Talk Rubric

This is a "bottom line" rubric; the most important aspect of the presentation is making sure your classmates get 25 minutes of value out of your talk. Other than the quality of the results, this is how you would be implicitly graded if you were to give a talk at a conference or at CS theory lunch.

There are three things we will be watching for:

Basic Comprehensibility

• Speaking clearly, slowly, and loudly enough for people to easily understand the words that you are saying. (Tip: it is better to say fewer things slower than to cram material at the risk of no one understanding anything.)

• Making sure the overall structure of the presentation is clear enough for someone to "get back on the train" if they accidentally doze off for a bit. (Tip: remind people where you are in the structure at every transition.)

Answering the Basic Questions

By the end of the talk, the audience (your fellow students) should at minimum be able to answer the following questions:
  1. What problem is the paper trying to solve?
  2. Why is the problem interesting?
  3. What is the primary contribution?
You can assume everyone looked at the paper, but don't assume everyone managed to get correct answers to the questions above.
At some point in the presentatation, you should also answer
  1. What are the key takeaways?
This can come at the end, or somewhere in the middle. Finally, you should answer
  1. How did they do it?
as much as is possible in the time that you have.

See how to read a paper for a full explanation.

Correctness

Is your understanding of 1-5 correct. If you are having trouble understanding the paper, feel free to talk to the TA (Monday after class, or by appt).

Notes

• Bring a video recording device if you'd like, to the practice talk and/or to the final presentation. Watching yourself give a talk can be a surprising and motivating experience if you've never done it before.

• You will be graded on absolute quality, not on the improvement between your practice talk and your final talk. So if you give a great practice talk it is completely acceptable to give the same talk again.