Ownership is the essence of the American Dream – or is it? The
mortgage crisis certainly led many people to rethink the virtues of owning a
home, but even in less dramatic markets, it’s a fair question. There are many assumptions to be challenged
and hidden costs to be considered.
Warren Buffett continues
to bet heavily on housing, while Yale economist Robert Shiller contends that
housing is an investment fad, with no net appreciation in the US market
over 100 years. Of course, as author of
the Shiller post points out, most of us are living in our homes, and the benefit
is partly intangible. But how much does
the intangible actually depend on ownership as opposed to just being there?
Rental has always been a popular alternative for major, long-term-use assets with high maintenance costs. Traditionally this has meant homes and cars, but they are just the beginning. The convergence of low-friction technology, on-demand efficiencies, expanding tastes, and shrinking wallets has led to the explosion of the sharing economy, as reported by The Economist. There are countless examples, each with its own intricacies: Rent The Runway, Amazon Web Services/EC2, ZipCar, Uber, even BlackJet. It’s about deciding not to own something you really don’t need to own yourself (and achieving better financial health as a result).). Increasingly, we have the option to spread out the long-term maintenance cost, which actually exceeds the acquisition cost for more assets than people tend to realize, while maintaining high availability.
The sharing economy ranges from necessities such as housing and transportation to luxuries such as designer dresses and private jets but necessities quickly become luxuries when acquired carelessly. This is especially pertinent for government, but it’s not always obvious which costs justify themselves. Traditionally, the Forest Service, Coast Guard, police, et cetera all maintained their own helicopters, for example. Even if they were grounded 90% of the time, no one wanted to give up ownership if they had a choice. Now that states are going broke, sharing is a much more palatable option, but it’s not just about cutting costs – you have to re-examine the incentives. In government, one of the major drivers of ownership is funding. It’s easier to get large capital funds for new assets because they are assumed to be investments— and investment has a return. It’s much harder to get operational funding because that is a cost - and costs are bad, right? (how many times have you heard the renting is throwing money away?) But what if that helicopter fleet is just a really bad investment? It becomes a lot easier to make that case if you can get a helicopter on short notice, probably based on a retainer and/or hourly use fee (similar to ZipCar).
Separating the emotional appeal of ownership (as difficult as that may be), my thesis is that it is generally a bad idea to own an asset unless you have a specific and special competency to own it. This is the same for everything: housing, cars, servers - and especially software.
Cars are a tricky case, famously depreciating (up to 10%) the minute you drive them off the lot (a phrase so commonplace you probably finished it in your head). Many of us don’t know how to truly maintain our cars beyond the basics. For occasional drivers, there is the lesser option, such as ZipCar, but US infrastructure is still designed around individual drivers, and giving up your car can be very difficult if you don’t live in a city. However, something like Sebastian Thrun's self-driving car work could someday open up a whole new world of on-demand transportation that is more efficient and safer than anything we have now. Think about it: 97% of the time, your car is sitting around, taking up space, idle.
Servers, beyond the fixed costs, require hardware maintenance, networking, power and cooling. Many servers require replacement after just a few years. It’s much easier and lower overhead to simply rent the capacity you need - unless you are Google, Amazon, or the like, and have a special competency that requires you to maintain your own servers.
Software is often perfectly suited to on-demand delivery for predictable use cases, and software-as-a-service (SaaS) certainly qualifies as one of the major technology waves of recent years. More and more, the prevailing sentiment is “why buy software when you can rent it?”, as reflected in Salesforce’s now-iconic logo.
Of course, not all software needs can be satisfied by SaaS. Then the relevant question is whether to build or buy, as opposed to rent or own, but the underlying considerations are similar (if quite a bit more complex). My guiding principle is that you shouldn’t be building your own software unless you have a particular competency that requires it, or need to develop such a competency.
In keeping with the theme of recognizing our own biases, it’s important to separate the emotional resonance of ownership from the practical reality. With software, the reality is that code depreciates incredibly fast, not to mention the continuous iteration and improvement required for software to stay relevant. Ownership bias is perhaps most frequent (and outsized) in government, where the idea of “owning” the code base has become hugely and irrationally popular. In the vast majority of cases, “building” and subsequently owning your own software actually means contracting with private vendors to develop complex, bespoke systems that cost 10, even 100 times as much as an off-the-shelf product.
There is an attractive yet perniciously false idea that once you build the software, it’s yours, free and clear. The appeal is simple - people enjoy the feeling of ownership, and are naturally wary of being beholden to outside vendors. But the reality is that you are paying down accrued technical debt all the time – just as you would maintain a house or car, except that a house or car isn’t expected to fundamentally change in a matter of months. Furthermore, a bespoke project concentrates that debt with one client instead of amortizing it across all customers the way a productized solution does. In a very cynical way, bespoke developers are smart to let the government own the source code. Not only does this prevent other customers from re-using the IP (and saving money on development), but it also makes the ongoing maintenance costs easier to justify because now, it’s their baby.
The final point is that if you are going to buy, you need to make sure that the seller has a specific competency in software. It might seem obvious, but more than any other product, you want to buy software from a software company. Rolls-Royce can build world-class cars and jet engines alike, but there isn’t really an analog in the world of aerospace companies and systems integrators that also attempt to build software. The product lifecycle, pace of innovation, maintenance considerations, and above all the deltas between good and great all make software unique among industries.