Value Usefulness

I've been contemplating how to spend my time effectively and concluded that maximizing time on task is crucial[1][2]. I developed some heuristics that might help in selecting the types of tasks to focus on, potentially reducing planning time without sacrificing quality[3].

Value tasks that produce inherently useful output. For instance, in my scientific machine learning work, I should prioritize creating my own explanations and implementations of research papers. This means valuing deep engagement—reading, implementing, and critically evaluating papers—over merely skimming through a large quantity or doing superficial reimplementations. The benefits include:

  1. Producing useful outputs like code and documentation that I fully understand and can confidently adapt. This includes validating authors' claims, improving explanations, or even disproving results[4].
  2. Gaining a practical understanding of the methods.
  3. Enhancing my ability to quickly grasp theoretical concepts.
  4. Improving technical expertise in both practice and theory.

This approach involves creating GitHub repositories or documents of a quality suitable for public release—work that would be genuinely useful to others. Over a semester, this might result in 10-50 valuable repositories, which is not the norm for most researchers.

Another focus is on generating useful GitHub repositories[5]. These could include implementations of working papers, my own ideas, or topics I'm learning. For my own ideas, the goal is to test them immediately—either pursue them if they're promising or move on if not. For topics I'm learning, the repositories should feature well-written explanations and functional, useful code implementations.

Notes:

  1. There's a heuristic here about distinguishing tasks that require extensive planning versus those that benefit from practice—where all practice is essentially equal. I want to explore this further.
  2. I suspect I underestimate the importance of avoiding interruptions. I was interrupted while writing this and lost my train of thought.
  3. Once you've proven something to yourself, you've created a rapid decision-making process that yields optimal results immediately. Revisiting it unnecessarily can introduce errors, leading to inferior outcomes. Therefore, derive something once, verify its truth, and then integrate it into your process with confidence.
  4. Avoid internalizing false information. Test the validity of new information, and once accepted, incorporate and trust it.
  5. Writing should follow a similar principle. Confidence in one's process leads to openness in sharing ideas. People often hesitate to share publicly due to fear of marginalization or exposing a weak process. Releasing work into the public domain involves relinquishing some control over its interpretation, but confidence in the process can mitigate this hesitation.

Back to Home