How to speak

 

https://x.com/aakashgupta/status/2037566060485083399

Transcript of the video (Patrick Winston's MIT lecture on "How to Speak"):

The uniform code of military justice specifies court-martial for any officer who sends a soldier into battle without a weapon. There ought to be a similar protection for students because students shouldn't go out in their life without the ability to communicate.

And that's because your success in life will be determined largely by your ability to speak, your ability to write, and the quality of your ideas — in that order.

I know that I can be successful in this because the quality of communication — your speaking, your writing — is largely determined by this formula. It's a matter of how much knowledge you have, how much you practice with that knowledge, and your inherent talent. Notice that the T is very small, but what really matters is what you know.

At this point it came to me suddenly a few decades ago when skiing at Sun Valley. I had heard that it was celebrity weekend and one of the celebrities was Mary Lou Retton, famous Olympic gymnast, perfect tens in the vault. And I heard she was a novice at skiing. So when an opportune moment arrived, I looked over on a novice slope and saw this young woman who, when she became unbalanced, went like that. And I said, "That's got to be her. That must be the gymnast."

But then it occurred to me I'm a much better skier than she is. She was an Olympic athlete — not only an ordinary Olympic athlete but an outstanding one — and I was a better skier because I had the K and P, and all she had was the T. So you can get a lot better than people who may have inherent talents if you have the right amount of knowledge.

That's what my objective is today. Here is my promise: Today you will see some examples of what you can put in your armamentarium of speaking techniques. And it will be the case that some one of those examples, some heuristic, some technique — maybe only one — will be the one who gets you a job. And so this is a very nonlinear process. You never know when it's going to happen. But that is my promise: By the end of the next 60 minutes we have been exposed.

In order to do that, we have a rule of engagement. And that is: No laptops. No cellphones. So if you could close those, I'll start as soon as I'm done.

Some people ask why that is a rule of engagement. And the answer: We humans only have one language processor. If your language processor is engaged browsing the web or reading email, you're distracted. Worse yet, you distract all the people around you. Studies have shown that. And worse yet, if I see an open laptop somewhere back there or up here, it drives me nuts and I do a worse job. So that ensures all of your friends who are paying attention don't get the performance they came to have.

So for preamble, let's start. First thing we talk about is how to start.

Some people think the right thing to do is start a talk with a joke. I don't recommend it. And the reason: In the beginning of a talk, people are still putting their laptops away, becoming adjusted to your speaking parameters, to your vocal parameters, and not ready for a joke. So it doesn't work very well — usually falls flat.

What you want instead: Start with an empowerment promise. Tell people they're going to know at the end of the hour what they didn't know at the beginning of the hour. It's an empowerment promise — it's the reason for being here.

What would be an example? "Oh, I see. End of this 60 minutes you will know things about speaking you don't know now, and something among those things you know will make a difference in your life." Yeah, that's an empowerment promise. So that's the best way to start.

So now that I talked about how to start, what I want to do is give some samples of heuristics that are always on my mind when I give a talk.

First of these heuristics is that it's a good idea to cycle on the subject — go around again. Some people say: Tell them what you want to tell them, tell them, tell them again. Then third time — as if people weren't intelligent. But the point is: There are many reasons. One of which: At any given moment, about 20% will be fogged out no matter what lecture. So if you want to ensure that the probability everybody gets it is as high as needed, you need to say it three times. So cycling is one thing I always think about when I give a talk.

Another thing I think about is: In explaining my idea, I want to build the fence around it so that it's not confused with somebody else's idea. So if you were from Mars and I was teaching about what an arch might say, well that's an arch. And that's not to be confused with some other things people might think is this — not an arch. That's not an arch. I'm building a fence around my idea so it can be distinguished from somebody else's idea.

So in a more technical sense I might say: Well, my algorithm might seem similar to Jones' algorithm except his is exponential and mine is linear. That's putting a fence around your idea so people cannot be confused about how it might relate to something else.

