Debian Codegraph @

texmf-dist/doc/generic/pgf/pgfmanual-en-guidelines.tex

raw
1 % Copyright 2018 by Till Tantau
2 %
3 % This file may be distributed and/or modified
4 %
5 % 1. under the LaTeX Project Public License and/or
6 % 2. under the GNU Free Documentation License.
7 %
8 % See the file doc/generic/pgf/licenses/LICENSE for more details.
9
10
11 \section{Guidelines on Graphics}
12
13 The present section is not about \pgfname\ or \tikzname, but about general
14 guidelines and principles concerning the creation of graphics for scientific
15 presentations, papers, and books.
16
17 The guidelines in this section come from different sources. Many of them are
18 just what I would like to claim is ``common sense'', some reflect my personal
19 experience (though, hopefully, not my personal preferences), some come from
20 books (the bibliography is still missing, sorry) on graphic design and
21 typography. The most influential source are the brilliant books by Edward
22 Tufte. While I do not agree with everything written in these books, many of
23 Tufte's arguments are so convincing that I decided to repeat them in the
24 following guidelines.
25
26 The first thing you should ask yourself when someone presents a bunch of
27 guidelines is: Should I really follow these guidelines? This is an important
28 question, because there are good reasons not to follow general guidelines. The
29 person who set up the guidelines may have had other objectives than you do. For
30 example, a guideline might say ``use the color red for emphasis''. While this
31 guideline makes perfect sense for, say, a presentation using a projector, red
32 ``color'' has the \emph{opposite} effect of ``emphasis'' when printed using a
33 black-and-white printer. Guidelines were almost always set up to address a
34 specific situation. If you are not in this situation, following a guideline can
35 do more harm than good.
36
37 The second thing you should be aware of is the basic rule of typography is:
38 ``Every rule can be broken, as long as you are \emph{aware} that you are
39 breaking a rule.'' This rule also applies to graphics. Phrased differently, the
40 basic rule states: ``The only mistakes in typography are things done in
41 ignorance.'' When you are aware of a rule and when you decide that breaking the
42 rule has a desirable effect, break the rule.
43
44
45 \subsection{Planning the Time Needed for the Creation of Graphics}
46
47 When you create a paper with numerous graphics, the time needed to create these
48 graphics becomes an important factor. How much time should you calculate for
49 the creation of graphics?
50
51 As a general rule, assume that a graphic will need as much time to create as
52 would a text of the same length. For example, when I write a paper, I need
53 about one hour per page for the first draft. Later, I need between two and four
54 hours per page for revisions. Thus, I expect to need about half an hour for the
55 creation of \emph{a first draft} of a half page graphic. Later on, I expect
56 another one to two hours before the final graphic is finished.
57
58 In many publications, even in good journals, the authors and editors have
59 obviously invested a lot of time on the text, but seem to have spend about
60 five minutes to create all of the graphics. Graphics often seem to have been
61 added as an ``afterthought'' or look like a screen shot of whatever the
62 authors's statistical software shows them. As will be argued later on, the
63 graphics that programs like \textsc{gnuplot} produce by default are of poor
64 quality.
65
66 Creating informative graphics that help the reader and that fit together with
67 the main text is a difficult, lengthy process.
68 %
69 \begin{itemize}
70 \item Treat graphics as first-class citizens of your papers. They deserve
71 as much time and energy as the text does. Indeed, the creation of
72 graphics might deserve \emph{even more} time than the writing of the
73 main text since more attention will be paid to the graphics and they
74 will be looked at first.
75 \item Plan as much time for the creation and revision of a graphic as you
76 would plan for text of the same size.
77 \item Difficult graphics with a high information density may require even
78 more time.
79 \item Very simple graphics will require less time, but most likely you do
80 not want to have ``very simple graphics'' in your paper, anyway; just
81 as you would not like to have a ``very simple text'' of the same
82 size.
