thing is or when it will come. Without that next big thing, won’t the current equilibrium require a higher rate of picking the right talent than venture capitalists are, in fact, able to do?

ALTMAN: I will talk you out of that one, happily. The most expensive investing mistake in the world to make is to be a pessimist, and it’s a common one. I think that’s actually the most common mistake to make in life. It is true that we are in a lull right now, but it is absolutely, categorically false that — unless the world gets destroyed in a very short term — that we will not have a bigger technological wave then we’ve ever had before.

COWEN: Why can’t I be an optimist but not an optimist about VC? I think new ideas will come through established companies. They’ll be funded by private equity. They’ll happen in China. But the exact formula where you can afford to make so many mistakes because the hits are so big — to what extent does VC rely on that kind of rapid scalability that may not come back?


COWEN: Young Napoleon shows up. What do you think after 5 minutes?

ALTMAN: How young? Like 18-year-old Napoleon or 5-year-old?

COWEN: Before he’s famous, 21-year-old Napoleon.

ALTMAN: From everything I’ve read that would be a definite yes. In fact, the best book I read last year is called The Mind of Napoleon, which is a book of quotes about his views on everything. Just that thick on Napoleon quotes. Obviously deeply flawed human, but man, impressive.

Definitely recommended.


Seems this guy is sharp, but if he fails to mention PATENT he's a fool. Truth.

You must be a lot of fun at parties, Ray.

With my tremendous family wealth--did you know we're in the top 1%, minimum net worth of $10M?--I'm not just the life of the party when I'm in the Philippines, but, I *am* the party, as I sponsor parties. Like a character out of a F. Scott Fitzgerald book, my job is to ask others how they like the party and voyeuristically observe others enjoying themselves. Emperidor Light! They drink this cheap brandy like water, and it's actually cheaper than some imported bottled water. I'm a beer and red wine man, but recently I've enjoyed the ginger vodka drink by Smirnoff, a sort of "Moscow Mule". Besides alcohol, we also have music (typically about 20-50 people show up), video karaoke, and on occasion people slip into the nearby nipa hut I built and for all I know they have sex in there too. No illegal drugs however. I don't need an excuse for Duterte supporters--or rather people harboring a grudge--to murder us.

You're the Timon to their Lucilius.

Maybe he is the Timon to their Pumbaa.

That makes you Max Goof.

Hi, I think your site might be having browser compatibility issues. When I look at your website in Safari, it looks fine but when opening in Internet Explorer, it has some overlapping. I just wanted to give you a quick heads up! Other then that, fantastic blog!

I am more like Max Power.

It is a lie. There are no homosexual people in Brazil.

That's a laugh, Thiago.

Unless you are a drug company, most startups don't waste their time on patents until they get big. They're not IBM or Qualcomm.

False. That would be a huge mistake.

Altman likes the licensing fees and royalties model for generating income (and achieving scale), which means he relies on intellectual property as the source of income for business. In other words, he likes firms that don't produce the products people use but the intellectual property necessary for the firms that do produce the products we use. It used to be that the overly optimistic became real estate developers, but now they become venture capitalists in tech.

Most venture funded startups don't use licensing nor royalties as their main driver for income. It's all about sale of product or service with the major exception of ad funded businesses.

"And I’ve updated myself to thinking, now, that things can actually get much worse in the US much more quickly than I thought."

But keep arming Red China. What can go wrong?

It wouldn't be nuclear but techno-biological apicomplexa blown over from sub-tropical trade winds.

Who knows? The point is, Red China plots the destruction of the West.

The distribution mechanism, I imagine, would be scrubber systems in Hainan coal fired smock stacks. By outward appearances, they would look like carbon sequestration devices, rather than particulate bio augmentation systems. Living soot turns up in the front range of the Colorado Plateau.

Maybe. We are moving into uncharted waters.

for a change! normally history just follows a chart.

I see Cowen didn't try to defend centaur chess. That idea as a paradigm for human-AI cooperation has not aged well.

"...but it is absolutely, categorically false that — unless the world gets destroyed in a very short term — that we will not have a bigger technological wave then we’ve ever had before"

I think he's right, but I expect the most important innovations to come outside of the SV tech industry going forward. For many decades in the 20th century, the auto and aviation businesses were the most important and innovative. But they (inevitably) matured. It appears to me that the same has already happened in what we think of as the tech industry.

