In my glacial metamorphosis from worker into entrepreneur—or at least into a person who does entrepreneurial things—I've learned the hard way about the kind of attitude I need to cultivate toward customers. Most of the influence we get from observing businesses looks like: cast a wide net, book the sale, get the money, and get out of your obligation. There's no sense of who these people are, what they're going to do with what you sell them, and how their experience of dealing with you will impact your future. Luckily I have my own domain of expertise to fall back on for inspiration:

First ask yourself How would I do this without a computer? Then have the computer do it the same way. Mark-Jason Dominus

To which I offer a corollary: if you want to understand real-life processes, examine how they have been modeled on a computer. This is not so much an issue of precession of simulacra, or even art imitating life. It's more that form follows function: certain patterns and structures have to exist in order for a simulation to work, and those are what we examine.

The transaction is a fundamental concept in computing science, not only because of its practical importance in business applications. A transaction is how we ensure that the state of a system is either changed completely in a specified way, or not at all. This is a property known as atomicity. What's interesting about transactions is what you need to do in order to make them atomic.

In particular, we require enough physical resources to contain both the before and the after states of a given transaction, on top of the workspace required to realize the result. We need to be able to present the system in its initial state to all those requestors that do not depend on the outcome of the transaction, and queue up the requests of all those that do—which consumes even more resources than not securing operations this way.

The concrete, quotidian example I use to think about this process is a jewelry store, with its elaborate and expensive air-lock-style entrance. Customers aren't allowed in until there is space and a salesperson to attend to them, and not allowed out until they either purchase something or leave it behind. Notwithstanding exogenous disruptions, like armed robbery, this system works well at ensuring that transactions either happen completely or not at all.

Returning to the computer, the transaction is considerably more mundane. Like the jewelry store, we use them to ensure that an intended outcome happens in no uncertain terms, or the universe is reset to the beginning as if nothing happened—like making a calculation in a database and writing a result to the hard disk, and getting confirmation that the system is at rest with the result safely stored. Just as a person won't be permitted to walk out of the store wearing a brand new Omega without paying for it, the machine is not permitted to exit its process in an irrevocable, half-finished state.

I submit, then, that the transaction, in the abstract, is an important sociocultural adaptation. Indeed it is a technology, in the way that, say, management is a technology. It is conceived, consciously or otherwise, by human beings, for a particular adaptive purpose.

Transactions represent the export of trust into the built environment, if we define trust as an ongoing mental account of the past behaviour of people and systems, and a set of expectations about their future behaviour. Building trust is time-consuming, and the transaction—again, the abstract concept—makes doing so less important. The salesperson doesn't need to build a relationship with the customer who just put on the watch, because they can't leave until they pay for it. Similarly, I don't need to worry about losing my work in a power outage, if my computer flushes it to disk at every turn.

(The matter of the disk itself, as many lamenters can attest, is indeed one of trust.)

Human civilization has committed considerable resources, effort and ingenuity to create environments that enable us to interact with positive outcomes without incurring the overhead of developing trust relationships. I believe this is not only okay, but essential for human survival and prosperity. Our mental capacity for forging, vetting and policing relationships is quite modest, and transactional environments extend our abilities enormously, particularly as economic and political actors. What's important to recognize is that these environments are artificial, only apply to certain well-defined domains, and are inherently fragile. When they fail, or when they can't apply, we're punted out into a broader operating context, one that is considerably murkier.

I've been mulling this line of thinking for some time, but the rhetorical support to materialize it only came to me after reading David Graeber's Debt: The First 5000 Years. In it were two interesting revelations. The first is that the canon of the maturation of economic exchange is apparently backwards. According to Graeber it begins with credit, which is inherently asynchronous, then proceeds into cash, and finally subverts into barter. The second is that many traditional, often quite formal relationships, are maintained slightly out of phase, with favours always being repaid to a slightly greater or lesser extent. To settle one's account in exactitude is like saying never speak to me again.

If these analyses are true, they speak to the notion of the economic context required to make atomic transactions possible. They also lend an inkling to how human beings operate when not enthralled in this artificial system. To the first point, we recall the thought experiment in which the most powerful Egyptian Pharaoh is imbued with all the knowledge necessary to build a Model T Ford, only to exhaust his empire in the endeavour—there simply isn't enough stuff around to pull it off. Transactional environments are artifacts just like cars are, and can't come into being until the context is rich enough to support them. To the second point, the extent that cultures lacking the means to create effective transactional environments eschew them, is a testament to their artificiality.

This is not a Luddite argument, or even a moral argument at all. Instead, I am suggesting that it is a strategic mistake to view what is actually a relationship as a mere transaction. Again, transactions can only operate over the set of well-understood results, and only in an appropriate, nicely-padded context. Absent these, we have to rely on the strength of our relationships. This is significant when what you have to offer is oddly-shaped or poorly understood, which necessarily describes the province of many innovative offerings.

The lesson is clear: If your offering is new or strange, you can't treat your early adopters like mere counterparties, like passive consumers of widgets. These people are your partners. Your success is inextricably linked to theirs. A time may come when such a transactional interface is possible, but it will only emerge with considerable effort. Until then, make the best of your relationships.