WEBVTT

00:00:00.709 --> 00:00:02.235
Type is something we consume

00:00:02.235 --> 00:00:04.157
in enormous quantities.

00:00:04.157 --> 00:00:05.239
In much of the world,

00:00:05.239 --> 00:00:07.107
it's completely inescapable.

00:00:07.107 --> 00:00:10.130
But few consumers are concerned to know

00:00:10.130 --> 00:00:12.462
where a particular typeface came from

00:00:12.462 --> 00:00:15.226
or when or who designed it,

00:00:15.226 --> 00:00:18.579
if, indeed, there was any human agency involved

00:00:18.579 --> 00:00:21.376
in its creation, if it didn't just sort of materialize

00:00:21.376 --> 00:00:25.088
out of the software ether.

NOTE Paragraph

00:00:25.088 --> 00:00:28.613
But I do have to be concerned with those things.

00:00:28.613 --> 00:00:30.130
It's my job.

00:00:30.130 --> 00:00:32.340
I'm one of the tiny handful of people

00:00:32.340 --> 00:00:34.454
who gets badly bent out of shape

00:00:34.454 --> 00:00:37.048
by the bad spacing of the T and the E

00:00:37.048 --> 00:00:38.948
that you see there.

00:00:38.948 --> 00:00:40.390
I've got to take that slide off.

00:00:40.390 --> 00:00:42.461
I can't stand it. Nor can Chris.

00:00:42.461 --> 00:00:44.122
There. Good.

NOTE Paragraph

00:00:44.122 --> 00:00:45.890
So my talk is about the connection

00:00:45.890 --> 00:00:49.290
between technology and design of type.

00:00:49.290 --> 00:00:51.953
The technology has changed

00:00:51.953 --> 00:00:54.825
a number of times since I started work:

00:00:54.825 --> 00:00:59.486
photo, digital, desktop, screen, web.

00:00:59.486 --> 00:01:01.224
I've had to survive those changes and try

00:01:01.224 --> 00:01:03.928
to understand their implications for what I do

00:01:03.928 --> 00:01:05.319
for design.

00:01:05.319 --> 00:01:10.489
This slide is about the effect of tools on form.

00:01:10.489 --> 00:01:13.390
The two letters, the two K's,

00:01:13.390 --> 00:01:16.715
the one on your left, my right, is modern,

00:01:16.715 --> 00:01:18.126
made on a computer.

00:01:18.126 --> 00:01:20.168
All straight lines are dead straight.

00:01:20.168 --> 00:01:22.978
The curves have that kind of
mathematical smoothness

00:01:22.978 --> 00:01:26.724
that the Bézier formula imposes.

00:01:26.724 --> 00:01:29.221
On the right, ancient Gothic,

00:01:29.221 --> 00:01:33.280
cut in the resistant material of steel by hand.

00:01:33.280 --> 00:01:35.405
None of the straight lines are actually straight.

00:01:35.405 --> 00:01:37.570
The curves are kind of subtle.

00:01:37.570 --> 00:01:42.284
It has that spark of life from the human hand

00:01:42.284 --> 00:01:44.198
that the machine or the program

00:01:44.198 --> 00:01:46.131
can never capture.

00:01:46.131 --> 00:01:48.110
What a contrast.

NOTE Paragraph

00:01:48.110 --> 00:01:50.689
Well, I tell a lie.

00:01:50.689 --> 00:01:53.677
A lie at TED. I'm really sorry.

00:01:53.677 --> 00:01:55.798
Both of these were made on a computer,

00:01:55.798 --> 00:01:57.724
same software, same Bézier curves,

00:01:57.724 --> 00:01:59.416
same font format.

00:01:59.416 --> 00:02:01.950
The one on your left

00:02:01.950 --> 00:02:04.204
was made by Zuzana Licko at Emigre,

00:02:04.204 --> 00:02:05.639
and I did the other one.

00:02:05.639 --> 00:02:08.995
The tool is the same, yet the letters are different.

00:02:08.995 --> 00:02:10.518
The letters are different

00:02:10.518 --> 00:02:11.724
because the designers are different.

00:02:11.724 --> 00:02:14.985
That's all. Zuzana wanted hers to look like that.

