Advice needed for giving a conference presentation

[AlphaSheeppig]

I've given a small number of presentations before, and they've all gone pretty well. The catch is that they have all been to my class mates, a whole lot of parents at my valediction when I graduated, and the board that currently funds my research... i.e. the sort of people who aren't really interested in details, and are easily impressed with pretty graphs and flashy videos. So I know the basics, i.e. keep it simple, lots of pictures and little writing to keep attention, speak loudly and clearly, and practice and all that.

 

The tricky bit is that my school has quite kindly agreed to fly me half way around the world in July to give a presentation at the most important conference in my field. My research has revealed a fundamental flaw in the currently accepted theories which proves the underlying assumptions completely wrong. I have absolutely no idea what alternative theory to suggest, only suggestions as to what new variables the theory would have to include. My research advisor is confident with my results, and the paper for the conference proceedings has made it through both stages of peer review with mostly positive comments, but I still can't help being very nervous about the whole thing. And the audience will be experts who will actually listen and understand what I'm saying and be able to pick out mistakes...

 

My question is, does anyone have any tips for making the whole thing go smoothly? Basically, what's the best way to present my research without everyone saying "This kid's an idiot."

[ajb]

Three points of general advice that I try to live by are:

 

1. Set the motivation and background well.
   
2. Don't get bogged down with technical details, but be careful not to say something incorrect.
   
3. Have a clear take home message.
   

[Marat]

Be prepared for flak from the scientific establishment with its academic reputations heavily invested in whatever it is that you are challenging. I was at a presentation of a very clever computer-based proof the four-color problem years ago, of which the presenters were quite confident, since it was logically unimpeachable. They weren't prepared, however, for some outraged old combinatorics professors in the audience who hadn't been able to develop a solution to the problem themselves, however, and who furiously attacked them for offering 'a proof which is no proof at all,' since they defined 'proof' to mean some process which humans could perform in a finite amount of time on a finite number of pages. As Henry Kissinger once said, 'Academic disputes are so vicious because so little depends on them.'

[timo]

My question is, does anyone have any tips for making the whole thing go smoothly? Basically, what's the best way to present my research without everyone saying "This kid's an idiot."

A few obvious, and perhaps a few less obvious ones:

• Be polite. There's no gain to give a talk in the line of "those idiots [cite here] assumed [whatever]. As I will show, this is a completely stupid assumption". In other words: try to offend the least amount of people possible.
  
• Be more passive. The experts you talk to are no idiots, so there's no need to teach them how groundbreaking your idea is. So replace "My research has revealed a fundamental flaw in the currently accepted theories which proves the underlying assumptions completely wrong" with something like "in [field], [assumption] is usually made. However, we investigated this assumptions and found that it might not be justified".
  
• Don't be too fond of yourself. No "revealed by my superior intellect". (ok, that was a really obvious one).
  
• Related: You should consider if you should really present it as your individual research. Usually, the advisor's share is at least half of the game, at least on the intellectual level (the boring work is of course done by the student). This is also the impression that most of the audience will have, anyways. After my last conference talk, one of the many interesting feedbacks I got was: "very interesting research. I consider inviting your advisor to give a talk about it in our group seminar".
  

 

While the problem mentioned by Marat is one that will appear rather frequent in various forms, it's often not really problematic. In almost any field and with almost any results, some inherent assumptions have to be made. To stay with the example: there is no problem with being honest and admitting that it's not a proof in the traditional sense, and that if one strictly does not accept computer proofs, then this is not a proof. That's really no big deal; most people in the audience are likely to accept computer proofs, or at least take them as what is usually called "an indication for X". Or to be a bit more cynic: scientists are very willing to accept an extended definition of a "proof" if it serves their current interest, e.g. when they want to publish work that foots on the four-color theorem to be correct.

Edited May 23, 2011 by timo

[imatfaal]

Nice way to build confidence there Marat!

 

Agree with all of AJB's points - would add a few more; practice (again), know your audience and make sure you know the format, if you are using materials do not trust that the organiser has even half a brain and be a control-freak, and listen real hard to the questions (it's damned easy to be totally phased by a misheard question).

