How to Do a Technical Reading Program
This approach focuses reading time by emphasizing the planning of the reading program before one sets out to read anything, and then continual maintenance of the reading plan.
Fred Brooks, 4 February 2008, with minor amendments by Neil Dodgson, 4 October 2011.
- Search for the citations
- Back-chaining from current papers
- Specialized bibliographies
- Computing Reviews and the annual indexes thereof
- Forward-chaining to papers that cite this one (e.g., via Google Scholar)
- Grade each citation with a priori (that is, guessed) "grades":
- A, B, C, D on expected relevance, based on
- 1, 2, 3, 4 on expected quality, based on
- Number of citations to it
- Sort into order on one variable within the other. You may want to keep both orders.
This yields you a matrix of paper titles, with the most promising in the upper left.
- Discuss the graded list with an advisor; revise grades.
- Confirm relevance for the easy entries, at least the A ones, revising as you go.
- Many journals are now online, so you can scan abstracts. Use Google Scholar to find an online version.
- For books, copy the title from your list into Amazon's Advanced Search page. Then use Amazon's Look Inside feature to examine Back Cover, Table of Contents, Excerpt.
- Read from cell A1 outward fairly uniformly in both directions.
- Write summaries, critiques as you read! These often carry over verbatim into your dissertation.
- Maintain your plan. Insert into the existing list, by guessed grade, new citations based on the papers you read. Occasionally, re-estimate the grades based on what you have learned. Re-sort.
- Quit reading in either dimension when it becomes unprofitable.