Ed Tufte [Was Re: Let tonight be advice to you all....]

Rohit Khare (khare@pest.w3.org)
Sat, 10 Feb 96 23:59:45 -0500

[First, Adam stands guilty once again of Poor Subject Lining, and gluing
together two separate topics in a single message. I hereby increase his
punishment by 6 hours. Case Dismissed. -- RK]

Adam brought in good bits with Tufte's talk summary (by Ted Romer,

Now, Adam and I both attended Tufte's seminars in LA, but I neglected to
raise my dissent with a few of these 'rules'. I've now become a pretty decent
public speaker, and I violate a few of these rules constitutionally...

> 7. Avoid overheads.

Now, I use one slide per minute. This is astonishingly high, I will admit;
call it an MTV mentality. Now, this may have to do with the kinds of talks I
give, which are usually insanely technical and detail-oriented, rather than
general and exhortatory.

Frankly, almost all of these slides are text. For an example, see

> * An overhead can convey only a fraction of the information content
> of printed material.

This is the *entire point* when the 'printed material' is a 20 page spec
written in standardese. The slides are a narrative story that 'tours' a
technical area, stopping at all the sights and providing context. I use slides
to remind people where we are...

> * If your slides are just misshapen trapezoidal note cards for your benefit,
> why not speak from note cards?

See above: it gives people the structure that's so expensive to establish (in
terms of bandwidth). If not, I think I'd have to cut the content by half to
incorporate the here's-what-I'm-going-to-tell-you // tell-you //

> * If you want to give the audience something to pay attention to when
you're > saying "er, ah", give them handouts.

I don't adhere to a rule about handing out copies of the slides in advance or
not. If it's a large talk, I try to prepare a 'roadmap' outline that includes
the upper parts of the tree, but not the entire thing. Since it's my job, I
dump everything to the Web, so I'm not worried about canonical-copies, I can
just provide a URL.

Besides, nowadays my audience is crawling with laptops, cell phones, cell
modems, Game Boys, etc... they have *enough* to fidget with without my giving
away my talk! :-)

> 11. Work hard.

I have settled on what might be a self-fulfiling prophecy: It takes an order
of magnitude more time to prepare than to present. Sounds extremely high,
doesn't it? But that's what I measure. For me, that comes to 10 minutes a
slide. And thats not including 'developing the details' -- this is all AFTER
you understand the material and scope cold.

> 14. Show your enthusiasm. Don't hide behind a lectern.
> Use gestures.
> * Walk around, directly engaging audience members'
> attention. (Tufte did this remarkably effectively.)

I CANNOT emphasize this enough. Some of what I think are my 'worst' talks
seem to be perceived well, because I am searching for feedback from the
audience. I wouldn't talk impassively to one person at a table, so I certainly
wouldn't to a large crowd...

> 4. Speak from notes, don't read a prepared text.

This is one I'm poor at. Since I use slides, I need to keep the same state as
my audience. I've also rehearsed my talk to the slides. But I cannot possibly
seem to synchronize my paper speaker's notes and acetate slides (this is
Tufte's point exactly). Most always, though, I present on-line, and I work
best if I can see the laptop while at the lectern.

The subtext of 4, though is spontaneity and connection. Modulo time wasted
craning over my shoulder to read slides, I think I try to do that well.

> 16. Avoid dehydration.
> * Avoid dehydrating beverages: caffeine and alcohol.

I'm sorry, I think Caffeine Does Your Body Right (TM). I have never deviated
from my two-soda minimum.

Most importantly, the resulting diuretic effect is very, very important to
meeting Rule #15:

> 15. Finish early. Everyone will be happier.

I can very easily stand corrected. I would be glad to accept divergent
opinions and criticisms -- let's talk about it...

Rohit Khare