83 \end{itemize}
84
85
86 \subsection{Workflow for Creating a Graphic}
87
88 When you write a (scientific) paper, you will most likely follow the following
89 pattern: You have some results/ideas that you would like to report about. The
90 creation of the paper will typically start with compiling a rough outline.
91 Then, the different sections are filled with text to create a first draft. This
92 draft is then revised repeatedly until, often after substantial revision, a
93 final paper results. In a good journal paper there is typically not be a single
94 sentence that has survived unmodified from the first draft.
95
96 Creating a graphics follows the same pattern:
97 %
98 \begin{itemize}
99 \item Decide on what the graphic should communicate. Make this a
100 conscious decision, that is, determine ``What is the graphic supposed
101 to tell the reader?''
102 \item Create an ``outline'', that is, the rough overall ``shape'' of the
103 graphic, containing the most crucial elements. Often, it is useful to
104 do this using pencil and paper.
105 \item Fill out the finer details of the graphic to create a first draft.
106 \item Revise the graphic repeatedly along with the rest of the paper.
107 \end{itemize}
108
109
110 \subsection{Linking Graphics With the Main Text}
111
112 Graphics can be placed at different places in a text. Either, they can be
113 inlined, meaning they are somewhere ``in the middle of the text'' or they can
114 be placed in stand-alone ``figures''. Since printers (the people) like to have
115 their pages ``filled'', (both for aesthetic and economic reasons) stand-alone
116 figures may traditionally be placed on pages in the document far away from the
117 main text that refers to them. \LaTeX\ and \TeX\ tend to encourage this
118 ``drifting away'' of graphics for technical reasons.
119
120 When a graphic is inlined, it will more or less automatically be linked with
121 the main text in the sense that the labels of the graphic will be implicitly
122 explained by the surrounding text. Also, the main text will typically make it
123 clear what the graphic is about and what is shown.
124
125 Quite differently, a stand-alone figure will often be viewed at a time when the
126 main text that this graphic belongs to either has not yet been read or has been
127 read some time ago. For this reason, you should follow the following guidelines
128 when creating stand-alone figures:
129 %
130 \begin{itemize}
131 \item Stand-alone figures should have a caption than should make them
132 ``understandable by themselves''.
133
134 For example, suppose a graphic shows an example of the different
135 stages of a quicksort algorithm. Then the figure's caption should, at
136 the very least, inform the reader that ``the figure shows the
137 different stages of the quicksort algorithm introduced on page xyz''.
138 and not just ``Quicksort algorithm''.
139 \item A good caption adds as much context information as possible. For
140 example, you could say: ``The figure shows the different stages of
141 the quicksort algorithm introduced on page xyz. In the first line,
142 the pivot element 5 is chosen. This causes\dots'' While this
143 information can also be given in the main text, putting it in the
144 caption will ensure that the context is kept. Do not feel afraid of a
145 5-line caption. (Your editor may hate you for this. Consider hating
146 them back.)
147 \item Reference the graphic in your main text as in ``for an example of
148 quicksort `in action', see Figure~2.1 on page xyz''.
149 \item Most books on style and typography recommend that you do not use
150 abbreviations as in ``Fig.~2.1'' but write ``Figure~2.1''.
151
152 The main argument against abbreviations is that ``a period is too
153 valuable to waste it on an abbreviation''. The idea is that a period
154 will make the reader assume that the sentence ends after ``Fig'' and
155 it takes a ``conscious backtracking'' to realize that the sentence
156 did not end after all.
157
158 The argument in favor of abbreviations is that they save space.
159
160 Personally, I am not really convinced by either argument. On the one
161 hand, I have not yet seen any hard evidence that abbreviations slow
162 readers down. On the other hand, abbreviating all ``Figure'' by
163 ``Fig.'' is most unlikely to save even a single line in most documents.
164 I avoid abbreviations.
165 \end{itemize}
166
167
168 \subsection{Consistency Between Graphics and Text}
169
170 Perhaps the most common ``mistake'' people do when creating graphics (remember
171 that a ``mistake'' in design is always just ``ignorance'') is to have a
172 mismatch between the way their graphics look and the way their text looks.