 

Best way to stop them saying "this kid is an idiot!" is to be correct. If your supervisor and department have enough confidence to send you to a conference - repay that trust by being sure of your own abilities. And let us know how it went when it is all over

Edited May 23, 2011 by imatfaal

[CharonY]

Dependent on the audience it can be a little bit tricky. The stronger your evidence the better it is, though. I was in a similar spot in grad school and at least to my audience (mostly senior scientists towards the end of their careers) had to be slowly convinced. Acknowledge the existing expertise in the audience, slowly build up to the parts that could not be satisfyingly explained by existing theories.

 

Basically make them agree with each step you take and then build up to the point where you show the flaw. Do not drop a bombshell, but rather try to ease them into it. Also use "we" instead of "I", especially if you are junior (you can highlight your contributions, if appropriate during 1-1 discussions afterwards).

 

Be prepared for questions regarding alternatives, though. Based on what you said it would have to be an evasion of some sorts, which is a polite way to say "no bloody idea". But most of the time it is OK.

Edited May 23, 2011 by CharonY

[ajb]

Be prepared for flak from the scientific establishment with its academic reputations heavily invested in whatever it is that you are challenging.

 

While the problem mentioned by Marat is one that will appear rather frequent in various forms, it's often not really problematic.

 

Not that I have given many talks, but I have been to a few. Most researchers are not going to have the attitude of trying to belittle you or catch you out. Unless you make a clear mistake or say something confusing.

 

However, some people may appear a bit aggressive in their questioning. In my experience this is due to them wanting to understand your work. Never take this personally, and avoid getting worked up about it.

 

Good luck.

[timo]

Also use "we" instead of "I", especially if you are junior (you can highlight your contributions, if appropriate during 1-1 discussions afterwards).

I just wanted to highlight it. It's a variation of what I said, so I do agree with it. I'd not go as far as to use make it a strict rule "always say 'we'", but the intent is similar to mine: don't sell something as being exclusively your work which in reality is the product of collaborative work (to whatever degree).

 

However, some people may appear a bit aggressive in their questioning. In my experience this is due to them wanting to understand your work. Never take this personally, and avoid getting worked up about it.

Again, this is something I tend to agree with. My current official boss is a good example of that: she tends to ask a lot of question, even in-depth ones, when she's got the feeling that she's not fully understood what's being told. That's not necessarily a sign of disagreement of rejection, but more-so one of attempting to understand a point of view that is new and potentially interesting. So I agree in the sense that one should take questions, even questions that raise doubts about the validity of your work, as a sign of interest. But I have experienced comments on my ideas which I find hard to categorize anything else than being from a personal point of view (those comments were outside of the formal talk environment, admittedly). So I'd not make a general rule out "never take comments personally". Still, for a talk not taking comments personally is probably a good attitude because "this guy probably has some private problem with my work" is a conclusion that's better left for the resume of a talk, not for the talk itself. Remember: your audience is composed of scientist who are well capable of making their own judgment about your work and the concerns raised against it.

 

@ajb: While those "that's a novel idea, I'd like to understand it"-comments are very frequent and an important part of potential feedback, I don't think that's what Marat had in mind. Some topics in science simply are the way that there's a divide in the community, and that some people believe that some approach is valid, while others do not. An example from my field is whether a computer simulation has to treat each individual atom as a single entity, or if one can "coarse grain" the simulation and represent a certain number of atoms by a single blob. Since the average speaker presenting some result, say in this case from a coarse grained simulation, does not have the expertise to comment on this (especially not against a real expert who already took his stance on the issue for whatever reason), all that's left is admitting "of course, you have to believe X" (with X being "coarse grained simulations appropriately representing reality" in this case).

[DrRocket]

My research has revealed a fundamental flaw in the currently accepted theories which proves the underlying assumptions completely wrong. I have absolutely no idea what alternative theory to suggest, only suggestions as to what new variables the theory would have to include.

 

You have submitted at least an abstract to the conference, and your advisors, who are undoubtedly with communication with others in the field, know of your findings.

 

The bombshell is already out.

 

All that you need to do is present your results and the data and logic supporting it in a cogent manner. If you are correct, then your result is a major result in your field (whatever it is), and it will be recognized as such. Being correct counts for more than any number of opinions and any amount of ruffled feathers.

 

Don't let Marat's "ghost story" frighten you. Appel and Haken announced their solution to the four color theorem at a conference in 1976. I don't know what was said by the audience at that conference, but I do know that combinatorists at my institution were intrigued and impressed by their technique of reducing the problem to a large but finite number of cases that could be checked by computer.