00:02:14.985 --> 00:02:18.106
I wanted mine to look like that. End of story.

00:02:18.106 --> 00:02:20.310
Type is very adaptable.

00:02:20.310 --> 00:02:23.631
Unlike a fine art, such as sculpture or architecture,

00:02:23.631 --> 00:02:27.030
type hides its methods.

00:02:27.030 --> 00:02:29.827
I think of myself as an industrial designer.

00:02:29.827 --> 00:02:31.393
The thing I design is manufactured,

00:02:31.393 --> 00:02:33.269
and it has a function:

00:02:33.269 --> 00:02:35.107
to be read, to convey meaning.

00:02:35.107 --> 00:02:36.753
But there is a bit more to it than that.

00:02:36.753 --> 00:02:38.601
There's the sort of aesthetic element.

00:02:38.601 --> 00:02:41.317
What makes these two letters different

00:02:41.317 --> 00:02:44.349
from different interpretations by different designers?

00:02:44.349 --> 00:02:46.362
What gives the work of some designers

00:02:46.362 --> 00:02:49.273
sort of characteristic personal style,

00:02:49.273 --> 00:02:51.810
as you might find in the work of a fashion designer,

00:02:51.810 --> 00:02:54.885
an automobile designer, whatever?

NOTE Paragraph

00:02:54.885 --> 00:02:56.660
There have been some cases, I admit,

00:02:56.660 --> 00:02:57.810
where I as a designer

00:02:57.810 --> 00:03:00.899
did feel the influence of technology.

00:03:00.899 --> 00:03:03.959
This is from the mid-'60s,

00:03:03.959 --> 00:03:06.194
the change from metal type to photo,

00:03:06.194 --> 00:03:07.896
hot to cold.

00:03:07.896 --> 00:03:09.093
This brought some benefits

00:03:09.093 --> 00:03:12.439
but also one particular drawback:

00:03:12.439 --> 00:03:15.112
a spacing system that only provided

00:03:15.112 --> 00:03:19.095
18 discrete units for letters

00:03:19.095 --> 00:03:21.558
to be accommodated on.

00:03:21.558 --> 00:03:23.534
I was asked at this time to design

00:03:23.534 --> 00:03:26.272
a series of condensed sans serif types

00:03:26.272 --> 00:03:29.435
with as many different variants as possible

00:03:29.435 --> 00:03:33.282
within this 18-unit box.

00:03:33.282 --> 00:03:34.921
Quickly looking at the arithmetic,

00:03:34.921 --> 00:03:38.230
I realized I could only actually make three

00:03:38.230 --> 00:03:41.617
of related design. Here you see them.

00:03:41.617 --> 00:03:44.208
In Helvetica Compressed, Extra Compressed,

00:03:44.208 --> 00:03:48.019
and Ultra Compressed, this rigid 18-unit system

00:03:48.019 --> 00:03:49.592
really boxed me in.

00:03:49.592 --> 00:03:51.425
It kind of determined the proportions

00:03:51.425 --> 00:03:53.714
of the design.

00:03:53.714 --> 00:03:57.744
Here are the typefaces, at least the lower cases.

00:03:57.744 --> 00:04:00.436
So do you look at these and say,

00:04:00.436 --> 00:04:03.575
"Poor Matthew, he had to submit to a problem,

00:04:03.575 --> 00:04:07.382
and by God it shows in the results."

00:04:07.382 --> 00:04:08.689
I hope not.

00:04:08.689 --> 00:04:10.930
If I were doing this same job today,

00:04:10.930 --> 00:04:13.756
instead of having 18 spacing units,

00:04:13.756 --> 00:04:16.840
I would have 1,000.

00:04:16.840 --> 00:04:19.293
Clearly I could make more variants,

00:04:19.293 --> 00:04:23.989
but would these three members
of the family be better?

00:04:23.989 --> 00:04:25.841
It's hard to say without actually doing it,

00:04:25.841 --> 00:04:27.548
but they would not be better in the proportion

00:04:27.548 --> 00:04:30.637
of 1,000 to 18, I can tell you that.

