Talk Guidelines

Last updated: January 8, 2024

These are general rules for giving talks composed primarily by Stephen Mann and Rick Kazman but with input from a variety of sources. The rules are broken up into four sections:

Designing the talk
Designing the slides
The presentation
Jokes

Note that all "rules" are guide lines. However, if you break one, you should have a good reason, and even then it is better to get a second opinion.

Some other talk guidelines pages include:

Bruce Randall Donald

1. DESIGNING THE TALK

Ask yourself who your audience is. In particular, determine the technical expertise of your audience, and give a talk that is technical significantly below that level.
Gear your talk for your audience. Examples of audiences that require different talks:
1. General Audience.
2. General Technical Audience. This is at a departmental level.
3. Technical Audience. This is at a discipline level.
4. Specialist Audience. These people are the ones who write the papers you read and read the papers you write.
In general, as you move down the list, you can give more detail and expect the audience to know more. However, it is easy to give too much detail and lose even a specialist audience.
Make sure that your talk has an obvious organization (this can be one of your first slides so that the audience knows what is coming), so that, if you get lost or stalled, it is easy to jump back in.
Decide what the main idea you are trying to present is. This becomes important when you try to shorten a talk: you remove all material not related to this idea.
Make sure that your talk has a goal. Make sure to describe the relevance of the information described in the talk. Give a short historical context.
The amount of material and detail about this material will depend upon the time allotted for the talk. Cutting material is preferred to doing a poor job in presenting it. When presenting a technical paper at a seminar, remember: you do not have to present the entire paper. Concentrate on what is interesting and/or novel.
The audience should be able to answer the following questions after the talk:
1. What was the talk about?
2. Why was this interesting?
3. What was new? [research talk]
4. What still needs to be done? [research talk]
Remember: a technical talk is different from a class room lecture. Far less detail is acceptable in a technical talk.

The following talk structure is taken from a set of guidelines written by >Farhad Mavaddat.

Here is a typical, concept-based, presentation format:

Start with a cover page. Give the title of presentation and your name on that page.
Next, give a point-wise outline of the talk. This way, you prepare the listener for what is to come; you also make it easier for him or her to relate the topics that you may be presenting in a disjoint form (of course not a good practice).
Next, give a definition of what the topic is all about and what issues, inefficiencies, or class of problems it tries to solve.
Next, if relevant, give an outline of the methods and/or techniques used to achieve the goals. Make sure any assumptions made (for example `caching' assumes program locality) are stated.
If the subject is too broad, or covers too many techniques and methods, decide which are more important; concentrate on those. It is preferable to cover one or two important methods or techniques well, rather than trying to cover all of them superficially.
When concentrating on only a few of the methods or techniques, you may also want to say something about other techniques or methods; at least mentioning their names ... .
Present your understanding of the topic in your own words, using figures and drawings, when possible.
Present the main ideas, assumptions, results, and techniques. Don't spend time on proofs and the details of theory. They can be read later.
In planning your presentation, keep in mind that a good sequence of topics for presentation is often different from a good sequence for writing them.
It is a good practice to close the talk by pointing to the shortcomings, problems, and controversies (if any) surrounding the proposed methods or techniques.

2. DESIGNING THE SLIDES

Avoid giving equations. If you do show an equation, then it should either be a "reminder" equation that is familiar to the audience and merely informing them of your notation, a trivial equation (A x = B) or you should spend a large amount of time discussing it.

There are at least two ways to overwhelm an audience with equations. First, you can give complex equations with inadequate explanation. Second, give a whole lot of equations. Even if the equations adequately explained, only the specialist in the audience will be willing to follow what you are doing, and a lot of them will tune you out, too.

In equations, avoid using variables that look similar. In particular, avoid using u and v together.

Proofread and spell-check your slides. Then, get a friend to proofread them, because neither you nor the spell-checker will notice things like using "if" for "is" or "a" for "an".
For those who feel spell-checkers are adequate, here's a further discussion of the merits of spell-checkers.
Avoid acronyms, unless they are central to the talk, and used frequently throughout the talk. Make sure you define any acronyms that you use (unless they are universally known, like RCMP, or have become regular words, like scuba).
Use a visually appealing slide style--give some thought to the use of icons, borders, shadowing, colour, but...
Keep slides visually simple, uncluttered.
If you are using colour, make sure that your combination of background colour and font colour is readable from a distance.
Use LARGE fonts--you should be able to read your (unprojected) transparencies from a distance of 2 meters.
Do not put too many points on a slide: 3-6 is a good rule of thumb. Points should be just that, not essays.
Use charts/graphs to indicate trends. Use tables of data only where precise data values are crucial.
Use figures or pictures where appropriate. These helps to hold an audience's attention.