The third thing on this list of samples is the idea of verbal punctuation. The idea here is that because people will occasionally fog out and need to get back on the bus, you need to provide some landmark places where you're announcing it's a good time to get back.

So I might in this talk say something about this being my outline: First thing we're going to do is talk about how to start. Then we're going to deal with these four samples. And among these four samples I've talked about the first idea — that's cycling. Second idea — building a fence. Third idea — verbal punctuation. Some enumerating, providing numbers. I'm giving you a sense that there's a seam in the talk and get back on.

Okay, so now we're on a roll. Since we're on a roll, can you guess what the fourth idea might be here? An idea that helps people get back on the bus? Yes? Ask a question? Yes? Thank you.

So ask a question. And I will ask the question: How much dead air can there be? How long can I pause? I counted seven seconds. Because then people will be embarrassed to say what the answer is. Can't be too hard because nobody will have anything to say.

So here are some sample heuristics you can put in your armamentarium and build up your repertoire of ideas about presentation.

What I hope this persuades you of is: If you watch the speakers you admire and feel are effective, ask yourselves why they are successful. Then build up your own personal repertoire and develop your personal style. That's my fundamental objective.

The rest of this talk is about some things in the armamentarium I think are effective.

So next thing on our agenda as we start to discuss these other things is a discussion of time and place.

What do you think is a good time for a lecture? 11 a.m. The reason: Most people at MIT are awake by then. Hardly anyone has gone back to sleep. It's not right after a meal. People aren't fatigued from this or that. It's a great time to have a lecture.

So that brings the next question of what about place. And the most important thing about the place is: Well lit.

The problem with other kinds of rooms is that we humans — whenever the lights go down or whenever the room is dimly lighted — it signals that we should go to sleep. So whenever I go somewhere and give a talk, even today, the first thing I do when I speak to audiovisual people is say: Keep the lights full up. Oh, they might reply: People will see the slides better if we turn the lights off. And then I reply: It's extremely hard to see slides through closed eyelids.

What else can you say about place? Well, the place should be cased. And I mean that in the cloak-and-dagger sense — like if you're robbing a bank, you would go to the bank on some occasions before to see what it's like so there are no surprises when you do your robbery.

So whenever I go somewhere to speak, the first thing I ask my host to do is take me to the place where I'll be speaking. Are there any weirdnesses I will be able to deal with? Sometimes it might require some intervention. Sometimes it just may require me to understand what the challenges are.

So when I came here this morning, I did what I typically do: Imagine that all seats were filled with disinterested farm animals. No way I knew — no matter how bad it was — wouldn't be as bad as that.

So finally, it should be reasonably populated. It should be the case there are 10 people in this hall — everyone would be wondering what's going on. It's so much more interesting. Nobody here. So you want to get a right-size place that's not packed but it has more than half full.

So those are some thoughts about time and place.

Next thing I want to talk about is the subject of boards and props and slides. Well, these tools in the trade — I believe this is the right tool for speaking when your purpose is informing. Other slides are good when your purpose is exposing. But it's what I use when I'm informing, teaching, lecturing. And there are several reasons why I use it.

For one thing, when you use the board you have a graphic quality. It's the case that when you have a board, then you can easily exploit the fact that you can use graphics in your presentation. So that's a graphic quality I like.

And next thing is the speed with which you write on the blackboard — approximately the speed people can absorb ideas. If you go flipping through a bunch of slides, nobody can go that fast.

Finally, one great property of the board is that it can be a target. Many people who are novices at speaking find themselves suddenly aware of their hands — as if their hands were private parts that shouldn't be exposed in public. So right away they go into the pockets. And this is considered insulting in some parts of the world. Or alternately, maybe hands will go in back like this.

I was once at a convent in Serbia and my host — as soon as we entered, a nun came up to us, offered refreshment. About thank you. When he said eat that stuff or die — it's a question of local custom and politeness. But then before anything happened there, the nun pulled my hands off like this. Because it's extraordinarily insulting in that culture to have your hands behind your back.