173
174 It is quite common that authors use several different programs for creating the
175 graphics of a paper. An author might produce some plots using \textsc{gnuplot},
176 a diagram using \textsc{xfig}, and include an |.eps| graphic a coauthor
177 contributed using some unknown program. All these graphics will, most likely,
178 use different line widths, different fonts, and have different sizes. In
179 addition, authors often use options like |[height=5cm]| when including graphics
180 to scale them to some ``nice size''.
181
182 If the same approach were taken to writing the main text, every section would
183 be written in a different font at a different size. In some sections all
184 theorems would be underlined, in another they would be printed all in uppercase
185 letters, and in another in red. In addition, the margins would be different on
186 each page. Readers and editors would not tolerate a text if it were written in
187 this fashion, but with graphics they often have to.
188
189 To create consistency between graphics and text, stick to the following
190 guidelines:
191 %
192 \begin{itemize}
193 \item Do not scale graphics.
194
195 This means that when generating graphics using an external program,
196 create them ``at the right size''.
197 \item Use the same font(s) both in graphics and the body text.
198 \item Use the same line width in text and graphics.
199
200 The ``line width'' for normal text is the width of the stem of letters
201 like T{}. For \TeX, this is usually $0.4\,\mathrm{pt}$. However, some
202 journals will not accept graphics with a normal line width below
203 $0.5\,\mathrm{pt}$.
204 \item When using colors, use a consistent color coding in the text and in
205 graphics. For example, if red is supposed to alert the reader to
206 something in the main text, use red also in graphics for important
207 parts of the graphic. If blue is used for structural elements like
208 headlines and section titles, use blue also for structural elements
209 of your graphic.
210
211 However, graphics may also use a logical intrinsic color
212 coding. For example, no matter what colors you normally use, readers
213 will generally assume, say, that the color green as ``positive, go,
214 ok'' and red as ``alert, warning, action''.
215 \end{itemize}
216
217 Creating consistency when using different graphic programs is almost
218 impossible. For this reason, you should consider sticking to a single graphics
219 program.
220
221
222 \subsection{Labels in Graphics}
223
224 Almost all graphics will contain labels, that is, pieces of text that explain
225 parts of the graphics. When placing labels, stick to the following guidelines:
226 %
227 \begin{itemize}
228 \item Follow the rule of consistency when placing labels. You should do
229 so in two ways: First, be consistent with the main text, that is, use
230 the same font as the main text also for labels. Second, be consistent
231 between labels, that is, if you format some labels in some particular
232 way, format all labels in this way.
233 \item In addition to using the same fonts in text and graphics, you
234 should also use the same notation. For example, if you write $1/2$ in
235 your main text, also use ``$1/2$'' as labels in graphics, not
236 ``0.5''. A $\pi$ is a ``$\pi$'' and not ``$3.141$''. Finally,
237 $\mathrm e^{-\mathrm i \pi}$ is ``$\mathrm e^{-\mathrm i \pi}$'', not
238 ``$-1$'', let alone ``-1''.
239 \item Labels should be legible. They should not only have a reasonably
240 large size, they also should not be obscured by lines or other text.
241 This also applies to labels of lines and text \emph{behind} the
242 labels.
243 \item Labels should be ``in place''. Whenever there is enough space,
244 labels should be placed next to the thing they label. Only if
245 necessary, add a (subdued) line from the label to the labeled object.
246 Try to avoid labels that only reference explanations in external
247 legends. Reader have to jump back and forth between the explanation and
248 the object that is described.
249 \item Consider subduing ``unimportant'' labels using, for example, a gray
250 color. This will keep the focus on the actual graphic.
251 \end{itemize}
252
253
254 \subsection{Plots and Charts}
255
256 One of the most frequent kind of graphics, especially in scientific papers, are
257 \emph{plots}. They come in a large variety, including simple line plots,
258 parametric plots, three dimensional plots, pie charts, and many more.