00:04:30.637 --> 00:04:32.575
My instinct tells you that any improvement

00:04:32.575 --> 00:04:35.653
would be rather slight, because they were designed

00:04:35.653 --> 00:04:38.574
as functions of the system they were designed to fit,

00:04:38.574 --> 00:04:40.983
and as I said, type is very adaptable.

00:04:40.983 --> 00:04:43.770
It does hide its methods.

00:04:43.770 --> 00:04:46.452
All industrial designers work within constraints.

00:04:46.452 --> 00:04:48.944
This is not fine art.

NOTE Paragraph

00:04:48.944 --> 00:04:50.822
The question is, does a constraint

00:04:50.822 --> 00:04:53.489
force a compromise?

00:04:53.489 --> 00:04:55.000
By accepting a constraint,

00:04:55.000 --> 00:04:57.449
are you working to a lower standard?

00:04:57.449 --> 00:04:59.411
I don't believe so, and I've always been encouraged

00:04:59.411 --> 00:05:01.535
by something that Charles Eames said.

00:05:01.535 --> 00:05:03.080
He said he was conscious of working

00:05:03.080 --> 00:05:04.118
within constraints,

00:05:04.118 --> 00:05:07.344
but not of making compromises.

00:05:07.344 --> 00:05:09.905
The distinction between a constraint

00:05:09.905 --> 00:05:12.280
and a compromise is obviously very subtle,

00:05:12.280 --> 00:05:17.951
but it's very central to my attitude to work.

NOTE Paragraph

00:05:17.951 --> 00:05:20.842
Remember this reading experience?

00:05:20.842 --> 00:05:22.267
The phone book. I'll hold the slide

00:05:22.267 --> 00:05:26.802
so you can enjoy the nostalgia.

00:05:26.802 --> 00:05:29.548
This is from the mid-'70s early trials

00:05:29.548 --> 00:05:32.177
of Bell Centennial typeface I designed

00:05:32.177 --> 00:05:33.951
for the U.S. phone books,

00:05:33.951 --> 00:05:37.270
and it was my first experience of digital type,

00:05:37.270 --> 00:05:41.440
and quite a baptism.

00:05:41.440 --> 00:05:43.039
Designed for the phone books, as I said,

00:05:43.039 --> 00:05:46.407
to be printed at tiny size on newsprint

00:05:46.407 --> 00:05:48.725
on very high-speed rotary presses

00:05:48.725 --> 00:05:51.477
with ink that was kerosene and lampblack.

00:05:51.477 --> 00:05:55.318
This is not a hospitable environment

00:05:55.318 --> 00:05:58.520
for a typographic designer.

00:05:58.520 --> 00:06:00.419
So the challenge for me was to design type

00:06:00.419 --> 00:06:01.920
that performed as well as possible

00:06:01.920 --> 00:06:06.665
in these very adverse production conditions.

00:06:06.665 --> 00:06:09.520
As I say, we were in the infancy of digital type.

00:06:09.520 --> 00:06:12.339
I had to draw every character by hand

00:06:12.339 --> 00:06:14.054
on quadrille graph paper --

00:06:14.054 --> 00:06:16.006
there were four weights of Bell Centennial —

00:06:16.006 --> 00:06:19.359
pixel by pixel, then encode
them raster line by raster line

00:06:19.359 --> 00:06:20.340
for the keyboard.

00:06:20.340 --> 00:06:24.764
It took two years, but I learned a lot.

00:06:24.764 --> 00:06:26.394
These letters look as though they've been chewed

00:06:26.394 --> 00:06:27.838
by the dog or something or other,

00:06:27.838 --> 00:06:29.780
but the missing pixels at the intersections

00:06:29.780 --> 00:06:31.385
of strokes or in the crotches

00:06:31.385 --> 00:06:34.559
are the result of my studying the effects

00:06:34.559 --> 00:06:37.505
of ink spread on cheap paper

00:06:37.505 --> 00:06:41.242
and reacting, revising the font accordingly.

00:06:41.242 --> 00:06:44.492
These strange artifacts are designed to compensate

00:06:44.492 --> 00:06:47.244
for the undesirable effects of scale

00:06:47.244 --> 00:06:49.530
and production process.

