Skip to content

Latest commit

 

History

History
44 lines (35 loc) · 1.8 KB

philosophy.md

File metadata and controls

44 lines (35 loc) · 1.8 KB

Philosophy

Science is necessarily an iterative process. Ideas build on previous ideas. It's only by following a field's history (past successes and failures) that you can move forward. New researchers in a field are particularly disadvantaged by not experiencing the evolution first-hand.

Ideally, you could follow the "tree" of cited works. This rarely works in practice.

First, most papers include approx. 40 references. Following a lineage back two steps means you're reading 64,000 papers. Authors have a vested interest in obscuring the relationship of their publications to others. They may want to hide the similarity to a previous or simultaneous work. They may want to draw attention away from results that disagree with the published literature for which they have no explanation.

Second, you can only look backwards. If you want to build off an older paper and want to see how their work panned out, you are even worse off. You can look at works citing, although this is even more indiscriminate than worked cited. If there are few works citing, you don't know if the paper was wrong, not useful, or just not noticed.

You might try to get a broader overview of the literature by scanning abstracts. Of course, these abstracts were written by the authors to sell the paper. They aren't updated to highlight the contributions from the paper that stand the test of time.

My philosophy is one of hand-curation. Only relevant papers should be included. Only relevant citations and cross-references should be included and it should be deadly explicit what ideas were borrowed or built upon.

There is value in opinionated descriptions of papers. If you want to be completely fair and correct, your description won't have any content. If you want the authors to be completely happy, you'll just get the abstract.