259
260 Unfortunately, plots are notoriously hard to get right. Partly, the default
261 settings of programs like \textsc{gnuplot} or Excel are to blame for this since
262 these programs make it very convenient to create bad plots.
263
264 The first question you should ask yourself when creating a plot is: Are there
265 enough data points to merit a plot? If the answer is ``not really'', use a
266 table.
267
268 A typical situation where a plot is unnecessary is when people present a few
269 numbers in a bar diagram. Here is a real-life example: At the end of a seminar
270 a lecturer asked the participants for feedback. Of the 50 participants, 30
271 returned the feedback form. According to the feedback, three participants
272 considered the seminar ``very good'', nine considered it ``good'', ten ``ok'',
273 eight ``bad'', and no one thought that the seminar was ``very bad''.
274
275 A simple way of summing up this information is the following table:
276
277 \medskip
278 \begin{tabular}{lp{3.75cm}r}
279 \emph{Rating given} & \raggedright\emph{Participants (out of 50) who gave this rating} &
280 \emph{Percentage} \\[1.75em]
281 ``very good'' & \hfil\hphantom{0}3\hfil & \hphantom{0}6\% \\
282 ``good'' & \hfil\hphantom{0}9\hfil & 18\% \\
283 ``ok'' & \hfil10\hfil & 20\% \\
284 ``bad'' & \hfil\hphantom{0}8\hfil & 16\% \\
285 ``very bad'' & \hfil\hphantom{0}0\hfil & \hphantom{0}0\% \\[2mm]
286 none & \hfil20\hfil & 40\% \\
287 \end{tabular}
288
289 \bigskip
290 What the lecturer did was to visualize the data using a 3D bar diagram. It
291 looked like this (except that in reality the numbers where typeset using some
292 extremely low-resolution bitmap font and were near-unreadable):
293
294 \bigskip
295 \par
296 \begin{tikzpicture}[y=0.03cm,z=3mm]
297 \foreach \y in {0,20,40,60,80,100}
298 \draw[dashed] (0,\y,0) node[left] {\y} -- (0,\y,1) -- (6,\y,1);
299
300 \draw (0,0,0) -- (0,100,0) (0,0,1) -- (0,100,1);
301 \draw (0,0,0) -- (6,0,0);
302
303 \foreach \x/\xtext/\height in {1/very good/10,2/good/30,3/ok/33,4/bad/27,5/very bad/0}
304 {
305 \draw (\x,0) node[rotate=90,anchor=east] {\xtext};
306
307 \begin{scope}[xshift=\x cm]
308
309 \filldraw[fill=blue!50] (-.3,0,0) rectangle (.3,\height,0);
310 \filldraw[fill=blue!30] (.3,0,0) -- (.3,0,1) -- (.3,\height,1) -- (.3,\height,0) --cycle;
311 \filldraw[fill=blue!20] (-.3,\height,0) -- (.3,\height,0) --
312 (.3,\height,1) -- (-.3,\height,1) --cycle;
313 \end{scope}
314 }
315 \end{tikzpicture}
316 \bigskip
317
318 Both the table and the ``plot'' have about the same size. If your first thought
319 is ``the graphic looks nicer than the table'', try to answer the following
320 questions based on the information in the table or in the graphic:
321 %
322 \begin{enumerate}
323 \item How many participants were there?
324 \item How many participants returned the feedback form?
325 \item What percentage of the participants returned the feedback form?
326 \item How many participants checked ``very good''?
327 \item What percentage out of all participants checked ``very good''?
328 \item Did more than a quarter of the participants check ``bad'' or ``very
329 bad''?
330 \item What percentage of the participants that returned the form checked
331 ``very good''?
332 \end{enumerate}
333
334 Sadly, the graphic does not allow us to answer \emph{a single one of these
335 questions}. The table answers all of them directly, except for the last one. In
336 essence, the information density of the graphic is very close to zero. The
337 table has a much higher information density; despite the fact that it uses
338 quite a lot of white space to present a few numbers. Here is the list of things
339 that went wrong with the 3D-bar diagram:
340 %
341 \begin{itemize}
342 \item The whole graphic is dominated by irritating background lines.