00:06:49.530 --> 00:06:52.376
At the outset, AT&T had wanted

00:06:52.376 --> 00:06:55.617
to set the phone books in Helvetica,

00:06:55.617 --> 00:06:57.282
but as my friend Erik Spiekermann said

00:06:57.282 --> 00:06:59.785
in the Helvetica movie, if you've seen that,

00:06:59.785 --> 00:07:01.820
the letters in Helvetica were designed to be

00:07:01.820 --> 00:07:04.561
as similar to one another as possible.

00:07:04.561 --> 00:07:07.835
This is not the recipe for legibility at small size.

00:07:07.835 --> 00:07:10.425
It looks very elegant up on a slide.

00:07:10.425 --> 00:07:12.615
I had to disambiguate these forms

00:07:12.615 --> 00:07:15.615
of the figures as much as possible in Bell Centennial

00:07:15.615 --> 00:07:17.910
by sort of opening the shapes up, as you can see

00:07:17.910 --> 00:07:20.823
in the bottom part of that slide.

NOTE Paragraph

00:07:20.823 --> 00:07:23.480
So now we're on to the mid-'80s,

00:07:23.480 --> 00:07:26.036
the early days of digital outline fonts,

00:07:26.036 --> 00:07:28.393
vector technology.

00:07:28.393 --> 00:07:30.431
There was an issue at that time

00:07:30.431 --> 00:07:32.287
with the size of the fonts,

00:07:32.287 --> 00:07:35.171
the amount of data that was required to find

00:07:35.171 --> 00:07:40.141
and store a font in computer memory.

00:07:40.141 --> 00:07:41.799
It limited the number of fonts you could get

00:07:41.799 --> 00:07:44.789
on your typesetting system at any one time.

00:07:44.789 --> 00:07:48.938
I did an analysis of the data,

00:07:48.938 --> 00:07:51.462
and found that a typical serif face

00:07:51.462 --> 00:07:52.921
you see on the left

00:07:52.921 --> 00:07:54.937
needed nearly twice as much data

00:07:54.937 --> 00:07:57.473
as a sans serif in the middle

00:07:57.473 --> 00:07:59.535
because of all the points required

00:07:59.535 --> 00:08:04.043
to define the elegantly curved serif brackets.

00:08:04.043 --> 00:08:07.477
The numbers at the bottom of the slide, by the way,

00:08:07.477 --> 00:08:09.179
they represent the amount of data

00:08:09.179 --> 00:08:12.984
needed to store each of the fonts.

00:08:12.984 --> 00:08:15.150
So the sans serif, in the middle,

00:08:15.150 --> 00:08:18.114
sans the serifs, was much more economical,

00:08:18.114 --> 00:08:20.313
81 to 151.

NOTE Paragraph

00:08:20.313 --> 00:08:23.970
"Aha," I thought. "The engineers have a problem.

00:08:23.970 --> 00:08:26.205
Designer to the rescue."

NOTE Paragraph

00:08:26.205 --> 00:08:28.552
I made a serif type, you can see it on the right,

00:08:28.552 --> 00:08:30.499
without curved serifs.

00:08:30.499 --> 00:08:32.915
I made them polygonal, out
of straight line segments,

00:08:32.915 --> 00:08:34.898
chamfered brackets.

00:08:34.898 --> 00:08:39.266
And look, as economical in data as a sans serif.

00:08:39.266 --> 00:08:41.565
We call it Charter, on the right.

NOTE Paragraph

00:08:41.565 --> 00:08:43.535
So I went to the head of engineering

00:08:43.535 --> 00:08:45.993
with my numbers, and I said proudly,

00:08:45.993 --> 00:08:48.121
"I have solved your problem."

NOTE Paragraph

00:08:48.121 --> 00:08:51.834
"Oh," he said. "What problem?"

NOTE Paragraph

00:08:51.834 --> 00:08:53.480
And I said, "Well, you know, the problem

00:08:53.480 --> 00:08:56.897
of the huge data you require
for serif fonts and so on."

NOTE Paragraph

00:08:56.897 --> 00:09:00.444
"Oh," he said. "We solved that problem last week.

00:09:00.444 --> 00:09:02.600
We wrote a compaction routine that reduces

00:09:02.600 --> 00:09:05.180
the size of all fonts by an order of magnitude.