So why is it? Well, usually it's supposed that it has to do with whether you're concealing a weapon. So if your hands are in your pockets or behind your back, then it looks like you might have a weapon. That's what I mean by the virtue — one of these virtues of the board. Now you have something to do with your hands. You can point out stuff.

I was once watching Seymour Papert give a lecture and thought it was terrific. So I went to a second time. First time to absorb the content, second time to note style. What I discovered is that Papert was constantly pointing at the board. Then I thought about it a little while and I noted that none of the stuff he was pointing to had anything to do with what he was saying. Nevertheless, there is an effective technique.

So that's just a little bit about the virtue of blackboards.

Now we talk about props. You know, the custodians of knowledge about props are playwrights. Many decades ago I saw a play by Henrik Ibsen. It was Hedda Gabler. I remember vaguely there is about a woman in an unhappy marriage and her husband was in competition for an academic job with somebody else. He was going to lose — probably because he was boring and partly because the competitor had just written a magnificent book. By the way, this is back in the days before there were copying machines or computers.

Anyhow, as the play opens there is a potbellied stove. And at the beginning of the play on the potbellied stove with its open door just has some slightly glowing embers. But the potbelly stove is always there. As tension mounts into the play you see this manuscript prop that Ibsen artfully used to know something will happen. Because as the play goes on, the fire gets bigger and hotter — finally all-consuming. You just know that manuscript is going to go into that fire. It's a memorable thing.

What I remember about why playwrights have got this all figured out — but in their hand they are not only people who can use props. Here's an example of these with props, also due to Seymour Papert.

He was talking about how it's important to look at the problem in the right way. And here's an example that not only teaches that but makes it possible for you to embarrass your friends in mechanical engineering.

So here's what you do: Take this bicycle wheel, start it spinning. Then put some torque on the axle — or equivalently you blow on the edge. And the issue is: Does it go that way or does it go that way?

Now mechanical engineers will immediately say, "Oh yes I see — right-hand screw rule." Put their fingers in this position but forget exactly how to align our fingers with various aspects of the problem. And so it's usually the case that they get it right about a 50% probability. So their very fancy education gets them up to the point where it's equivalent to flipping a coin.

But it doesn't have to be that way because you can think about the problem a little differently. So here's what you do: Take some duct tape and put it around part of the wheel like that. Now you start to think about not the whole wheel but just a little piece underneath the duct tape.

So here that piece comes rolling over top. At this point you blow on it with a puff of air — forgetting about the rest of the wheel. What happens to that little piece under the duct tape? It must want to go that way because you banged on it like that. So it's already going down. And what about the next piece? Same thing. Next piece? Same thing. So the only thing that can happen is the wheel goes over like that.

Now you'll never wonder again. Thinking about the problem the right way — demonstrated by use of a prop.

Try this after we are done.

Another example I like to remember is one from when I was taking 8.01. Alan Lazarus was the instructor at the time. And he was talking about conservation of energy — kinetic, potential. There was a long wire in the ceiling of 26-100 attached to a much bigger steel ball — but one not unlike this.

And Lazarus took the ball up against a wall like this. He put his head flat against the wall to steady himself and then let go. And the pendulum takes many seconds to go over back and then gently kisses Lazarus's nose.

So you have many seconds to think: This guy really believes in the conservation of energy.

Do not try this at home. The problem is that the first time you do this you may just let go. There's a natural human tendency to push.

So that's a little bit on the subject of props.

You know it's interesting — whenever surveys are taken, students always say more chalk, less PowerPoint. And why would that be? Our props were also very effective. But what would that be? I'll give my lunatic fringe view on this.

It has to do with what I would call empathetic mirroring. When you're sitting up there watching me write on the board, all those little mirrored neurons in your head I believe become actuated and you can feel yourself writing on the blackboard. And even more so when I talk about this steel ball going out this way and this way — you can feel the ball as if you were me.

And you can't do that with a slide. You can't do it with a picture. You need to see it in the physical world. That's why I think — oh yes, of course there are speed questions involved too that have to be separated out. But I think the empathetic mirroring is why props and these blackboards are so effective.