343 \item It is not clear what the numbers at the left mean; presumably
344 percentages, but it might also be the absolute number of
345 participants.
346 \item The labels at the bottom are rotated, making them hard to read.
347
348 (In the real presentation that I saw, the text was rendered at a very
349 low resolution with about 10 by 6 pixels per letter with wrong
350 kerning, making the rotated text almost impossible to read.)
351 \item The third dimension adds complexity to the graphic without adding
352 information.
353 \item The three dimensional setup makes it much harder to gauge the
354 height of the bars correctly. Consider the ``bad'' bar. Is the number
355 this bar stands for more than 20 or less? While the front of the bar
356 is below the 20 line, the back of the bar (which counts) is above.
357 \item It is impossible to tell which numbers are represented by the
358 bars. Thus, the bars needlessly hide the information these bars are
359 all about.
360 \item What do the bar heights add up to? Is it 100\% or 60\%?
361 \item Does the bar for ``very bad'' represent 0 or~1?
362 \item Why are the bars blue?
363 \end{itemize}
364
365 You might argue that in the example the exact numbers are not important for the
366 graphic. The important things is the ``message'', which is that there are more
367 ``very good'' and ``good'' ratings than ``bad'' and ``very bad''. However, to
368 convey this message either use a sentence that says so or use a graphic that
369 conveys this message more clearly:
370
371 \medskip
372 \par
373 \begin{tikzpicture}
374 \colorlet{good}{green!75!black}
375 \colorlet{bad}{red}
376 \colorlet{neutral}{black!60}
377 \colorlet{none}{white}
378
379 \node[align=center,text width=3cm]{Ratings given by 50~participants};
380
381 \begin{scope}[line width=4mm,rotate=270]
382 \draw[good] (-123:2cm) arc (-123:-101:2cm);
383 \draw[good!60!white] (-36:2cm) arc (-36:-101:2cm);
384 \draw[neutral] (-36:2cm) arc (-36:36:2cm);
385 \draw[bad!60!white] (36:2cm) arc (36:93:2cm);
386
387 \newcount\mycount
388 \foreach \angle in {0,72,...,3599}
389 {
390 \mycount=\angle\relax
391 \divide\mycount by 10\relax
392 \draw[black!15,thick] (\the\mycount:18mm) -- (\the\mycount:22mm);
393 }
394
395 \draw (0:2.2cm) node[below] {``ok'': 10 (20\%)};
396 \draw (165:2.2cm) node[above] {none: 20 (40\%)};
397 \draw (-111:2.2cm) node[left] {``very good'': 3 (6\%)};
398 \draw (-68:2.2cm) node[left] {``good'': 9 (18\%)};
399 \draw (65:2.2cm) node[right] {``bad'': 8 (16\%)};
400 \draw (93:2.2cm) node[right] {``very bad'': 0 (0\%)};
401 \end{scope}
402 \draw[gray] (0,0) circle (2.2cm) circle (1.8cm);
403 \end{tikzpicture}
404
405 \bigskip
406 The above graphic has about the same information density as the table (about
407 the same size and the same numbers are shown). In addition, one can directly
408 ``see'' that there are more good or very good ratings than bad ones. One can
409 also ``see'' that the number of people who gave no rating at all is not
410 negligible, which is quite common for feedback forms.