00:09:05.180 --> 00:09:07.168
You can have as many fonts on your system

00:09:07.168 --> 00:09:08.726
as you like."

NOTE Paragraph

00:09:08.726 --> 00:09:11.340
"Well, thank you for letting me know," I said.

NOTE Paragraph

00:09:11.340 --> 00:09:12.970
Foiled again.

00:09:12.970 --> 00:09:15.015
I was left with a design solution

00:09:15.015 --> 00:09:19.488
for a nonexistent technical problem.

NOTE Paragraph

00:09:19.488 --> 00:09:21.957
But here is where the story sort
of gets interesting for me.

00:09:21.957 --> 00:09:24.552
I didn't just throw my design away

00:09:24.552 --> 00:09:25.953
in a fit of pique.

00:09:25.953 --> 00:09:27.690
I persevered.

00:09:27.690 --> 00:09:29.858
What had started as a technical exercise

00:09:29.858 --> 00:09:33.122
became an aesthetic exercise, really.

00:09:33.122 --> 00:09:36.171
In other words, I had come to like this typeface.

00:09:36.171 --> 00:09:38.490
Forget its origins. Screw that.

00:09:38.490 --> 00:09:40.980
I liked the design for its own sake.

00:09:40.980 --> 00:09:43.363
The simplified forms of Charter

00:09:43.363 --> 00:09:45.446
gave it a sort of plain-spoken quality

00:09:45.446 --> 00:09:46.997
and unfussy spareness

00:09:46.997 --> 00:09:49.487
that sort of pleased me.

00:09:49.487 --> 00:09:52.040
You know, at times of technical innovation,

00:09:52.040 --> 00:09:53.560
designers want to be influenced

00:09:53.560 --> 00:09:55.296
by what's in the air.

00:09:55.296 --> 00:09:57.527
We want to respond. We want to be pushed

00:09:57.527 --> 00:10:00.938
into exploring something new.

00:10:00.938 --> 00:10:03.837
So Charter is a sort of parable for me, really.

00:10:03.837 --> 00:10:07.627
In the end, there was no hard and fast causal link

00:10:07.627 --> 00:10:10.820
between the technology and the design of Charter.

00:10:10.820 --> 00:10:14.582
I had really misunderstood the technology.

00:10:14.582 --> 00:10:17.910
The technology did suggest something to me,

00:10:17.910 --> 00:10:20.027
but it did not force my hand,

00:10:20.027 --> 00:10:22.744
and I think this happens very often.

NOTE Paragraph

00:10:22.744 --> 00:10:25.370
You know, engineers are very smart,

00:10:25.370 --> 00:10:26.926
and despite occasional frustrations

00:10:26.926 --> 00:10:28.453
because I'm less smart,

00:10:28.453 --> 00:10:30.180
I've always enjoyed working with them

00:10:30.180 --> 00:10:32.258
and learning from them.

00:10:32.258 --> 00:10:34.600
Apropos, in the mid-'90s,

00:10:34.600 --> 00:10:37.291
I started talking to Microsoft

00:10:37.291 --> 00:10:39.679
about screen fonts.

00:10:39.679 --> 00:10:42.100
Up to that point, all the fonts on screen

00:10:42.100 --> 00:10:44.853
had been adapted from previously existing

00:10:44.853 --> 00:10:47.230
printing fonts, of course.

00:10:47.230 --> 00:10:49.733
But Microsoft foresaw correctly

00:10:49.733 --> 00:10:51.883
the movement, the stampede

00:10:51.883 --> 00:10:54.670
towards electronic communication,

00:10:54.670 --> 00:10:56.700
to reading and writing onscreen

00:10:56.700 --> 00:10:59.833
with the printed output as being sort of secondary

00:10:59.833 --> 00:11:02.056
in importance.

NOTE Paragraph

00:11:02.056 --> 00:11:05.635
So the priorities were just tipping at that point.

00:11:05.635 --> 00:11:07.829
They wanted a small core set of fonts

00:11:07.829 --> 00:11:11.134
that were not adapted but designed for the screen

00:11:11.134 --> 00:11:13.707
to face up to the problems of screen,

00:11:13.707 --> 00:11:17.549
which were their coarse resolution displays.

