One of the best program managers I have had the chance to work with observed that companies have challenges when transactional-based decision making and arrangement-based decisions come into conflict.
Transactional-based decision-making is what I call strict enumeration and documentation in the form of a contract. Once both sides agree on the document, then even if people change, the decision remains. The trust is in the record, and the change controls surrounding the document. This kind of decision depersonalizes the decision. These decisions tend to be transparent.
Arrangement-based decision-making relies on trust between the two parties. The parties establish trust in various ways, but typically it’s about making sure that the personal goals between both leaders align. Arrangements are remarkably durable and transportable. Meaning that once you establish an arrangement with someone, trust is the basis of subsequent decisions much faster. It also means that decisions without going through a complex process. The problem is that these decisions tend to be opaque.
Both models taken to extremes can be a disaster. I have worked in both. And they both suck.
The law is THE example of a transactional system. Overly transactional environments turn into bureaucratic, legalistic environments.
A pure transactional system can turn into a totalitarian state. A great example is – “the only way to change this spec is to call a meeting of the change control board and submit a request.”
A pure arrangement system can turn into a cult where belief in the great leader is paramount and the phrase “blah said” is used to justify or denounce everything. That disagreement with the great leader is an unforgivable sin. A great example of this is when someone says – “but SR ARCHITECT FOO said”. Or alternatively – an old-boys network – that is impenetrable.
I’ll also observe that both approaches feel natural to different people. My personal experience is being Greek and Canadian and living in the USA. As a Greek, I believe that laws are suggestions. That relationships, and in particular, family relationships, trump everything. As someone who lives in Sunnyvale, CA, I have learned that in the USA, laws matter, but that the laws are structured to satisfy arrangements among the wealthy.
A transactional system feels like a spectacular waste of time because personal relationships trump everything and that, ultimately, transactional systems are a facade for arrangements.
But I have learned that that is naive. The critical flaw of arrangements is the opacity of building trust and the boundaries of trust. Transactional-based decision-making creates a public record of the moment trust was built and describes the trust boundary. Without such a record, trust-building naturally devolves to family, culture, background, and other attributes.
This brings me to how I think about architecture reviews. The purpose of the review is to create a trust boundary between the approver and the author. The more high-level the spec, the more trust that has to exist. That trust is typically built over a series of smaller successes or previous professional successes. The more detailed the spec, the less trust that exists. A key trust moment as an architect is when you are willing to stop being the sole approver of an area.
The key illusion in all of this is that as an architect the spec actually defines what is built. The reality is that unless you are writing the software, some other human being is going to take that document and do what they think the right thing is. My job is to make sure that they are thinking about the problem in a way that aligns with a reasonable solution.
Decrying arrangements because they are old boys networks is wrong. Decrying transactional-based systems because Process is also wrong. Like most everything in life, there is a balance. And like most everything in life, navigating that balance is the art of living and being a software architect.
Leave a Reply