411
412 Charts are not always a good idea. Let us look at an example that I redrew from
413 a pie chart in \emph{Die Zeit}, June 4th, 2005:
414
415 \bigskip
416 \par
417 \begin{tikzpicture}
418 \begin{scope}[xscale=3.2,yscale=1.2]
419
420 \sffamily
421 \coordinate (right border) at (2.0cm,-1.7cm);
422 \coordinate (left border) at (-2.5cm,2.1cm);
423
424 \fill[black!25] ([xshift=-2mm,yshift=1.1cm]left border) rectangle ([xshift=2mm,yshift=-.3cm]right border);
425
426 \node[below right,text width=10cm,inner sep=0pt] at ([yshift=.9cm,xshift=-1mm]left border)
427 { {\color{black!75} \Large Kohle ist am wichtigsten}\\
428 Energiemix bei der deutschen Stromerzeugung 2004};
429
430 \filldraw[draw=gray,fill=white] ([xshift=-1mm]left border) node[below right,black]
431 {\footnotesize Gesamte Netto-Stromerzeugung in Prozent, in
432 Milliarden Kilowattstunden (Mrd.\ kWh)}
433 rectangle ([xshift=1mm]right border);
434
435 % The 3D stuff
436 \pgfdeclarehorizontalshading{zeit}{100bp}
437 {color(0pt)=(black);
438 color(25bp)=(black);
439 color(37bp)=(white);
440 color(50bp)=(black);
441 color(62bp)=(white);
442 color(75bp)=(black);
443 color(100bp)=(black)}
444
445 \shadedraw[very thin,shading=zeit,yshift=-1.5mm] (0,0) circle (1cm);
446
447 \fill[green!20!gray] (0,0) -- (90:1cm) arc (90:-5:1cm);
448 \fill[white!20!gray] (0,0) -- (-5:1cm) arc (-5:-105:1cm);
449 \fill[orange!20!gray] (0,0) -- (-105:1cm) arc (-105:-180:1cm);
450 \fill[orange!60!white] (0,0) -- (180:1cm) arc (180:150:1cm);
451 \fill[black!75!white] (0,0) -- (150:1cm) arc (150:145:1cm);
452 \fill[blue!90!white] (0,0) -- (145:1cm) arc (145:135:1cm);
453 \fill[blue!50!white] (0,0) -- (135:1cm) arc (135:92:1cm);
454 \fill[yellow!50!black] (0,0) -- (92:1cm) arc (92:90:1cm);
455
456 \begin{scope}[very thin]
457 \draw (0,0) -- (90:1cm);
458 \draw (0,0) -- (-5:1cm);
459 \draw (0,0) -- (-105:1cm);
460 \draw (0,0) -- (-180:1cm);
461 \draw (0,0) -- (150:1cm);
462 \draw (0,0) -- (145:1cm);
463 \draw (0,0) -- (135:1cm);
464 \draw (0,0) -- (92:1cm);
465
466 \draw(0,0) circle (1cm);
467 \end{scope}
468
469 \node (Regenerative) at (115:.75cm) {\bfseries 9,4\%};
470 \node (Kernenergie) at (30:.5cm) {\bfseries 27,8\%};
471 \node (Braunkohle) at (-45:.6cm) {\bfseries 25,6\%};
472 \node (Steinkohle) at (-135:.6cm) {\bfseries 22,3\%};
473 \node (Erdgas) at (168:.75cm) {\bfseries 10,4\%};
474 \coordinate (Mineral) at (147:.9cm);
475 \coordinate (Sonstige) at (140:.9cm);
476
477 \small
478 \draw (Regenerative.north) |- ([yshift=.25cm]Regenerative.north -| right border) coordinate (Regenerative label);
479 \draw (91:.9cm) |- (Regenerative label);
480 \node[above left] at (Regenerative label) {Regenerative\
481 {\footnotesize (53,7 kWh)/davon} Wind \textbf{4,4\%} \footnotesize (25,0 kWh)};
482
483 \draw (Kernenergie.base east) -- (Kernenergie.base east -| right border) coordinate (Kernenergie label);
484 \node[above left] at (Kernenergie label) {Kernenergie};
485 \node[below left] at (Kernenergie label) {\footnotesize (158,4 kWh)};
486
487 \draw (Braunkohle.south) |- ([yshift=-.75cm]Braunkohle.south -| right border) coordinate (Braunkohle label);
488 \node[above left] at (Braunkohle label) {Braunkohle\ \ \footnotesize (146,0 kWh)};
489
490 \draw (Steinkohle.south) |- ([yshift=-.75cm]Steinkohle.south -| left border) coordinate (Steinkohle label);
491 \node[above right] at (Steinkohle label) {Steinkohle\ \ \footnotesize (127,1 kWh)};
492
493 \draw (Erdgas.base west) -- (Erdgas.base west -| left border) coordinate (Erdgas label);
494 \node[above right] at (Erdgas label) {Erdgas\ \ \footnotesize (59,2 kWh)};
495
496 \draw (Mineral) -- (Mineral -| left border) coordinate (Mineral label);
497 \node[above right] at (Mineral label) {Mineral\"olprodukte\ \
498 \footnotesize (9,2 kWh) \ \ \normalsize\textbf{1,6\%}};
499
500 \draw (Sonstige) |- (Regenerative label -| left border) coordinate (Sonstige label);
501 \node[above right] at (Sonstige label) {Sonstige\ \
502 \footnotesize (16,5 kWh) \hskip1.5cm\
503 \normalsize\textbf{2,9\%}};
504 \end{scope}
505 \end{tikzpicture}
506
507 This graphic has been redrawn in \tikzname, but the original looks almost
508 exactly the same.