00:11:17.549 --> 00:11:21.080
I said to Microsoft, a typeface designed

00:11:21.080 --> 00:11:22.649
for a particular technology

00:11:22.649 --> 00:11:26.024
is a self-obsoleting typeface.

00:11:26.024 --> 00:11:28.118
I've designed too many faces in the past

00:11:28.118 --> 00:11:31.697
that were intended to mitigate technical problems.

00:11:31.697 --> 00:11:34.532
Thanks to the engineers, the
technical problems went away.

00:11:34.532 --> 00:11:37.021
So did my typeface.

00:11:37.021 --> 00:11:40.152
It was only a stopgap.

00:11:40.152 --> 00:11:41.693
Microsoft came back to say that

00:11:41.693 --> 00:11:43.323
affordable computer monitors

00:11:43.323 --> 00:11:44.520
with better resolutions

00:11:44.520 --> 00:11:47.116
were at least a decade away.

00:11:47.116 --> 00:11:49.770
So I thought, well, a decade, that's not bad,

00:11:49.770 --> 00:11:52.182
that's more than a stopgap.

NOTE Paragraph

00:11:52.182 --> 00:11:54.183
So I was persuaded, I was convinced,

00:11:54.183 --> 00:11:56.505
and we went to work on what became Verdana

00:11:56.505 --> 00:11:58.177
and Georgia,

00:11:58.177 --> 00:12:00.517
for the first time working not on paper

00:12:00.517 --> 00:12:04.477
but directly onto the screen from the pixel up.

00:12:04.477 --> 00:12:08.330
At that time, screens were binary.

00:12:08.330 --> 00:12:11.370
The pixel was either on or it was off.

00:12:11.370 --> 00:12:14.225
Here you see the outline of a letter,

00:12:14.225 --> 00:12:15.692
the cap H,

00:12:15.692 --> 00:12:18.493
which is the thin black line, the contour,

00:12:18.493 --> 00:12:21.369
which is how it is stored in memory,

00:12:21.369 --> 00:12:23.039
superimposed on the bitmap,

00:12:23.039 --> 00:12:25.187
which is the grey area,

00:12:25.187 --> 00:12:27.024
which is how it's displayed on the screen.

00:12:27.024 --> 00:12:30.170
The bitmap is rasterized from the outline.

00:12:30.170 --> 00:12:32.411
Here in a cap H, which is all straight lines,

00:12:32.411 --> 00:12:34.499
the two are in almost perfect sync

00:12:34.499 --> 00:12:38.867
on the Cartesian grid.

00:12:38.867 --> 00:12:41.993
Not so with an O.

00:12:41.993 --> 00:12:44.720
This looks more like bricklaying
than type design,

00:12:44.720 --> 00:12:47.621
but believe me, this is a good bitmap O,

00:12:47.621 --> 00:12:49.636
for the simple reason that it's symmetrical

00:12:49.636 --> 00:12:52.116
in both x and y axes.

00:12:52.116 --> 00:12:54.974
In a binary bitmap, you actually can't ask

00:12:54.974 --> 00:12:56.694
for more than that.

00:12:56.694 --> 00:12:59.098
I would sometimes make, I don't know,

00:12:59.098 --> 00:13:01.454
three or four different versions of a difficult letter

00:13:01.454 --> 00:13:02.960
like a lowercase A,

00:13:02.960 --> 00:13:06.500
and then stand back to choose which was the best.

00:13:06.500 --> 00:13:08.595
Well, there was no best,

00:13:08.595 --> 00:13:11.020
so the designer's judgment comes in

00:13:11.020 --> 00:13:12.409
in trying to decide

00:13:12.409 --> 00:13:15.450
which is the least bad.

00:13:15.450 --> 00:13:17.900
Is that a compromise?

00:13:17.900 --> 00:13:19.446
Not to me, if you are working

00:13:19.446 --> 00:13:22.533
at the highest standard the technology will allow,

00:13:22.533 --> 00:13:24.742
although that standard may be

00:13:24.742 --> 00:13:27.221
well short of the ideal.

00:13:27.221 --> 00:13:28.812
You may be able to see on this slide

00:13:28.812 --> 00:13:30.974
two different bitmap fonts there.