Well, let's see. Oh yes, there is one more thing by way of tools. It has to do with the use of slides. I repeat: I think they're for exposing ideas, not teaching ideas. But that's what we do in a job talk or conference talk — expose ideas. We don't use them for teaching.

So let me tell you a little bit about my views on it. I remember once I was in Terminal A at Logan Airport. I had just come back from a really miserable conference and the flight was really horrible. It was one of those that feels like an unbalanced washing machine. For the only time in my life I decided to stop on the way to my car and have a cup of coffee, relax a little bit.

As I was there for a few minutes someone came up to me and said, "Are you Professor Winston?" I think so. I said I don't know — I guess I was trying to be funny. In any event he said, "On my way to Europe — job talk. Would you mind critiquing my slides?" Not at all. Too many words. How did you know? He said, thinking perhaps I had seen a talk of his before. My reply was: Because it's always true. There are too many slides, too many words.

So let me show some extreme examples of how not to use slides.

Well, this demonstration needs to be way over here. And when I get over here then I can start to say things like: One of the things you shouldn't do is read your transparencies. People in your audience know how to read and reading will just annoy them.

Also, you should be sure that you have only a few words on each transparency. See? And that their words are easy to read. And I hope they're driving you crazy because I'm committing all kinds of crimes. The first of which is that there are too many words on this line. Second of which is way over there in the slides — away over there. See? You get into a tennis match feeling — shifting back and forth between the slide and the speaker.

You want the slides to be condiments to what you're saying, not the main event — or the opposite way around.

So how can we fix this? Step number one is get rid of background junk. It's always a distraction. Step number two is to get rid of the words. When I reduce the words to these, then everything I read a previous time — not licensed to say because it's not on this line. I'm not reading my slides anymore, but I'm saying what was written on the slide in a previous example.

So what else can we do to simplify this? Well, get rid of the logos. They don't need them. Simplification. What else can we do? Give it a title. Now I want to talk to you about some rules for slide preparation. I'm telling you the title doesn't have to be up there. By reducing the number of words on a slide and allowing people to pay more attention to me — unless what's written on this line. I mentioned it before: We have only one language processor and we can either use it to read stuff or listen to the speaker.

So if you have too many words on this slide it forces people in the audience to read this stuff and not listen.

A student of mine did an experiment a few years ago. He taught some students web-based programming ideas. Half the information was on slides. He said the other half. Then for a control group you reverse that. Question: What is it the subjects — freshmen in his fraternity — what are the subjects remember best? What he said or what they read on slides?

And the answer is what they read on the slide when their slides have a lot of material on them. They pay attention to the speaker.

In fact, no reason even for those bullets.

So the too-many-words problem is a consequence of a crime Microsoft has committed by allowing you to use fonts that are too small.

So usually I'll have a sample slide like this that can be used to determine what the minimum font size is easily legible. What do you think of those? What's up right? Yeah, maybe 40 or 50. Yeah, he says 40 or 50. I think it's about right. 35 is too small — not necessarily because you can't read it but because you're probably using it to get too many words on the slide.

What other crimes do we have? Well, we have laser pointer crime. And for that, in old days when we didn't have lasers, people would go waving these things around. And pretty soon it became almost like a baton swirling contest.

So here's what I recommended in the old days for dealing with this kind of pointer. Just an example — use of a prop. Jim Glass up there saw this talk about 20 years ago and said, "Oh yeah, I remembered that talk — that one where you broke the pointer." Amazing how props tend to be the things that are remembered.

Well now we don't have pointers anymore. We've got laser pointers. It's a wonder more people aren't driven into epileptic fits over this sort of stuff. But here is what tends to happen: Look at that lovely recursive picture. And I can become part of it by putting that laser beam right on the back of my head up there.

And what do you see? You see the back of my head. I have no eye contact, no engagement, nothing.

I was sitting with a student watching a talk one day. She said, "You know what? We could all leave. He wouldn't know."

