There’s a Python library called monty that supports a convention for moving between JSON objects and Python class instances. The major components are a mix-in class, MSONable, along with subclasses of json.encoder.JSONEncoder and json.decoder.JSONDecoder. An appropriate JSON object will have two special keys: @module and @class.

I’ve heard the phrase “just the tip of the iceberg” used as a positive phrase when revealing value of which an audience might not have been previously aware. In the context of disseminating scientific data, this tip might be a publication. A reader sees paragraphs and figures that describe and show data. A Supplemental Information section might link to a much greater volume of data – the rest of the iceberg.

In his 1962 classic "The Architecture of Complexity", Herbert Simon described the hierarchical structure found in many complex systems, both natural and human-made. But even though complexity is recognized as a major issue in software development today, the architecture described by Simon is not common in software, and in fact seems unsupported by today's software development and deployment tools.

An effective data unification system is like having an ATM network for your data. Instead of having to get to a branch (and only during open hours), wait in line, and talk to a teller, even for simple withdrawals and balance inquiries, with an ATM you can do many common operations on-demand and quickly.

I was fascinated as I skimmed over a publication detailing a network analysis of crystalline materials. ¹ It’s “Open Access”, meaning I don’t have to pay to read the text and figures. There was one figure that had neat data-driven graphics for two separate materials, one stacked on top of the other. I wondered what the graphics would look like for some of the other 22,600 materials considered by the study.

I’d like have a working mental model for this chain relating data, information, knowledge, and wisdom. Weiss and Wilkerson define data, information, knowledge, and wisdom as follows ² : The above is a helpful starting point for me, but it falls short in two ways. First, it unnecessarily complicates by branching both knowledge and wisdom from information. This is why I included the “[ sic ]” above.

I came across The DataOps Manifesto, ¹ which states five values with regard to developing and delivering analytics. I was surprised that I disagreed with all five.

Shoot for success, not perfection. Success is the path of confidence. Perfection is the path of arrogance. If you’ve achieved perfection in something, by your estimation, there is nothing that can be improved upon, nothing to be learned. This belief, that you have nothing more to learn, is the key difference between confidence and arrogance.

At my last full-time job, I had a performance evaluation each year with my supervisor. The format was batch call-and-response. First, I self-evaluated and filled out all sections. Then, my supervisor responded to my content. Finally, we met briefly to go over the content before submitting to HR. In the last section of the evaluation, I had listed my goals for the next year.

Malleable systems are software systems that are designed to be modified and extended by their users, eliminating the usually strict borderline between developers and users.