I might have a different opinion if I were more optimistic about AI, but I think that the low-hanging fruit in robotics has already been picked (in factory automation) and expect that in the coming years/decade, AI systems will be able to beat humans in every game ever invented, but still not be able to do your laundry or drive your car in other than limited, ideal circumstances.

Well said. Legacy tech limited to silicon. More fun to be had in biological systems engineering. We needed the big processing power and the shift to speed over precision to drive the pattern analysis, modeling, and scenario execution with distributed networks which happened around language AI the past few years. And new techniques from cancer research. The next wave of innovation will not be in the valley. But iphones... how cute!

"...and expect that in the coming years/decade, AI systems will be able to beat humans in every game ever invented, but still not be able to do your laundry or drive your car in other than limited, ideal circumstances."

Last year, an A.I. research team at Princeton showed that their software beat humans at every game on the internet. "Tech" is not close to being mature.

Cowen: "We’re now in a slight lull. We’re not sure what the next big thing is or when it will come."

What is a "slight lull" since if slight couldn't be seen as a lull. The percentage of American and world users of virtual reality has been growing exponentially at about the same rate as internet users in the early and mid 90s and the VR systems improve every 12 to 18 months.

U.S. internet users
1993... 2%
1994... 5%
1995... 9%
1996.. 16%
1997.. 22%
1998.. 30%

U.S. virtual reality users
2016... 3%
2017... 7%
2018.. 11%
2019.. 16%
2020.. 22% (?)
2021.. 30% (?)

The idea that VR is going to be a widely used (or as broadly useful) as the Internet doesn't pass the giggle test. I don't think it really matters how much VR systems improve -- imagine a pair of googles that is as light-weight, powerful, and high-res as you like with a battery that lasts for days, and the problem remains in finding uses for the technology. The fundamental problem is that your head is the VR environment and your body remains stuck in the real world. Nothing in the VR environment is tactile. You're just waving your hands around in empty space. It may look like there are objects there, but you can never feel, touch or grasp them. And if you get up and move, you're going to be stumbling over things you can't see but that stubbornly persist in existing. The fundamental disconnect between eyes/ears and hands/body is unbridgeable. Given that, what are most people going to do with these near-future VR systems? Sit still and watch Netflix? Are they really going to prefer pretending to be a in a virtual theater over their plain old TV? Or are people going to want to watch 3D entertainment? The (lack of) success of 3D televisions suggests not. My guess is VR users will never be more than a minority of the population, and the numbers may start shrinking once the novelty wears off (and the VR startups start going bust).

I'm similarly skeptical about AR -- what would I really want on my full-time, heads-up display? Would I really rather wear an AR headset than just glance at my phone as needed? I really don't think so. I expect HoloLens to go the way of Google Glass. Or maybe -- maybe -- there'll be legit, practical industrial (or military!) uses this time, but then that means AR remaining a niche product with nothing like the impact of the Internet.

I just wonder if you will ever see a decrease in the need for business travel. I suspect not - I think even as communication technology improves, showing your commitment by taking on the burden of travel to visit your customers is as much the point as anything.

It's even more fundamental than that. Humans are a social species, and the most important things happen when people are face to face, or in groups, physically. This is why people go to the movies and concerts and so on, to be around other humans. To close the deal people want to look you in the eye and shake your hand. That's part of being human (for most).

The internet/video chat was supposed to make travel obsolete, didn't happen. It was also supposed to make economic clusters obsolete. Especially high tech, no need for all the talent and companies to be in high cost Silicon Valley when everyone can work remotely and video chat. Didn't happen that way, people still need to be in the mix, physically, with others.

I disagree. There's still more innovation to be had in Tech. Not just AI applications like self-driving cars or writing news articles but remodeling the finance industry along more innovative ways like micropayments and newer forms of digital money or mass adoption of virtual reality with the bandwidth needed to boot or the rise of quantum computing, which will usher in new ways of modeling chemical and quantum processes which will lead to new materials and medical treatments. I prefer to bet with smart entrepreneurs than against them. Maybe that is why I'm in Silicon Valley.

Tyler and most commenters here have never heard of quantum computing so a tough sell.

So here's a guy who believes he is good at picking people, and he also likes Napoleon, who said "I know he's a good general, but is he lucky?"

Are you seriously second-guessing Napoleon's ability to pick successful marshals of France?