So what happens when you use a laser pointer without turning your head, pointing at something? And when you do that you lose contact with the audience. You don't want to do it.

So what do you do if you need to identify something in your image and you don't point with a laser? This is what you do: Put an arrow on there. Look at guy number one at the end of arrow number one. You don't need to have a laser pointer. Too heavy.

Crime when people ask me to review a presentation: I print it out and lay it on the table. When they do that it's easy to see whether the talk is too heavy — too much text, not enough air, white space, imagery.

This is a good example — way too heavy. The presenter has taken advantage of small font sizes to get as much on a slide as he wanted.

Lots of other crimes here but it is too heavy.

What I want to illustrate — so by contrast, another talk when I gave a few years ago. It wasn't a deeply technical talk but I show it to you because there is air in it — mostly pictures of things. Three or four slides that have text on them. When they come, those give the audience time to read them and they are there because it might have some historical significance.

The first slide with a lot of text on it is an extraction from the 1957 proposal for the 1957 AI conference at Dartmouth. Extraordinary interesting event. In that historical extraction from the proposal helps drive that point home.

What else have we got here? Oh yeah — your vocabulary word for today. This is an apex legomenon. But it means this kind of slide you can get away with exactly once in your presentation.

This is a slide that got some currency some years ago because it shows the complexity of governing in Afghanistan by showing how impossibly complex it is. Something you and any audience can't understand. And that's the point. But you can't have many of these. You can have one per work — one for presentation, one for paper, one for book. That's what an apex legomenon is. And this is an example of it.

Well I showed some crimes. So you might be asking: Do these crimes actually occur? So they do.

There's the hands-in-the-pockets crime. There is a time and place selection. Here this is how you get to Bartos Theater. First thing that you do is get on these steps over the Media Lab and you cross this large open space, then turn right down this corridor. At this point whenever I go in there I wonder if there are torture implements around the corner when you get into the dark gloomy place. So it's well named what they call the Bartos Theater because it's a place where you can watch a movie but is not a place where you can give a talk.

Now on the subject of "doesn't happen" — here's a talk I attended a while back in Stata. Note that the speaker is far away from this line. Speaker is using a laser pointer. And you say to me: Well what is happening here? By way of the 80th sliding presentation. Notice distance with words. This is the first of 10 conclusion slides.

So what's the audience reaction? That's the sponsor of the meeting. He was reading his email. This is a co-sponsor of the meeting examining the lunch menu. About this person — this person looks like paying attention but just because it's a still picture. If you were to see the video then you would say something like this.

So yeah, it does happen.

Well now that's a quick review of tools.

I want to talk about some special cases. We could talk a little bit about informing — or say another way, doing what I'm doing now. But I'll just say a few words about them. And that kind of presentation you want to start with the promise like I did for this hour that we're going now.

Then it comes to the question of how do you inspire people? I've given this talk for a long time and a few years ago our department chairman said, "Would you please give this talk to new faculty? Be sure to emphasize what it takes to inspire students." And strangely I hadn't thought about that question before. So I started a survey. I talked to some of my incoming freshman advisees and I talked with senior faculty at everything between about how they've been inspired.

What I found from the incoming freshmen is that they were inspired by some high school teacher who told them they could do it. What I found in the senior faculty: They were inspired by someone who helped them see a problem in a new way.

Then what I saw from everyone is that they were inspired when someone exhibited passion about what they are doing.

Yeah, so that's one way to be inspiring. It's easy for me because — you know, I do artificial intelligence. And how can you not be interested in artificial intelligence? If you're not interested in artificial intelligence you are probably not interested in interesting things.

So when I am lecturing my AI class it's natural for me to talk about what is cool and how exciting some new idea is. That is the kind of expression of passion that makes a difference while informing.

With respect to this question of inspiring — oh yeah, during this promise phase you can also express how cool stuff is. Let me give an example of a lecture that starts this way. I'm talking about resource allocation. It's the same sort of stuff you would think when your source is the same sort of ideas you need if you're allocating aircraft to flight schedule or trying to schedule a factory or something like that.