509
510 At first sight, the graphic looks ``nice and informative'', but there are a lot
511 of things that went wrong:
512 %
513 \begin{itemize}
514 \item The chart is three dimensional. However, the shadings add nothing
515 ``information-wise'', at best, they distract.
516 \item In a 3D-pie-chart the relative sizes are very strongly distorted.
517 For example, the area taken up by the gray color of ``Braunkohle'' is
518 larger than the area taken up by the green color of ``Kernenergie''
519 \emph{despite the fact that the percentage of Braunkohle is less than
520 the percentage of Kernenergie}.
521 \item The 3D-distortion gets worse for small areas. The area of
522 ``Regenerative'' somewhat larger than the area of ``Erdgas''. The
523 area of ``Wind'' is slightly smaller than the area of
524 ``Mineral\"olprodukte'' \emph{although the percentage of Wind is
525 nearly three times larger than the percentage of
526 Mineral\"olprodukte.}
527
528 In the last case, the different sizes are only partly due to
529 distortion. The designer(s) of the original graphic have also made
530 the ``Wind'' slice too small, even taking distortion into
531 account. (Just compare the size of ``Wind'' to ``Regenerative'' in
532 general.)
533 \item According to its caption, this chart is supposed to inform us that
534 coal was the most important energy source in Germany in 2004.
535 Ignoring the strong distortions caused by the superfluous and
536 misleading 3D-setup, it takes quite a while for this message to get
537 across.
538
539 Coal as an energy source is split up into two slices: one for
540 ``Steinkohle'' and one for ``Braunkohle'' (two different kinds of
541 coal). When you add them up, you see that the whole lower half of
542 the pie chart is taken up by coal.
543
544 The two areas for the different kinds of coal are not visually
545 linked at all. Rather, two different colors are used, the labels are
546 on different sides of the graphic. By comparison, ``Regenerative''
547 and ``Wind'' are very closely linked.
548 \item The color coding of the graphic follows no logical pattern at all.
549 Why is nuclear energy green? Regenerative energy is light blue,
550 ``other sources'' are blue. It seems more like a joke that the area
551 for ``Braunkohle'' (which literally translates to ``brown coal'') is
552 stone gray, while the area for ``Steinkohle'' (which literally
553 translates to ``stone coal'') is brown.
554 \item The area with the lightest color is used for ``Erdgas''. This area
555 stands out most because of the brighter color. However, for this
556 chart ``Erdgas'' is not really important at all.
557 \end{itemize}
558 %
559 Edward Tufte calls graphics like the above ``chart junk''. (I am happy to
560 announce, however, that \emph{Die Zeit} has stopped using 3D pie charts and
561 their information graphics have got somewhat better.)
562
563 Here are a few recommendations that may help you avoid producing chart junk:
564 %
565 \begin{itemize}
566 \item Do not use 3D pie charts. They are \emph{evil}.