00:13:30.974 --> 00:13:32.670
The "a" in the upper one, I think,

00:13:32.670 --> 00:13:34.583
is better than the "a" in the lower one,

00:13:34.583 --> 00:13:37.159
but it still ain't great.

00:13:37.159 --> 00:13:39.066
You can maybe see the effect better

00:13:39.066 --> 00:13:42.442
if it's reduced. Well, maybe not.

NOTE Paragraph

00:13:42.442 --> 00:13:44.838
So I'm a pragmatist, not an idealist,

00:13:44.838 --> 00:13:46.341
out of necessity.

00:13:46.341 --> 00:13:48.270
For a certain kind of temperament,

00:13:48.270 --> 00:13:49.970
there is a certain kind of satisfaction

00:13:49.970 --> 00:13:53.608
in doing something that cannot be perfect

00:13:53.608 --> 00:13:57.477
but can still be done to the best of your ability.

00:13:57.477 --> 00:14:02.334
Here's the lowercase H from Georgia Italic.

00:14:02.334 --> 00:14:04.607
The bitmap looks jagged and rough.

00:14:04.607 --> 00:14:06.466
It is jagged and rough.

00:14:06.466 --> 00:14:08.462
But I discovered, by experiment,

00:14:08.462 --> 00:14:11.679
that there is an optimum slant

00:14:11.679 --> 00:14:13.625
for an italic on a screen

00:14:13.625 --> 00:14:15.957
so the strokes break well

00:14:15.957 --> 00:14:18.440
at the pixel boundaries.

00:14:18.440 --> 00:14:21.221
Look in this example how, rough as it is,

00:14:21.221 --> 00:14:23.271
how the left and right legs

00:14:23.271 --> 00:14:25.220
actually break at the same level.

00:14:25.220 --> 00:14:28.740
That's a victory. That's good, right there.

00:14:28.740 --> 00:14:31.918
And of course, at the lower depths,

00:14:31.918 --> 00:14:33.841
you don't get much choice.

00:14:33.841 --> 00:14:38.886
This is an S, in case you were wondering.

NOTE Paragraph

00:14:38.886 --> 00:14:41.040
Well, it's been 18 years now

00:14:41.040 --> 00:14:43.690
since Verdana and Georgia were released.

00:14:43.690 --> 00:14:45.780
Microsoft were absolutely right,

00:14:45.780 --> 00:14:48.194
it took a good 10 years,

00:14:48.194 --> 00:14:50.474
but screen displays now do have

00:14:50.474 --> 00:14:52.927
improved spatial resolution,

00:14:52.927 --> 00:14:56.399
and very much improved photometric resolution

00:14:56.399 --> 00:14:59.553
thanks to anti-aliasing and so on.

00:14:59.553 --> 00:15:03.250
So now that their mission is accomplished,

00:15:03.250 --> 00:15:04.980
has that meant the demise

00:15:04.980 --> 00:15:06.860
of the screen fonts that I designed

00:15:06.860 --> 00:15:09.511
for coarser displays back then?

00:15:09.511 --> 00:15:12.906
Will they outlive the now-obsolete screens

00:15:12.906 --> 00:15:15.078
and the flood of new web fonts

00:15:15.078 --> 00:15:16.581
coming on to the market?

00:15:16.581 --> 00:15:18.441
Or have they established their own

00:15:18.441 --> 00:15:20.679
sort of evolutionary niche

00:15:20.679 --> 00:15:24.416
that is independent of technology?

00:15:24.416 --> 00:15:26.087
In other words, have they been absorbed

00:15:26.087 --> 00:15:29.400
into the typographic mainstream?

00:15:29.400 --> 00:15:33.058
I'm not sure, but they've had a good run so far.

00:15:33.058 --> 00:15:35.553
Hey, 18 is a good age for anything

00:15:35.553 --> 00:15:37.704
with present-day rates of attrition,

00:15:37.704 --> 00:15:39.618
so I'm not complaining.

NOTE Paragraph

00:15:39.618 --> 00:15:42.457
Thank you.

NOTE Paragraph

00:15:42.457 --> 00:15:44.634
(Applause)