But the example is putting colors on a map of the United States. This is what I show in the beginning of class. This is a way to do that coloring.

And you might say: Well why don't we wait till it finishes? Would you like to do that? No. But we're not going to wait until it finishes because the sun will have exploded and consumed the Earth before this program finishes.

With a slight adjustment how the program works — which I tell my students you will understand in the next 50 minutes — this is what you get. Isn't that cool? Yeah. We got to be amazed by stuff that takes a computation from longer than the lifetime of the solar system into a few seconds.

So that's what I mean by providing a promise up front and expressing some passion about what we're talking about.

Well, last item in this little block here has to do with what people think you do at MIT. You ask faculty what the most important purpose is and I will say: Well it's the most important thing that I do is teach people how to think.

Then he says: Oh that's great. How do you teach people how to think?

Blank stare. No one can quite respond to that part. Natural next question: So how do you teach people how to think?

Well I believe that we are storytelling animals and that we start developing our story understanding and manipulating skills with fairy tales in childhood and continue on through professional schools like law, business, medicine, engineering — everything. And we continue doing that throughout life.

So if that is what thinking is all about, then when you want to teach people how to think you provide them the stories they need to know, the questions they need to ask about those stories, mechanisms for analyzing the story, ways of putting stories together and evaluating how reliable the story is. And that's what I think you need to do when you teach people how to think.

But it's all about education. And many here not necessarily with that — rather for this part: For persuading. Which breaks down into several categories: Oral exams (not shown), job talks, getting famous.

I won't say much about oral exams other than the fact that they used to be a lot scarier than today. In the old days, reading literature and foreign language was part of that. There was a high failure rate. When you look back on those failures, the most usual reason for people failing an oral exam is failure to situate and failure to practice. By situate I mean it's important to talk about your research in context.

This is a problem that has been pursued all over the world. There has not been any progress before me for 30 years. Everyone is looking for a solution because it will have impact on so many other things such as situating time and place and feel.

Then as far as practice is concerned — yes, practice is important. But that doesn't mean showing your slides to the people who share an office with you. The problem is that if you know what you're doing they will hallucinate there is material in their presentation it isn't there.

A variant on this thing by the way: As your faculty supervisor is not a very good person to help debug a talk because they know what you're doing and they will hallucinate there is material in your presentation that isn't there.

So you need to get together some friends who don't know what you're doing and have them — well, start the practice session by saying: If you can't make me cry I won't value you as a friend anymore. Then when we get to the faculty on an early exam it will be easy.

The difficulty or the amount of flack from somebody proportional to age of the older somebody is. The more they understand where in the world. But young people are trying to show old people how smart they are. So it will be vicious. Whenever you have an opportunity of an examining committee that is full with great hair — that's what you want.

Well just a word or two about something I haven't listed here. Let us get into job talks.

So I was sitting at a bar many years ago in San Diego. I was a member of the Navy Science Board and I was sitting with my colleagues on the board — Delores Edder from University of Colorado. She made me jealous that I could spit because she had written 21 books but only 17. Then another one is Bill Weldon from Texas. He was an electromagnetism guy who knew how to use rail guns to drive steel rod through tank armor. These are interesting people.

So I said: What do you look for in a faculty candidate? And within one microsecond Delores said: After they show us they have got some kind of vision. Quickly followed by Bill who said that he had to show us that they've done something.

Oh — sounds good I said. Then said to them: How long does the candidate have to establish these two things? What do you think? Compare your answer to theirs: Five minutes.

So if you haven't expressed vision and told people that you've done something in five minutes — here are you already lost.

So we have to be able to do that. Let me just mention a couple of things in connection here which is in the video: The problem is understanding the nature of human intelligence. And the approach is asking questions about what makes us different from chimpanzees and Neanderthals. Is it merely a matter of quantity or just a little bit smarter in some continuous way? Or do we have something that's fundamentally different?

