The video above is a two-parter. First part is demonstrating bad presentation skills. The second half (starting at 12:17) shows 'better' presentation skills. The two presentations were judged by 4 opponents, each preoccupied with a different aspect (Content, Design, Performance and Q&A). Each opponent gave scores from 0 to 4. Detailed feedback is below where ‘-‘ indicates the point was not given and ‘+’ indicates that the point was given. Passing grade is a 10/16. The presentation evaluation is based on Jean-Luc Lebrun's book: When a Scientist Presents. Jean-Luc teaches Scientific Presentation Skills every other year at UEF.

Bad presentation feedback (3/16)

Content (0/4)
- Presentation was describing a method but not comparing it against others = can't know if good or bad.
- Presentation was mostly about the fundamentals: set operations. Not obvious how it links to trajectories and why.
- Presenter assumed some things are known by the audience. Q&A revealed it was not the case.
- Q&A revealed that people stopped following during the first half.

Design (0/4)
- Flow was not good, topics were changing without warning = not easy to follow.
- Some slides contained too much information, some of it unnecessary.
- Often font was very small (screenshot of paper tables).
- Too much information presented at once = not clear where the viewers focus.

Performance (2/4)
- Speech was sometimes fast and using too technical terms without any support.
- Eye contact was often missing.
+ Audience was paying attention all the time, most likely because they had to spot the problems.
+ Equipment was managed well enough, could have used markers at times, but they are not typically available at conferences.

Q&A (1/4)
+ Presenter had answers to all the questions.
- Some answers did not satisfy.
- Answers were mostly correct, however, …
- Often answer to a different question was given.

Good presentation feedback (13/16)

Content (3/4)
+ Presentation presented a new method, which outperforms the current state of the art in terms of usability.
+ The transition between the application and the fundamental method was proper.
- The words Route and Trajectory were used interchangeably as synonyms. They are not exact synonyms. Proper definition was lacking. 
  Synonyms should be AVOIDED during scientific talks!
+ In Q&A people asked questions to different parts of the presentation. The questions were complementary to the content. 
  This implies that all parts were understood overall.

Design (4/4)
+ Flow was good, telling a story of how the method works.
+ Slides had the right amount of information.
+ Font sizes were adequate.
+ Complicated slides were not revealing all the information at once.

Performance (3/4)
+ Speech was clear and well paced.
+ Eye contact and posture were good.
+ Audience was paying attention.
- Equipment was managed well. But sometimes the next slide surprised the presenter.

Q&A (3/4)
- Presenter had answers to all the questions. Question about the GetRoutes function would have benefited from a separate slide. 
+ All that asked a question seemed satisfied.
+ Questions were answered correctly.
+ The presenter answered to the exact question that was asked.

Teacher observations

Bad presentation:
- Visual styles on slides can be OK, but usually they attract too much attention and prevent you from using all the available space.
- Figure on slide 2 could be much larger and most of the text should be removed. It should be memorized and just said by the presenter. 
- Presentation would have benefited without slide 3. The text there tries to put the work into context, but it is breaking the flow.
- The screenshots (slides 4-5) are not aligned properly when slide is changed. The interesting content is very small. 
  Zooming in, making the window smaller and cropping to the window contents should be done. 
  The result shows two trajectories on the list but only one on the map (there are 2 overlapping, but enough to raise questions). 
  Screen recording of the application can be considered. Don’t use live DEMOs unless enough testing was done. 
  Errors during the DEMO are not nice to have, however, successful DEMO is very impressive!
- Pseudocode text is very small (screenshots from paper). Acronyms are not defined or explained.
- Not all functions are defined and questions remain unanswered.
- Some terms like interpolation and b-tree seem important but are not well introduced.
- Not clear what mathematical notations mean.
- Presenter moves quite quickly through complicated slides.
- Slide 9 is very complicated. Font size is unacceptable. It is not clear at all what the presenter is talking about.
- Titles for the slides 3-10 are not proper.
- Slide 10 has a nice visual, but viewers are not prepared for it in any way
- Evaluation and conclusions are not convincing. How do we know that 84 seconds is good... It sounds quite bad.
- Slide 12 describes an unrelated, difficult topic: trajectory complexity without any support. Having text in slides and reading the text is a NO.
- Conclusions are not strong.
- The last slide stays on the screen a long time (during Q&A) it is a waste. Put there your contact information, at least. 
  Maybe remind what your presentation is about, somehow.

Good presentation:
- Should have practiced the presentation more. The talk was given few years back at a conference and prepared it in my head now a few times, but it was not enough.
- Some slides from the BAD presentation could have been used as Expert slides in the good presentation 
  Expert slides = additional slides placed at the end of the talk that are only used in case someone asks related questions.
- Slide 11 should make better use of the whitespace - larger figures can be used.
- Could have used a mouse or a pointer with slide changing functionality. Pressing the board can be disturbing to the audience.

Other recommendations (answers to students' questions):
- Slides can be numbered if you want, but keep in mind that viewer should look at your slides and focus on the content, not how much time is left. 
  Proper title should be given to each slide. Then the viewer can note down the title of the slide, not the number.
- Table of contents in the beginning?
 * If you give a very long presentation (few hours), then maybe.
 * Otherwise, begin with something interesting (hook slide) that prevents the people from going out of the room (bored) when you start.
 * And always have a proper flow for the presentation content.
- If your institute asks you to have the university / company mentioned on every slide, OK. Otherwise, beginning and end is enough. 
  You don’t want to attract attention to those during your talk.

Thoughts:
There was some bias because opponents knew that first presentation will be the bad one... I did consider switching them and not letting them 
know which is which. But if the good one would be first, the topic is better understood from the beginning. So then even a bad presentation 
can be understood after the good one. I considered giving presentations from different topics. But it is harder to compare the two in that case.