Software is very different. It’s not, “I’m going to do a new mode of production that’s going to do more with less.” It’s, “I’m going to create a new way to serve customers with existing stuff that we have. I’m going to do it by building an abstraction on top of existing hardware…”
That was Alex Danco on a very interesting episode of the Venture Stories podcast. Danco lays out the existing paradigm where innovators were able to build the future because the stories they generated create hype. The resulting financial speculation makes capital very cheap to these innovators, allowing them to build that ambitious future, which couldn’t have happened without the irrational bubble in which people become exuberant.
It was nice to hear Danco apply and expand on many of the concepts from Bill Janeway’s book, Doing Capital in the Innovation Economy. Here is a snippet of what I took away from Janeway:
What I think Janeway was saying is that Government needs to invest more on the front end of R&D, which will mature technology until it can be speculated upon by financial markets. Then, inevitably, financial speculators will be either wrong or over-exuberant, leading to bubbles and crashes. While this is a necessary step for scaling technologies into the consumer world, it can also lead to misallocation, recession, and job-loss. So the Government must also be there on the back-end, through entitlement spending and the like, to mitigate the downside of the crashes it precipitated so that we avoid Great Depression scenarios.
Like Janeway, Danco seems to be generally positive on government as a source of patient capital on the front end and social welfare on the back end. Danco also said he also was happy to see tech firms becoming this source of patient capital for ambitious R&D projects, and how their users are getting huge benefits from the firms which are able to charge third parties for access to them.
What Danco seemed really excited about, however, was the role of software as a service being deployed on top of existing hardware, and because vastly more users were able to access the service in many new and unforeseen use-cases, the new requirements would pull technology into the future. This, rather than speculative bubbles pushing hardware technology to users before they have given feedback. Here is Danco on how the requirements pull approach and accelerate innovation:
It’s a positive feedback cycle that’s reinforcing itself. Hey, here’s a chip, and we’re going to get this chip or this computer out to consumers. We’re going to make abstractions [i.e., software], and we are going to make abstractions more powerful than we previously could, and those abstractions are going to be exposed to more people and more use cases. They’re going to break the hardware. The hardware isn’t going to be able to keep up. But it creates a market pull that forces hardware makers to make better stuff, or meet that demand. As soon as you give that increased performance or next generation back to the consumers, we immediately take that performance and write more ambitious abstractions, and we break the computers again. But it creates this pull…
In the old way, it was very clear what was scarce. Units of production were scarce. Whatever it was that I was making in the factory was scarce. And my ratio of cost of capital to return on capital is what measured scarcity. If I owned something scarce, then capital will be available for cheap, and I’ll be able to make a return on it.
Now, it’s like this has migrated to the other side. Now it is users that are scarce… If I own something truly scarce, then my users will cost me less to acquire.
And here’s some more:
Before software, what was scarce was the unit of production, like a factory that makes cars, or a railroad that transports people, or electricity that you use to power your house, or whatever it might be… Instead of investing all this money into hardware, and hope that the hardware works and that users use the hardware, I can say, I’m just going to make an abstraction on top of hardware that the user’s already have, and I’m going to get all the users, because I believe they want this abstraction, and I’m going to force all the hardware, or force all the third parties, or force everyone else, to come along with me, and I’m going to steal all of their attractive economic profits. And the way I’m going to keep this position, is that I’m going to pass on most of those profits to users. Again, this is why people love Google.
It wasn’t clear to me to what extent Danco thinks that technology-push still plays a part. He was bullish on cryptocurrencies, which get momentum based on speculation. I think another way of looking at technology-push, as the internet and crypto are showing, is people’s genuine fascination with the technology, and the discovery that new opportunities have opened up.
I’m interpreting Danco, perhaps unfairly, as saying something like, tangible asset production used to be “technology-push” through bubbles and speculation in financials, but now tangibles are being “pulled” in a more guided and iterative way, because of software abstractions. So technology-push is now on the software platforms that have a large number of customers, and if a firm can demonstrate high user growth, there will be speculation and cheap capital, which can then create new abstractions (like AI interfaces) which will then pull technology along (like driverless cars).
Of course, we have long known that the growth of knowledge takes an interaction of requirements-pull and technology-push approaches, which itself can be framed more generally as an interaction of deductive and inductive methods. While Danco’s views in relation to this framework were not explicit, I really enjoyed his discussion. It forced me to ask new questions about old concepts.
Here is Alex Danco on cost disease:
Outside of cities where you have land and zoning and all this complicated stuff, the price of building a house as stayed more or less constant over the last couple decades. What’s inflating is, one, the price of land. That’s not a mystery, it’s straight up positional scarcity. And the second thing is that it’s more expensive to get permits, it’s more expensive to do bureaucracy, its more expensive to do zoning. That has a cost-disease element but I think that it’s mostly, oh, the world is more complicated, it gets slower. That’s just how governments and bureaucracies work. The more complicated and more entrenched they get, the more costly they are to use. That’s not new.
In cost disease, I tend to emphasize the second, problems of bureaucracy (or lack of trust) more than the first, problems of positional scarcity. What does he mean by positional scarcity? Land, for example, is only expensive because one person wants a “better” spot in relation to other people, so they bid up the price not because of the underlying value of the asset, but because they have more income from productivity in other sectors (e.g., manufacturing and tech) that the wealth generated can go into positional goods that are artificially scarce (e.g., housing, education).
It’s interesting to think about positional scarcity as a driver of cost disease in defense acquisition, but I think it has some power. For example, access to the budget is positional scarcity. Which organizations are the most important? Importance is measured by the size of the budget allocation. When one service adds something to the wish list, or experiences cost growth, it is asking for a larger share of a fixed pie. Certainly much of the squabbling in the Pentagon comes down to a year-after-year fight to justify their funds (particularly in a zero-base budget, where you have to build a case every year from scratch).
It seems to me, however, that Danco downplays the role of poor bureaucracy leading to vastly higher costs. He seems to say that bureaucratic ineptitude is a fact of life that grows with complexity. Nothing to be done.
Perhaps that’s OK in some sectors where government is on the sidelines and often finding itself behind the times and trying to catch up with regulations. But in places where government is active and the gatekeeper to innovation, like acquisition in the DOD, NASA, DHS, DOE, NIH, CIA, etc., the government doesn’t just provide regulations like zoning that are kind of like “rules of the game,” it is both regulator and participant in the sector. That’s different, and a big deal, considering Danco thinks (like Janeway) that government has this important role in innovation. Surly it does, but they neglect the way it actually functions and interacts with innovation, not just through regulations, but through the process by which it arrives at decisions.
Leave a Reply