But chimpanzees and Neanderthals either — and the answer is yes we do have something different. We are symbolic creatures. And because we're symbolic creatures we can build symbolic descriptions of relations in events, string them together to make stories. And because we can make stories that's what makes us different.

So that is my stump speech — how I start most of my talks on personal research.

How do you express the notion that you have done something? By listing the steps that need to be taken in order to achieve a solution of this problem. You don't have to do all those steps but here is what needs to be done.

An example: We need to specify some behavior and enumerate constraints that make it possible to deal with that behavior. We have to implement a system because we're engineers and we don't think that we understood something unless we can build it. And we built such a system about to demonstrate today.

An example of enumerating a series of steps needed to realize a vision.

So then blah blah blah blah blah blah blah blah blah blah blah. Conclude by enumerating your contributions.

It's kind of a mirror of these steps and helps to establish that you've done something.

So there is a kind of general purpose framework for doing a technical talk.

Now only a few more things left today.

Getting famous is the next item on our agenda because once you've got the job you need to think about how you are going to be recognized.

From what? Do first of all: Why should you care about getting famous?

I thought about this in connection with a fundraising event I attended once for raising money to save Venice from going under water and having all of its art destroyed. Anyway I was sitting here at JC. Was sitting here — that is Julia, the late Julia Child. As even he wore on more and more people would come up and ask Julia to autograph something or express a feeling that she had changed their life. And it just happened over and over again.

So eventually I turned to Julia and said: Miss Child, is it fun to be famous? And she thought about it for a second. As you get used to it. But you know what occurred to me? Never get used to being ignored.

So it's — you know, it's not the way to think about it. Your ideas are like children and you don't want them into the world in rags. So what you do is be sure that you have these techniques, mechanisms, thoughts — how to present the ideas that you have so they recognize the value that is in them.

So that's why it's a legitimate thing to concern yourself with packaging.

Now how do you get remembered? Well there's a lot. Every one of the items I'm about to articulate starts with an S. So if you want your presentation ideas to be remembered, one of the thing that you need to do is make sure that some kind of symbol is associated with the work.

So this arch example is actually from my PhD thesis many years ago. And in the course of my work at that time this work on arch learning became mildly famous. And I didn't know why it was. Only many years later that I realized the network accidentally had all of the elements on this star.

So first element: There was a kind of symbol. It's the arch itself.

Next thing you need: Some kind of slogan. A kind of phrase that provides a handle on the work. In this case the phrase was "one-shot learning." And it was one-shot because the program I wrote learned something definite from every example that is presented to it.

So in going from a model based on this configuration to something that isn't an arch based on that configuration, the program learns that it has to be on top — one-shot learning.

Such symbol, slogan.

Now we need a surprise. Yeah, the surprise is you don't need a million examples of something to learn if you do it with one example — if you're smart enough so you make use of that example appropriately.

So that was the surprise: You can learn something definite from each example.

Next item was a salient idea. Now when I say salient idea I don't mean important. What I mean is an idea that sticks out.

Because which one is it? So you need an idea that sticks out. And the idea that stuck here was the notion of "near miss." See? This isn't an arch but it doesn't miss by much.

Finally, tell the story — how you did it, why it works, why it's important.

So that's a bit on how to get famous — but ensure your work is recognized.

Well, we're almost finished because now we're down in this last item which is how to stop.

And when we come to that there's a question of: All right, what is the final slide and what are the final words?

So for the final slide let me give you some examples with possibilities.

About this one: While you might see that slide and think to yourself there are 1,000 faculty at MIT. Nice piece of work but not so much — it's only a tiny piece. If you divide by 1,000. When you show the whole gigantic list of collaborators at the end of the talk it's kind of a let-down because nobody knows who did anything significant.

Now you have to recognize your collaborators right on the first slide. All this was on the first slide — these are collaborators. So you don't want to put them at the end in a slide like this.

How about just one? It is the worst possible way to end a talk because this slide can be up there for 20 minutes. I've seen it happen. It squanders real estate and squanders an opportunity to tell people who you are. It's just what about this one? Often I've never seen anybody write it down. Also wastes opportunity.