 

Their complete proof, including microfiche, was published in the Illinois Journal of Mathematics, a prestigious journal, in 1977. Because their result was the first major theorem the proof of which involved computers, there was some controversy. Mathematicians, me included, prefer proofs that can be understood without recourse to checking so many cases that a computer is required. However, combinatorial theorems can simply require such methods. The reputation of Appel and Haken was never in doubt. Their reduction of the 4-color theorem to a finite number of cases, even a huge finite number, remains an impressive accomplishment.

[Marat]

There's a story that Wolfgang Pauli was always rude to anyone who gave a presentation in front of him at a conference, calling him and idiot and a fool for what he was saying. One researcher, knowing Pauli would object to his results about to be presented, went to speak with Pauli privately about them prior to the talk. Pauli told him then emphatically that he was a fool and an idiot, so when the conference came, Pauli just muttered under his breath, feeling that there was no point in telling the fool once again how dumb he was.

[ajb]

Still, for a talk not taking comments personally is probably a good attitude because "this guy probably has some private problem with my work" is a conclusion that's better left for the resume of a talk, not for the talk itself.

 

The last thing you want is a public argument that gets personal. However, that is how it used to be done!

 

The other extrema I have encountered is total lack of questions and the audience leaving straight after the talk rather than hanging about. The talk in question was not presented badly, just the topic was invalid and simply wrong. This won't happen to AlphaSheeppig.

 

Then we all have come across "horror stories" like Pauli and others. This I believe is rare and should not be a worry.

[DrRocket]

Then we all have come across "horror stories" like Pauli and others. This I believe is rare and should not be a worry.

 

I know of talks where the speaker was shown to be wrong, but no horror stories outside of folklore.

 

At a special session of the Annual American Mathematical Society/Mathematical Society of America Meeting (I gave a talk in this session also) I saw a gentleman who is now a full professor at Purdue be shown by another gentleman in the audience, that his theorem was incorrect. It was all very polite, and the fellow who gave the talk told me quite honestly that the gentleman in the audience was absolutely correct. There was no animosity and they still get along very well.

 

The fellow who was corrected felt a bit embarrassed, but there was no lasting damage. The audience was not large, but included several "big shots" -- Lou Auslander, Cal Moore, Joe Wolf, Roger Howe, .... None of them seemed to considered the mistake a tragedy or to think less of the presenter.

 

Anyone who has not made a mistake has probably never tried anything difficult or produced anything of note. Einstein made lots of them.

[AlphaSheeppig]

Thanks for the advice, everyone. I guess it depends on the audience and how attached they are to the old ideas. I hope there are no tricky questions, because I don't think quickly on my feet. I'll let you all know how it goes. The conference is in July though.

 

You have submitted at least an abstract to the conference

An abstract and a six page paper, in fact, which has been reviewed, and got the comment "there should be some confidence in your results"... which I'm hoping is true. There's always that nagging feeling that my advisor just accepts my findings because he trusts my calculations, and that the paper was accepted based on his name as a co-author.

[ajb]

Anyone who has not made a mistake has probably never tried anything difficult or produced anything of note. Einstein made lots of them.

 

I have made a few mistakes in the past, some more critical than others. I think mistakes and misunderstanding are simply part of research work. To some extent almost unavoidable when pushing frontiers. I hope I am now not inadvertently to Einstein as this is one of the signs of a quack!

 

 

@AlphaSheeppig Please do let us know how it all went. Best of luck.

[AlphaSheeppig]

The presentation went surprisingly well. The two presentations before me were really bad, so any nervousness I had been feeling vanished. The questions were tricky, but I was prepared for them... They were the generic sort though - along the line of asking me to elaborate on how I validated my simulations, the guy that kept asking obvious questions (to test whether I understood what I was saying, I guess), and then the one who asked me to speculate on where results from some other unrelated case would fit into my data - ironically, it was the exact same thing my advisor asked me when I first started getting results. Thanks to everyone who gave advice. All of it helped.

 

To anyone who stumbles across this at a later time looking for advice, the best advice I can possibly give after having been through it once is to have in depth knowledge what you are talking about.

[imatfaal]

Glad it went well! Are they publishing the proceedings?

[JohnB]

Second!

 

Don't keep us in suspense. You've challenged the basic assumptions in your field but you haven't told us what the field or assumptions were.

 

Do tell. Or at least advise when there will be a paper to be read. I can't speak for anyone else, but I'm as curious as hell.

 

It's exciting!