567 \item Consider using a table instead of a pie chart.
568 \item Do not apply colors randomly; use them to direct the readers's
569 focus and to group things.
570 \item Do not use background patterns, like a crosshatch or diagonal
571 lines, instead of colors. They distract. Background patterns in
572 information graphics are \emph{evil}.
573 \end{itemize}
574
575
576 \subsection{Attention and Distraction}
577
578 Pick up your favorite fiction novel and have a look at a typical page. You will
579 notice that the page is very uniform. Nothing is there to distract the reader
580 while reading; no large headlines, no bold text, no large white areas. Indeed,
581 even when the author does wish to emphasize something, this is done using
582 italic letters. Such letters blend nicely with the main text -- at a distance
583 you will not be able to tell whether a page contains italic letters, but you
584 would notice a single bold word immediately. The reason novels are typeset this
585 way is the following paradigm: Avoid distractions.
586
587 Good typography (like good organization) is something you do \emph{not} notice.
588 The job of typography is to make reading the text, that is, ``absorbing'' its
589 information content, as effortless as possible. For a novel, readers absorb the
590 content by reading the text line-by-line, as if they were listening to someone
591 telling the story. In this situation anything on the page that distracts the
592 eye from going quickly and evenly from line to line will make the text harder
593 to read.
594
595 Now, pick up your favorite weekly magazine or newspaper and have a look at a
596 typical page. You will notice that there is quite a lot ``going on'' on the
597 page. Fonts are used at different sizes and in different arrangements, the text
598 is organized in narrow columns, typically interleaved with pictures. The reason
599 magazines are typeset in this way is another paradigm: Steer attention.
600
601 Readers will not read a magazine like a novel. Instead of reading a magazine
602 line-by-line, we use headlines and short abstracts to check whether we want to
603 read a certain article or not. The job of typography is to steer our attention
604 to these abstracts and headlines, first. Once we have decided that we want to
605 read an article, however, we no longer tolerate distractions, which is why the
606 main text of articles is typeset exactly the same way as a novel.
607
608 The two principles ``avoid distractions'' and ``steer attention'' also apply to
609 graphics. When you design a graphic, you should eliminate everything that will
610 ``distract the eye''. At the same time, you should try to actively help the
611 reader ``through the graphic'' by using fonts/colors/line widths to highlight
612 different parts.
613
614 Here is a non-exhaustive list of things that can distract readers:
615 %
616 \begin{itemize}
617 \item Strong contrasts will always be registered first by the eye. For
618 example, consider the following two grids:
619
620 \medskip\par
621 \begin{tikzpicture}[x=40pt,y=40pt]
622 \draw[step=10pt,gray] (0,0) grid +(1,1);
623 \draw[step=2pt] (2,0) grid +(1,1);
624 \end{tikzpicture}
625
626 \medskip
627 Even though the left grid comes first in English reading order, the
628 right one is much more likely to be seen first: The white-to-black
629 contrast is higher than the gray-to-white contrast. In addition,
630 there are more ``places'' adding to the overall contrast in the right
631 grid.
632
633 Things like grids and, more generally, help lines usually should not
634 grab the attention of the readers and, hence, should be typeset with
635 a low contrast to the background. Also, a loosely-spaced grid is less
636 distracting than a very closely-spaced grid.
637 \item Dashed lines create many points at which there is black-to-white
638 contrast. Dashed or dotted lines can be very distracting and, hence,
639 should be avoided in general.
640
641 Do not use different dashing patterns to differentiate curves in
642 plots. You lose data points this way and the eye is not particularly
643 good at ``grouping things according to a dashing pattern''. The eye
644 is \emph{much} better at grouping things according to colors.
645 \item Background patterns filling an area using diagonal lines or
646 horizontal and vertical lines or just dots are almost always
647 distracting and, usually, serve no real purpose.
648 \item Background images and shadings distract and only seldom add
649 anything of importance to a graphic.
650 \item Cute little clip arts can easily draw attention away from the data.
651 \end{itemize}