Altman's strategy for 'picking people' is to pick ten and see who makes it work.

Napoleon recognized that luck was more than chance, but the ability to take advantages of opportunities as they arrive, the ability to pay attention to information and intelligence that is coming in and make sense of it and make decisions based on it, and the ability to surround yourself with people who will accomplish what needs doing.

It also meant that they had seen battle and not only survived but prevailed. Far better someone like that than the useless inbred son of an aristocrat.

It's good to see that investors are reporting for spring training.

COWEN: As our final segment, let me ask you a few questions about what I call the Sam Altman production function. Are you game?

I think this was mis-transcribed.

Spent fuel reprocessing is not as sexy as fusion but it solves 2 problems: (i) storage of present waste, and (ii) fuel supply for the next generation of reactors. Today, reprocessing is "uneconomical" when compared to uranium mining.

A better estimate of the costs of nuclear waste storage or CO2 emissions may make spent fuel reprocessing more attractive.

This is a don't care though. If the cost of nuclear waste storage grows enough, then you recycle it. It's only a distraction for people who don't understand that it is a solved problem. Lowering the cost here is not going to change the fundamental picture for energy provision.

" I agree that talent is equally distributed around the world"

The internet is filled with tautologically true statements that are also false. If the statement means intelligence, well we know that's not true. If it means that we all have some biological edge that outcompeted others who didn't, well, all our ancestors survived at least until they reproduced.

But if you want a QB, go to Texas.

If a statement like that has any truth it's with the caveat "...adjusting for observables", than that "talent" (defined by VC criteria) is equally plentiful per km2 of Fishtown, Belmont and Kalahari Desert.

"But if you want a QB, go to Texas."
Or California where Tom Brady is from.

And Aaron Rodgers and Jared Goff and....

That was an embarrassing "Emperor has no clothes" moment as Tyler hurried to throw in his assurance that he, too thought the king's clothes were beautiful.

Altman's YCombinator office is in San Francisco. If talent was equally distributed, those offices would be in all 50 states. But not all states are equally talented. Math and science is really hard. Most flyover states are filled with low IQ people who are only good for unskilled jobs. That's why they are so scared of Mexicans. Californians can upskill into a sexy tech job that's why they don't care if half of Mexico moves to their state.

I'm not sure Sam or Tyler really believes it, but it's a great thing to pretend to believe. Sure, the reason all the founders are upper-middle class Americans or demographically/culturally similar foreigners is that they have loving parents. Note that nobody is actually putting any money down on behalf of this belief.

I bristled when Tyler said this.

To me it sounded like:
Sam -- "I'm a religious fanatic."
Tyler -- "I assure you I too am a religious fanatic."

I enjoyed this. I wanted more discussion and pushback on his UBI ideas, particularly how Professor Cowen has said that people have a need to do work to find value in their communities. (UBI would work for Sam and the driven UMC people he surrounds himself with, but Joe Sixpack?)

But it was hard to think of what other part should have been cut to make room for that.

You rang?

Hah, "From everything I’ve read that would be a definite yes". seems more like a statement of Altman's judgement than a statement of obviousness, clarity of Napoleon's excellence to *any* observer and the ease of talent decision problems. (But, perhaps, would that it t'were so?).

Chateaubriand was pretty good on Napoleon, and on "spotting talent", as they say these days.

There is a book of Napoleon's quotes, full of insight on 'what to do when you have x regiments and the enemy has x regiments and there is a ridge here and a river there', that sort of thing, and the book has pictures.

Camille Paglia says young people should study military history.

That being said, Waterloo was a fiasco, but Napoleon was not feeling well during that battle. And all the books about tactics and strategy at Waterloo were published AFTER the battle *** who can deny that such books would have been worth their weight in gold! or Platinum! if written and published before the battle .....

Finnegans Wake, near the beginning, has a good few pages on Napoleon, verb. sap. sufficit

Starts slow and gets better, imho. The questions from the audience produced really great responses, btw. I often skip them entirely, but I found them a highlight this time.

If Altman doesn’t like the current NIMBY government if SF, why doesn’t he run for office? Or at least find & massively fund someone who has the time to be in public office.

If he feels the NIMBY government represents the preferences of the electorate, he would be correct to assume there is very little possibility of change. You can't just win one election and fix everything. Same problem leftists face with the American electorate. Change is only possible at the margins, generally.

Comments for this post are closed