While engaged in research, I tended to spend too much time learning how to do, sparing little time to actually do. Its helpful to map the terrain and consider related work, but this can quickly turn into procrastination. In software development, you may end up building the wrong thing.
Instead spend 1/3rd of your time writing and 2/3rds of your time rewriting. When composing, put yourself in Hemingway mode, pour ideas out and riff without judgement, and whilst editing, ruthlessly reorganise and reduce.
This works well for creative writing as well as coding, and can be incredibly freeing since you’re constantly making something. As a rule of thumb, I spend about 20% of my time on learning, design, and planning; 30% on problem solving; and 50% refining the solution.