Review and open questions

Now that we've spent a few weeks discussing an issue, spend some time thinking about the papers we read. What questions did they answer? What questions did they leave open? How do they all fit together?

For class, identify one particularly attractive open problem posed by these discussions. The problem should be (a) of interest to you, (b) important to solve, and (c) solvable (or a problem for which progress can be made).

For class, bring a summary of your thoughts on the problem, including:

• A crisp problem statement. Identify a specific hypothesis to be tested.
• A description of why this question is important (what is impact of answering the question?)
• A plan for answering the question (e.g., what experiment would you do to answer the question?)

In these discussions, aim for precision, specificity, and brevity. "Improve prefetch algorithms" is not a useful problem statement (unless you have a specific plan for accomplishing this). "Quantify the sensitivity of Duchamp's prefetching hyperlinks algorithm to each of the following four factors: (prefetch threshold, bandwidth threshold, PPM(1,1) v. PPM(2,4), and filter-by-hyperlink-structure) might be.

Work individually or in a team of at most 2 people

Print your summary using large print -- we'll use an overhead projector camera to display each proposal for discussion. You should also e-mail your summary to Arun. Arun will create a web page (visible only from within UT) of the class's ideas.

Class will be moved to ACES 6.TBD for these discussions

Feel free to provide multiple submissions...

Also, if there were any issues that in retrospect you think need more discussion or thought, feel free to bring those up.