3. THE PRESENTATION

Practice your talk out loud at least twice, in front of a mirror or (better still) in front of friends/family.

Time your talk. Put timing marks in your notes so that you know whether you are ahead or behind your target. Have slides near the end of the talk that have both a "long" and a "short" version so that you can adjust your time as needed.

Practice your talk. Out loud. You shouldn't be stumbling over words. If you get stuck on a sentence, practice it until you find a wording you like. "Out loud" is the key idea here.

When using overhead projectors or slides, stand near the projection screen and not the projector. This allows the audience to look both at you and the screen. A corollary is that you should point to the projection screen and not to the slide on the overhead projector. If the screen is too high to reach, then fine, point to the projector. Or better yet, see if you can get a pointer (either a stick type or a laser pointer).
Practice your talk. You shouldn't memorize your talk. Rather, you should let your slides trigger what you want to say.
Do not read your slides. Talk about them. You are adding insight--the audience can read.
Practice your talk. You should be able to give a practice by yourself without stumbling before you give a practice to others.
Never leave the screen blank. If you have nothing to show, turn off the projector.
Practice your talk. After practicing to an audience, make changes based on what they suggest.
Changing slides keeps the audience's attention. Don't spend too much time on one slide. If an idea is that complex, it should probably have multiple slides anyway.
Practice your talk. If needed, bring food and drink to bribe an audience to attend your practice. :-)
Talk to the audience, not the slides or the screen. If possible, target people in the audience (changing target frequently) and talk to each selected person for a few moments. This helps to engage the audience.
Practice giving talks. The more talks you give, the more relaxed and prepared you'll be for the important talks.
Defer questions which take more than 30 seconds to answer to the question period following the talk. Some questions may be able to be deferred to a point later in the talk. A presentation that flows well will cause people to ask questions that you will be providing answers for shortly in your talk. Defer the question by stating that you will answer those questions in a little while and simply continue with your planned presentation.
Practice your talk. Ideally, practice under the same conditions that you'll be giving the talk, preferrably in the same room. This may not be possible at SIGGRAPH, but it should be possible at smaller conferences.
If you are going to use a laser pointer, learn how to use it. In particular, note that your hand is not that steady in the best of circumstances, and you will be unable to hold the laser steady. So you should circle or underline with the pointer, rather than try to hold it steady at a point (if you try this, the laser light will bounce around).
Practice your talk. The more "important" the talk, the more practice that is needed. A conference talk or interview talk should be practiced multiple times before an audience (the number of times depends on how many changes you need to make, and how comfortable you are in giving it). Don't forget: if you're shakey during the practice, then you'll probably do a bad job on the real thing.
The reason "practice your talk" is on here so many times is because it is the easiest thing to NOT do (or avoid doing). This gets more and more true as you gain experience in giving talks...

4. Jokes

I pulled these off of rec.humor. Your goal is to not have any of these be said about you.

From: nancy.carson@alsbbs.org (Nancy Carson)
Newsgroups: rec.humor
Subject: SPEAKER
Date: Mon, 12 Feb 1996 00:16:00 GMT

       -Recently our speaker had to discontinue several of his long
        talks on account of his throat.   Several people threatened
        to cut it.

       -You have heard it said before that this speaker needs no
        introduction.   Well, I have heard him and he needs all the
        introduction he can get.

       -You haven't heard nothing until you've heard our speaker of the
        evening.  Then you've heard nothing.

       -Our speaker needs no introduction.   What he needs is a
        conclusion.

       -Our speaker will not bore you with a long speech...he can do it
        with a short one.

       -At the place our speaker last spoke, he drew a line three blocks
        long.   But then the police took his chalk away.

       -You've been a wonderful audience...you stayed.

From: "Coby&John" 
Newsgroups: rec.humor
Subject: A long speech......................
Date: Tue Sep 07 20:01:00 EDT 1999

 A man is giving a speech at his lodge meeting.
 He gets a bit carried away and talks for two hours.
 Finally, he realizes what he is doing and says;
 "I'm sorry I talked so long. I left my watch at home."
 A voice from the back of the room says, "There's a calendar behind you."

Stephen Mann