Arnold Kling

@arnoldkling

Cowen, Andreessen, and Horowitz: Annotated

December 31st 2018

This podcast, in which Tyler Cowen interviews Ben Horowitz and Marc Andreessen, contains a number of interesting observations. In what follows, I recapitulate some of the observations and add my own comments.

The view from 1995

Andreessen and Horowitz first collaborated in 1995, when Horowitz left an established, successful company, Lotus Notes, to work with Andreessen at a high-flying startup, Netscape. Andreessen points out that the big issue dividing the tech world was whether or not the Internet was going to actually solve the problem of connecting the world’s computers. As late as 1995, there were still many major companies that were working on technology that would be valuable if and only if the Internet did not work. That was a bad bet.

Not knowing any better, I made what turned out to be the right bet. I fell in love with Andreessen’s first graphical web browser, Mosaic, and I just assumed that everything else was going to fall into place. So in April of 1994 I quit my job to start a web-based business.

Whatever happened to AOL?

Andreessen and Horowitz explained that they got the idea for their next start-up, LoudCloud, from watching corporate Internet sites drown in the “firehose of traffic” directed at them when AOL linked to these sites. The two entrepreneurs determined that the corporations needed a solution for this.

This reminded me of the power of America Online in those days. It was the Facebook of its day. Why did it fade so quickly?

I believe that the answer is that AOL was optimally positioned to capture the market for people accessing the Internet from their homes over ordinary telephone lines using modems. But as the twenty-first century dawned, there was a pivot toward Internet connections at offices and broadband Internet connections at home. AOL was unable to come up with a compelling value proposition for consumers in that environment.

Facebook also had to contend with a major shift in how people accessed the Internet, as mobile phones took over. Facebook had achieved its early success back when personal computers were the standard technology for accessing the Internet. Had it stayed rooted in the PC, Facebook would today be forgotten. Instead, the company adapted to the mobile environment and became even stronger.

Don’t Replace the Founders?

Andreessen and Horowitz formed their venture capital firm in 2009. They took the view that the best firms were those that retained their founders for a long time. But most venture capital firms operated by replacing founders with veteran executives.

Andreessen and Horowitz make it sound obvious that founders should remain on the job. The founder has developed intuition about the business, including the problems that most need fixing and the opportunities that are most worth exploiting.

But there is another side to the story. As they point out, founders tend to be peculiar individuals. Being willing and eager to break rules may be a feature when you are innovating, but it becomes a bug when you need to develop reliable processes that can allow the business to scale up.

It is probably the case that the vast majority of founders are not capable of leading a major enterprise. They will become a hindrance to the business unless they are replaced.

A Talent Business

It seems to me that the Andreessen-Horowitz approach can only work if they are good at spotting those rare founders who do not need to be replaced as the business scales up. Although the podcast does not explain what their secret is for doing that, they do say that they see themselves in the talent business.

Their model is to maintain a network of outstanding people with the skills needed in the businesses that they fund. I infer that this allows them to ensure that founders do not have to learn everything the hard way. They do not have to crash and burn on their own. Instead, I assume that Andreessen and Horowitz can plug people from their network in as executives, coaches, or temporary help as needed.

Blockchain and Trust

Andreessen and Horowitz are bullish on blockchain. Horowitz sees blockchain as a general-purpose technology for providing trust.

I disagree. In the context of business, trust means that you believe that the other party will do what it has promised. For that belief to be justified, you need credible evidence that the other party has kept similar promises in the past and has the incentive to continue to do so in the future.

I think that trust is too subtle and nuance to be “solved” by blockchain. Instead, we need an intricate interlocking arrangement of institutions to provide trust. We need a government in which officials who don’t take bribes are generally better off than those who do. We need businesses with valuable reputations that they do not want to lose. We need low-cost mechanisms by which people can reward cooperators and punish defectors, so that the trustworthy survive and the untrustworthy get weeded out.

There is always an incentive for people to signal that they are trustworthy when in fact they are not. This means that we need institutions to develop that make it costly to signal trustworthiness unless you really deserve it. No single institution can solve this problem. It requires layers of institutions, including governmental institutions.

There are many, many ways to cheat in a transaction. I can sell you property where I do not actually have a title recognized by the authorities. I can sell you a property that has environmental hazards that I did not disclose. I can sell you a property that is not zoned for the type of improvement you want to put on it.

Recording the transaction by blockchain does not by itself prevent me from doing any of these things. Of course, if the government recognizes a blockchain as its official title record, then the blockchain technology would be involved in the transaction. But without government sanction, a blockchain title record would be of little or no use in a dispute.

As far as I can tell, blockchain can only help to prevent one type of cheating: digital forgery. If blockchain is going to have a killer app, then it has to enable a transaction to take place where the only impediment to undertaking such a transaction currently is the threat of digital forgery. I cannot think of such transactions.

Andreessen and Horowitz would say that I have not tried hard enough to come up with applications for blockchain. That is because they think of it as an all-purpose technology, like the Internet. But I think of it as a narrow-purpose technology, along the lines of a digital signature. I think we already know what the applications are for a digital signature, and I don’t think it’s a very exciting market.