Oh my God — even worse. All of these slides do nothing for you. They waste an opportunity for you to tell people, to leave people with who you are.

Well, about this — is a good one. Might seem so at first but here's the problem: If you say "Here are my conclusions" — it's a perfectly legitimate conclusion that nobody cares about. What they care about is what you have done. And that's why your final slide should have this label: Contributions.

It's a mirror of what I said over there about how job talks ought to be like. Sandwich in the final slide. Then when it's up while people are asking questions and filing out — here is an example from my own stump speech.

Yeah, this is what I talk about a lot. Yes — here are the things that typically demonstrate. And wait for people to read it.

Another final element there is: This is what we get out of it. So it's an example contribution slide.

All right. Now what about the other part? You know, if you've got your final slide up there as a contribution slide — somehow you have to tell people you're finished.

So let's check out a few possibilities.

One thing that you could do in the final words is you can tell a joke.

Okay — by the time we're done people have adjusted themselves to your voice parameters. They're ready for a joke.

I was sitting at another bar — this time in Austin, Texas — with a colleague of mine named Doug Lenat. And Doug is a fantastic speaker. So I said to Doug: You are a fantastic speaker. What is your secret? He said: Oh, I always finish with a joke. That way people think they have had fun all the time.

So yeah, a joke will work down there.

How about this one?

Thank you. Don't recommend it.

So we move — you will not go to hell if you conclude your talk by saying thank you. But it's a weak move. And here is why: When you say thank you — even worse, thank you for listening — it suggests that everybody has stayed that long out of politeness and they had a profound desire to be somewhere else but they were so polite they stuck it out.

That what everybody does is not necessarily the right thing.

And I like to illustrate how some talks can end without saying thank you. I like to draw from political speeches. But the ones that I've heard recently aren't so good. So what happens if we don't say it?

In our homes or is a Governor Christie who gave the Republican keynote address one year. This is at the end of his talk. Let's see what he does: And together — everybody together — we will stand up once again for American greatness, for our children and grandchildren. God bless you and God bless America.

That is some classic benediction ending: God bless you, God bless America.

Now I don't want to be partisan about this so I think better switch to the keynote address at the Democratic Convention that was delivered by Bill Clinton who knows something about how to speak.

If you believe you must vote and re-elect President Barack Obama. God bless America.

Now watch this. Let's go back a little bit and redo it. What I want you to see is that at one point he seems almost pressing his lips together forcing himself not to say thank you. Then there is another place where he does the salute. So watch those this time around.

If that is what you believe you must vote and you must re-elect President Barack Obama. God bless America.

Every special person — his lips as a sloop.

Yeah, I think that's pretty good.

Now what are we going to take away from this? Well I suppose I could include this talk by saying God bless you and God bless the Institute of Technology. But it might not work so well.

The way where you can get out of this: Don't have to say thank you. There are other things you do. And you know it — interesting things.

So in the Catholic Church and a good old Latin mass it ended with it: Ite, missa est — which translates approximately to: Okay, mass is over. You can go home now.

Of course at musical concerts you know that it's time to clap not at the end of this song but rather when the conductor goes over and shakes hands with the concert master. Those are conventions that tell you that the event is over.

So those are all possibilities for here. But one more possibility and that is that you can salute the audience. And by that I mean you could say something about how much we value your time at a place.

So I could say: Well, it's been great fun being here. It's been fascinating to see what you folks are doing here at MIT. I have been much stimulated and provoked by the kinds of questions that we ask. It's really great. And I look forward to coming back on many occasions in the future.

So a salute to the audience. If you could do this well.

There is — you know what? I'm glad you're here. And a reason is being here I think you have demonstrated an understanding that how you present and package your ideas is an important thing. I salute you for that. And I suggest to come back again and bring your friends.

End of transcript.

(Note: The video is the classic ~62-minute MIT lecture by the late Prof. Patrick Winston on public speaking/communication skills. Subtitles were extracted directly from the video file.)