Gestalt-driven Software Development

gestalt: Something that is made of many parts and yet is somehow more than or different from the combination of its parts

Merriam-Webster Dictionary

The ability to generate outputs greater than inputs is why organizations exist, and organizations are how all the multi-human achievements in history are possible. This axiom applies to product organizations no differently. And yet, after a certain period of time, all organizations find themselves at a slowing pace; leaders and followers no longer find themselves gaining from each other’s achievements, and it no longer makes sense to work together. Hence, the organization dies, the torch passes to another generation, and the cycle begins anew.

I’ve found that the strongest and most effective organizations stay this effect as long as they can, in part due to maximizing the gestalt of the organization and increasing the logic of cooperation.

The following tips might help spark ideas on how to increase these benefits:

  • Aggressively establishing high engineering standards: Setting standards and enforcing them (e.g. through a continuous integration and delivery pipeline) stops engineers from worrying about engineering aspects that don’t impact business value and establishes a ground truth for shared understanding of the codebase. In addition, it creates a high level of social trust within the organization, which helps lead to decentralized and faster decision making.

    Apply this principle by using linters like pylint and explictly enumerate every rule, strictly versioning all dependencies, and maintaining up-to-date issue and code review templates.

  • Keeping their data clean: Clean data, easily maintained and strictly regulated, accelerates business insights by deduplicating and unifying various code paths utilizing that data. It also lends itself well to creating data lakes within an organization, helping to break down barriers between departments (e.g. HR or Sales/Marketing and Engineering or Product).

    To apply this principle, establish and maintain high engineering standards for data management, while also ensuring data remains easily accessible to both engineers and non-engineers. A product like Airtable would really shine in this category.

  • Maximizing code reuse: Reused code is time saved and code better tested in production. Additionally, as all code is technical debt, minimizing the amount of code written helps de-risk your codebase, and keeps your options more open in choosing the direction of your codebase and hence your product offerings as your organization grows.

    Structure your code so that every functional unit is useful in general, even if you work on a monolithic application as opposed to a something more modular. If code cannot be reused, it should remain minimal. Clients (Web frontends, CLIs, etc.) should be comparatively thin, while backends and services can be comparatively thicker. Keep breaking this down as appropriate (e.g. once you have formulated and validated a design language, create a standardized UI component library and subtree into the appropriate frontend repositories). While there are exceptions to this rule, keep them few and far between, and only apply these exceptions after serious peer review of the design.

I hope you found these tips helpful! Leave your thoughts in the comments below.

Subscribe to my mailing list

comments powered by Disqus