No retailers, no automobile drivers

Andreessen foresees a future, perhaps as soon as fifteen years from now, without retailers and without automobile drivers.

I think of the death of retail as the unbundling of selection and fulfillment. Today, you go to a store to select a product and to fulfill the shipment of it (by taking it home yourself). In the unbundled future, you will select the product using the Internet and fulfillment will come from a company that specializes in maintaining inventories and shipping goods.

The current model requires too much inventory to be held in the vast array of retail stores, and households spend too much time driving around in the selection and fulfillment processes. The unbundled model will squeeze out those costs.

I think it is reasonable to believe that driverless cars and human-driven cars are not destined to share the roads. Just as cars and horses were not destined to share the roads. Once people see how a city can operate with only driverless vehicles, the human driver will find himself shunted off the main roads.

Wearables, AR, and VR

Andreessen sees ear buds as the vanguard of wearable technology. It turns out that people have a lot of time for listening, and ear buds and podcasts facilitate that.

New listening technology strikes me as incremental, not revolutionary. Portable radios are a very old technology. I listened to the Beatles sing “When I’m 64” on a transistor radio when I was 13. Now I’m 64.

Andreessen is more excited about virtual reality than augmented reality. He sees augmented reality as more difficult to execute, because of the need to combine perception of the existing world with creation of alternative worlds. With VR, you just need to create the alternative world. I think that this is a really important insight.

Could AR and VR become a big part of everyone’s life? In my opinion, yes. Have the breakthroughs necessary for that to happen occurred or are they on the verge of occurring? In my opinion, no.

Explorers, followers, and futile resisters

Cowen asks Andreessen and Horowitz what weaknesses Wall Street and government have regarding technological innovation. They answer that Wall Street is too slow to “get” tech. They say that government is too inclined to resist technology.

I think that the different approaches to technology have a reasonable basis. It is reasonable for venture capitalists to experiment with new technology, to operate on the basis of hope, and to try to stay ahead of the curve. It is reasonable for large institutional managers of financial capital to try to wait for some of the uncertainty to be resolved before they make big investment commitments. It is reasonable for government officials to worry about the destabilizing potential of technology.

Because institutional capital managers do not want to participate heavily in technology at the experimental stage, when a new technology starts to pan out Wall Street will necessarily have some catching up to do. As the new technology blossoms, the venture capitalists are going to have a better feel for its potential than will Wall Street. Explorers have that advantage over followers. But followers have the advantage of not having to suffer through all of the mishaps of the explorers.

Government’s inclination to resist new technology makes sense, also. The status quo is bound to look good to someone who enjoys the power and prestige of a high government position. Other things equal, stability appeals to someone high up in government. Officials are more naturally sensitive to threats than to opportunities.

In democracies, politicians compete for votes by preying on people’s fears. They have the incentive to trumpet the risks of technology more than its positive potential.

I am not saying that government resistance to technology is a good thing. I am not saying that it can be effective for very long. All I am saying is that it is natural.

Hollywood, Silicon Valley, and Cultural Mixing

Andreessen and Horowitz note that the disparate cultures of the technology industry and the entertainment industry seem to be mixing much more easily than was the case twenty years ago. I might put it that Silicon Valley is appropriating more of the culture of Hollywood, and conversely.

Going back to the previous section, would we want such cultural mixing to take place with Wall Street and tech? With government and tech? I suspect not.

In the case of government and tech, I think that the highest potential for mixing is in applications related to the military and to security. I assume that in Israel and China, security issues provide an arena for cultural mixing between government and technology. Presumably, there is also some cultural mixing between Silicon Valley and part of the American military and homeland security apparatus.

It is interesting to speculate about the effectiveness of the use of technology for security purposes. Anti-government forces have some advantages in exploiting new technology. They are more willing and able to experiment. They can suddenly find themselves with new capabilities available at low cost. They have many potential points of attack, and governments must try to defend many potential points of vulnerability.

Against this, governments have the ability to mobilize much more human and financial resources. They are hampered, however, by difficulties in deploying resources efficiently or effectively.

Systemizers and Empathizers

Horowitz describes the ideal executive as someone who can engage in systems thinking and also see things from the perspective of other people. Horowitz talks of intuitively seeing the perspective of employees, but I think it also is important to intuitively see the perspective of customers and partners.

This made me think of Simon Baron-Cohen’s distinction between systemizers and empathizers. I interpret Horowitz as saying that a top executive needs to be really strong at both systemizing and empathizing. Given what Baron-Cohen sees as a trade-off between the two, one can see how it might be the rare person who is strong at both.

I can imagine that there are many companies where one top executive is somewhat better at, say, systemizing and another top executive is somewhat better at, say, empathizing. They serve to strengthen each other. One is tempted to speculate about Gates and Balmer. Or about Andreessen and Horowitz.

More by Arnold Kling

More Related Stories